ORIGINAL ARTICLE doi:10.1111/evo.12177 SIZE VARIATION, GROWTH STRATEGIES, AND THE EVOLUTION OF MODULARITY IN THE MAMMALIAN SKULL Arthur Porto, 1,2,3 Leila Teruko Shirai, 4 Felipe Bandoni de Oliveira, 2 and Gabriel Marroig 2 1 Department of Anatomy and Neurobiology, Washington University in St Louis, St Louis, Missouri 2 Departamento de Gen ´ etica e Biologia Evolutiva, Instituto de Bioci ˆ encias, Universidade de S ˜ ao Paulo, CEP 05508-090, S ˜ ao Paulo, SP, Brasil 3 E-mail: [email protected]4 Instituto Gulbenkian de Ci ˆ encias, Oeiras, Portugal Received December 9, 2011 Accepted May 15, 2013 Data Archived: Dryad doi:10.5061/dryad.4d236 Allometry is a major determinant of within-population patterns of association among traits and, therefore, a major component of morphological integration studies. Even so, the influence of size variation over evolutionary change has been largely unap- preciated. Here, we explore the interplay between allometric size variation, modularity, and life-history strategies in the skull from representatives of 35 mammalian families. We start by removing size variation from within-species data and analyzing its influence on integration magnitudes, modularity patterns, and responses to selection. We also carry out a simulation in which we artificially alter the influence of size variation in within-taxa matrices. Finally, we explore the relationship between size variation and different growth strategies. We demonstrate that a large portion of the evolution of modularity in the mammalian skull is associated to the evolution of growth strategies. Lineages with highly altricial neonates have adult variation patterns dominated by size variation, leading to high correlations among traits regardless of any underlying modular process and impacting directly their potential to respond to selection. Greater influence of size variation is associated to larger intermodule correlations, less individualized modules, and less flexible responses to natural selection. KEY WORDS: Allometry, constraints, flexibility, life-history evolution, morphospace, V/CV matrix. Living organisms are no longer viewed as an array of discrete traits, but rather as complex systems in which several characters interact, sharing genetic, developmental, or functional pathways (Cheverud 1984; Emerson and Hastings 1998). Under this view, a trait cannot be considered as evolving independently of others, and the study of how traits are connected is central to evolution- ary biology (Steppan et al. 2002). Studies across several levels of biological organization, from proteins to morphology, have often found traits to be organized in modules, that is, semiau- tonomous complexes of highly intercorrelated traits (Schlosser and Wagner 2004). For studies of morphological evolution, the importance of such modular architecture relates to the observation that morphological variation is also modularly organized (e.g., Olson and Miller 1958; Berg 1960; Cheverud 1982; 1995; Marroig and Cheverud 2001; Porto et al. 2009). Such structure of varia- tion biases the direction, magnitude, and rate of morphological change, either constraining or facilitating evolution along differ- ent dimensions of the morphospace (Maynard Smith et al. 1985; Arnold 1992; Arnold et al. 2001; Marroig and Cheverud 2004a; 2005, 2010). Modularity is, therefore, a central umbrella concept to understand morphological evolution. There is a variety of methods available for describing and quantifying modularity in morphological data. Most methods are based on correlations or covariances among traits, and modularity 3305 C 2013 The Author(s). Evolution C 2013 The Society for the Study of Evolution. Evolution 67-11: 3305–3322
18
Embed
Size variation, growth strategies, and the evolution of modularity in the mammalian skull
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
ORIGINAL ARTICLE
doi:10.1111/evo.12177
SIZE VARIATION, GROWTH STRATEGIES,AND THE EVOLUTION OF MODULARITYIN THE MAMMALIAN SKULLArthur Porto,1,2,3 Leila Teruko Shirai,4 Felipe Bandoni de Oliveira,2 and Gabriel Marroig2
1Department of Anatomy and Neurobiology, Washington University in St Louis, St Louis, Missouri2Departamento de Genetica e Biologia Evolutiva, Instituto de Biociencias, Universidade de Sao Paulo, CEP 05508-090, Sao
DiscussionSize can be regarded as an emergent property of the growth pro-
cess that includes the integration of different modules into a sin-
gle, coherent, functional structure. This integration establishes
correlations among traits on top of local processes, and therefore
potentially obscures the individuality of modules. Module indi-
viduality, in turn, directly impacts the potential for evolutionary
responses to selection of a given structure. Our results point that
the greater the relative influence of size variation over the total
morphological variation, the greater the intermodule correlations,
the less distinct the modules, and the more constrained their po-
tential responses to natural selection.
Evidence supporting this perspective of the evolution of mod-
ularity has been published in four complementary papers (Porto
et al. 2009; Marroig et al. 2009; Shirai and Marroig 2010; and this
article). Our previous studies pointed to the importance of integra-
tion magnitudes in understanding the potential of the evolutionary
3 3 1 6 EVOLUTION NOVEMBER 2013
SIZE VARIATION AND ITS EVOLUTIONARY IMPLICATIONS
1 2 3 4 5 60.1
0.3
0.5
0.7
FLEX
IBIL
ITY
412
2 1514
26 24
3028
7231
2113
632 31
3433
22173527329
1620 918191058
2511
1 2 3 4 5 6ICV
0.4
0.5
0.6
0.7
0.8
0.9
1.0
CONS
TRAI
NTS
412
215
1426
24
30
2872312113
63231
3433
22
173527
3
2916
209
18
19
10582511
1 2 3 4 5 6ICV
0.0
12.5
25.0
37.5
50.0
62.5
75.0
87.5
100.0
% V
ar. P
max
4 12215 1426
2430
287231
2113632 31
343322
173527329
16
20
9181910582511
1 2 3 4 5 6ICV
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
MO
D IN
DEX
12
2
15
14
26
24
30
287
23
1
2113
63231
3433
221735
273291620
9
18
1910
5
8
25
11
ICV1 2 3 4 5 6
ICV
0.1
0.2
0.3
0.4
0.5
0.6
0.7
FLEX
IBIL
ITY
1 2 3 4 5 6ICV
0.40
0.55
0.70
0.85
1.00
CONS
TRAI
NTS
1 2 3 4 5 6ICV
0
25
50
75
100
% V
ar. P
max
1 2 3 4 5 6ICV
0.0000
0.0175
0.0350
0.0525
0.0700
MO
D IN
DEX
Figure 4. Comparison of results obtained for the mammalian families (right column) and the expected under evolution of the influence
of size variation over the total morphological variation (modifications of mammalian V/CV matrices; left column). Plots of the several
indexes (flexibility, constraints, percentage of variation along pmax, and modularity) against the measurement of overall integration
magnitudes are shown for both cases.
EVOLUTION NOVEMBER 2013 3 3 1 7
ARTHUR PORTO ET AL.
1 2 3 4 5ICV
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
EG
Hominidae
Molossidae
Cebidae
VespertilionidaePhyllostomidae Macroscelididae
Sciuridae
Tupaiidae
TapiridaeMustelidae
Cricetidae
Atelidae
Erethizontidae
Noctilionidae
Felidae
Dasyuridae
Tayassuidae
PeramelidaeMacropodidae
Muridae
Caviidae
Didelphidae
ProcaviiddaeCercopithecidae
Dasypodidae
Myrmecophagidae
Dasyproctidae
Delphinidae
Echimyidae
Cuniculidae
Cervidae
Canidae
ProcyonidaeLeporidae
Figure 5. Plot of the proportion of the total metabolic power
devoted to growth (Eg) against the overall integration magnitude
(ICV) calculated for each mammal family. The regression line is
shown in gray.
responses to selection produced by different species, because, as
a rule, covariation patterns remained very similar among mam-
mals while integration magnitudes changed rapidly (Marroig and
Cheverud 2001; Oliveira et al. 2009; Porto et al. 2009). Higher
integration magnitudes were associated with lower evolutionary
flexibilities and less distinct developmental/functional modules in
the skull, and vice versa (Marroig et al. 2009). This pattern was
consistently observed here in several mammal orders, including
taxa that diverged more than 130 million years ago (Bininda-
Emonds et al. 2007).
This study goes further and analyzes the relationship between
size variation, growth strategies, and integration/modularity in the
mammalian skull. The data set used is by far the largest of its kind
in terms of phylogenetic breadth, spanning a total of 35 mammal
families. We used three main analytical methods: (1) removed
size variation from within-species data and evaluated the conse-
quences of it; (2) simulated changes in the influence of size vari-
ation in each species matrix and compared expectations derived
from this manipulation to observed results; and (3) regressed in-
tegration magnitudes on postnatal growth measurements. Results
from these three lines of evidence support each other and paint a
common picture, detailed later.
Much of the overall magnitude of integration among mor-
phological elements can be attributed to size variation. This is
consistent with what was previously observed in a much smaller
data set in terms of phylogenetic scope (neotropical marsupials
and primates) but with a fairly good representativeness of within-
clade diversity (Shirai and Marroig 2010). Once size variation
was removed, integration magnitudes lowered, and became more
similar among groups than previously observed. The effect of size
removal was particularly dramatic in taxa exhibiting higher inte-
gration magnitudes in raw data (e.g., marsupials). In such groups,
integration magnitudes dropped to values more than 50% lower
than their raw counterparts. Concomitantly, modularity patterns
became more evident and modules that could not be detected in
raw matrices (see also Porto et al. 2009) were consistently detected
in all groups. Neurocranial traits, for example, only presented
significant integration in hominid’s raw matrices. In residual ma-
trices, significant neurocranial integration was detected in most
groups. A similar pattern was observed for the cranial vault and zy-
gomatic hypotheses. All families presented significant total inte-
gration in residual matrices, indicating that the failure to detect it in
raw matrices was due to the impact of size variation in integration.
The theoretical simulations involving modifications on the
first eigenvalue of V/CV matrices clearly showed that altering
the relative amount of variation along the pmax of V/CV matrices
leads to results that closely mirror those observed for all mam-
mal families (Fig. 4). Similarly, V/CV matrices exhibiting high
percentages of variation associated with the pmax yielded higher
overall integration magnitudes and less distinct modular patterns.
Results from both approaches support the notion that size acts
as a strong and general integration factor that blurs the largest
part of the modularity signal in morphological variation patterns.
Our results also corroborate the suggestion that studies of mod-
ularity will probably be more successful in detecting modules
once variation induced by global factors, such as size, is mini-
mized compared to variation due to local factors (Mitteroecker
and Bookstein 2007).
Our data further indicate that size variation cannot be con-
sidered a nuisance. Size variation is a genetically variable, bi-
ologically meaningful, and a major component of within-group
variation patterns and magnitudes. Moreover, we have shown that
removing size variation discloses hidden modularity, but not in
the same degree for all groups, nor equally among modules. This
implies that corrections for size variation cannot be treated ho-
mogeneously across taxa because each lineage will behave in
different ways according to how its evolution has been shaped
by size variation. We hope to encourage other researchers to con-
sider the impact and implications of removing size variation when
addressing evolutionary questions, especially in groups like mam-
mals, where growth is determinate and age classes can be estab-
lished with reasonable accuracy. When removing size variation,
a researcher might be neglecting a substantial part of the within
population morphological variation. Given that size variation has
important evolutionary and ecological implications, if size is
removed a priori without further consideration, one might miss
3 3 1 8 EVOLUTION NOVEMBER 2013
SIZE VARIATION AND ITS EVOLUTIONARY IMPLICATIONS
the role and impact of size on the evolution of the taxon under
study. The removal of the main axis of variation also leads to the
undesirable consequence of increasing the proportion of variance
due to error (Marroig et al. 2012). In marsupials, for example, size
removal could increase the proportion of variance due to error to
values above 20%, even when measurement repeatabilities are
higher than 0.95.
Before discussing the impact of size variation on evolution,
it is important to consider that the ability of a population to
evolve in the direction of selection has two aspects: one is the
response vector’s alignment with the direction of selection (flex-
ibility); the other is the amount of change in the direction of
selection (evolvability sensu Hansen and Houle 2008). Our previ-
ous studies found a scaling effect on evolvability among mammals
(Marroig et al. 2009) and a lack of association with integration
and modularity. Although these two aspects are mathematically
and biologically related (flexibility is the ratio between evolvabil-
ity and respondability; Hansen and Houle 2008), we focus here
on the relationship between flexibility and other life-history and
integration/modularity statistics.
The impact of size variation within a population can be
easily captured by the flexibility index. As mentioned earlier,
this metric was employed to represent the capacity of a given
structure to respond in the direction of selection (Marroig
et al. 2009). Thus, if evolutionary responses are closely aligned to
the selection gradient that generated them, one can argue that the
structure under concern is flexible to respond in that direction of
the morphospace. Although flexibility and constraints are obvi-
ously inversely related concepts capturing the overall evolutionary
consequence of modularity/integration, it is important to keep in
mind that flexibility and constraints are calculated through very
different procedures. Constraints are defined as the correlation
between the evolutionary response and pmax. Thus, if selection is
pushing along pmax, both flexibility and constraints should be high
in a strongly integrated matrix. In any case, once size variation
was removed, flexibility values increased in all groups, indicating
higher capacity of closely tracking selection, and also the poten-
tial for broader exploration of the morphospace. This potential
is constrained by the association among morphological elements
induced by size variation, as shown by the constraints index.
The same general pattern can be seen in the simulations alter-
ing the influence of size variation. Matrices with low percentages
of variation associated with pmax yielded lower overall integration
magnitudes, more distinct modular patterns, and more flexibility
in terms of evolutionary responses to selection. The observed pat-
tern for the evolution of the influence of size on morphological
variation of mammals is thus consistent with theoretical expecta-
tions (Fig. 4). These results clearly indicate that size variation can
act as a powerful constraint for evolutionary change in mammals,
which would bias evolutionary responses to selection along the
so-called line of least resistance (Schluter 1996), at least on a
short-term scale if the adaptive landscape is not changing through
time (Arnold et al. 2001). Even when selection gradients are not
strictly aligned with size, taxa with some size variation usually re-
spond along the size dimension (except if selection is orthogonal
to the size variation axis, see Marroig and Cheverud 2010). Large
portions of variation aligned with the size axis, therefore, greatly
reduce the flexibility of a population. In our sample of mammals,
within-population size variation ranged from 20% to almost 81%
of all phenotypic variation (Table 3). Accordingly, those lineages
with a smaller relative amount of size variation (like hominids
and bats) are far less constrained, more modular and flexible with
regard to the potential for selection responses than those lineages
for which the major part of the total variation is associated with
size (like marsupials). It is important to notice that pmax vectors are
not identical across taxa. Pairwise comparisons between the pmax
of different taxa indicate they are not perfectly aligned (Support-
ing Information—S.I.5). In other words, allometric size vectors
do evolve, to some extent, in mammals. However, their effect over
morphological integration is basically the same. Furthermore, al-
though we have been emphasizing this “attractor effect” of size
variation (pmax) with regard to the potential selection response,
other axes of the morphospace also represent lines of least re-
sistance. As long as the within-population variance is unequally
distributed, usually in the form of a negative exponential distribu-
tion of eigenvalues, some axes would present more variance than
others, and would act as attractors of evolutionary responses to
selection (e.g., Hansen and Voje 2011).
All these results lead to the question: which factors could
modify within-population size variation? Part of the answer might
be related to life-history differences among lineages, particularly
related to the growth process. Considering that multicellular com-
plex organisms are built from a single zygotic cell that undergoes
proliferation and differentiation that finally assemble a multi-
cellular integrated adult, one could argue that differential adult
size variation among lineages is a result of differences in the
growth process. Growth affects, to some extent, all traits in an
organism, bringing together modules into a higher hierarchical
structure, like a glue holding parts in a Francois Jacob bricolage
(Jacob 1977). The immediate consequence is that differences in
the growth processes will lead to differences in variance along
the size dimension within a population, with consequences for
modularity and for the evolutionary potential of a population to
respond to natural selection.
The postnatal growth data strongly suggests that the relative
amount of size variation and skull modularity are tightly asso-
ciated with the evolution of life-history strategies in mammals.
Those lineages with altricial neonates, such as marsupials and
rodents, allocate higher portion of the energy budget to growth,
have relatively more variation in size, higher integration, lower
EVOLUTION NOVEMBER 2013 3 3 1 9
ARTHUR PORTO ET AL.
flexibility, and are highly constrained. Conversely, those lineages
with precocial neonates, such as hominids and bats, allocate a
smaller portion of energy to growth, have a smaller portion of the
total variation in size and are less integrated, more flexible, and
less constrained in its evolutionary potential. This also suggests
that different growth strategies existent in different species have a
fundamental impact on their evolutionary flexibility and potential
for the exploration of the morphospace.
Although certainly speculative at this point, the genetic basis
for the evolution of size variation is clearly an important aspect
to be considered. Particular attention has been paid recently to
genetic variation in pleiotropy caused by differential epistatic in-
teractions (e.g., Pavlicev et al. 2011), mainly because genetic
variation in pleiotropy is necessary for it to evolve (Pavlicev
et al. 2007). The presence of relationship QTLs (r-QTL) has al-
ready been demonstrated to play important roles on long bone
growth, with an effect on body size variance (and its allometric
effects) that resembles the effects described here (see Pavlicev
et al. 2007). Differential epistasis causes differential canalization
of single traits and trait combinations, leading to varying grades
of integration among genotypes and allowing for the evolution of
modularity (Pavlicev et al. 2007).
Finally, it is clear from our results that understanding the
connection between life-history strategies, the genetic architec-
ture, and the influence of size variation will be fundamental to
understand how different mammal species respond to evolution-
ary processes at least on a microevolutionary scale. Depending
on the potentially complex relationship of the history of adaptive
peaks and G-matrix orientation in mammals, such dominance of
size variation in modularity might even extend its consequence to
a macroevolutionary scale as already show in New World mon-
keys (Marroig and Cheverud 2001, 2005, 2010). We showed that
mammal species differ in their apportionment of variation in the
morphospace and that these differences are associated with their
life-history strategies. In altricial species, most of the total mor-
phological variation is concentrated in a single size axis (pmax).
In those species, modularity is masked by the strong between-
module correlations associated with variation along the allometric
pmax. Furthermore, this masking effect upon modularity is not ex-
plained by differences in orientation of the allometric vectors but
instead it is a product of the relative length (norm or the relative
amount of variation) of these vectors. Those lineages with a larger
portion of the morphological variation associated with pmax tend to
be more integrated. Accordingly, pmax acts as a strong “attractor”
of evolutionary responses to selection, potentially constraining
their morphological diversification. In precocial species, varia-
tion is more homogeneously apportioned in the morphospace.
Modularity is evident and evolutionary responses more flexible.
As a consequence, we expect those species to be less constrained
in their short-term morphological diversification.
ACKNOWLEDGMENTSThe authors are grateful to A. Templeton and two anonymous review-ers for careful revision of previous drafts of this manuscript, and to J.Cheverud for helpful discussions. The authors are also grateful to peopleand institutions that provided generous help and access to museum col-lections (Supporting Information). This research was supported by grantsand fellowships from FAPESP, CAPES, CNPq, an American Museumof Natural History Collections Study Grant, and by the Department ofBiology from Washington University in St Louis.
LITERATURE CITEDArnold, S. J. 1992. Constraints on phenotypic evolution. Am. Nat. 140:S85–
S107.Arnold, S. J., M. E. Pfrender, and A. G. Jones. 2001. The adaptive landscape
as a conceptual bridge between micro- and macroevolution. Genetica112–113:9–32.
Armbruster, W. S., C. Pelabon, T. F. Hansen, and C. P. H. Mulder. 2004.Floral integration, modularity, and accuracy: distinguishing complexadaptations from genetic constraints. Pp. 23–49 in M. Pigliucci and K.Preston, eds. Phenotypic integration: studying the ecology and evolutionof complex phenotypes. Oxford Univ. Press, Oxford, U.K.
Badyaev, A. V., and K. R. Foresman. 2004. Evolution of morphological in-tegration. I. Functional units channel stress-induced variation in shrewmandibles. Am. Nat. 163:868–879.
Berg, R. L. 1960. The ecological significance of correlation pleiades. Evolu-tion 14:171–180.
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. Nature446:507–512.
Chernoff, B., and P. M. Magwene. 1999. Morphological integration: fourtyyears later. Pp. 316–360 in E. C. Olson and R. L. Miller. Morphologicalintegration. University of Chicago Press, Chicago, IL.
Cheverud, J. M. 1982. Phenotypic, genetic and environmental morphologicalintegration in the cranium. Evolution 36:499–516.
———. 1984. Quantitative genetics and developmental constraints on evolu-tion by selection. J. Theor. Biol. 110:155–172.
———. 1988. A comparison of genetic and phenotypic correlations. Evolution42:958–968.
———. 1995. Morphological integration in the saddle-back tamarin (Sagui-
nus fuscicollis) cranium. Am. Nat. 145:63–89.———. 1996. Quantitative genetic analysis of cranial morphology in the
cotton-top (Saguinus oedipas) and saddle-back (S. fuscicollis) tamarins.J. Evol. Biol. 9:5–42.
Delph, L. F., J. C. Steven, I. A. Anderson, C. R. Herlihy, and E. D. BrodieIII. 2011. Elimination of a genetic correlation between the sexes viaartificial correlational selection. Evolution 65:2872–2880.
Eble, G. 2004. The macroevolution of phenotypic integration. Pp. 253–273in M. Pigliucci and K. Preston, eds. Phenotypic integration: studyingthe ecology and evolution of complex phenotypes. Oxford Univ. Press,Oxford, U.K.
Emerson, S. B., and P. A. Hastings. 1998. Morphological correlations inevolution: consequences for phylogenetic analyses. Quart. Rev. Biol.73:141–162.
Hansen, T. F., and D. Houle. 2008. Measuring and comparing evolvability andconstraint in multivariate characters. J. Evol. Biol. 21:1201–1219.
Hansen, T. F., and K. L. Voje. 2011. Deviation from the line of least resistancedoes not exclude genetic constraints: a comment on Berner et al. (2010).Evolution 65:1821–1822.
3 3 2 0 EVOLUTION NOVEMBER 2013
SIZE VARIATION AND ITS EVOLUTIONARY IMPLICATIONS
Harmon, L. K., and R. E. Glor. 2010. Poor statistical performance of theMantel test in comparative phylogenetic analyses. Evolution 64:2173–2178.
Hayssen, V., Ari van Tienhoven, and Ans van Tienhoven. 1993. Asdell’spatterns of mammalian reproduction: a compendium of species-specificdata. Cornell Univ. Press, New York.
Huxley, J. S. 1932. Problems of relative growth. Johns Hopkins Univ. Press,Baltimore, MD.
Jacob, F. 1977. Evolution and tinkering. Science 196:1161–1166.Jones, K. E., J. Bielby, M. Cardillo, S. A. Fritz, J. O’Dell, C. D. L. Orme, K.
Safi, W. Sechrest, E. H. Boakes, C. Carbone, et al. 2009. PanTHERIA: aspecies-level database of life history, ecology, and geography of extantand recently extinct mammals. Ecology 90:2648.
Klingenberg, C. P. 2004. Integration, modules and development: moleculesto morphology to evolution. Pp. 213–230 in M. Pigliucci and K. Pre-ston, eds. Phenotypic integration: studying the ecology and evolution ofcomplex phenotypes. Oxford Univ. Press, Oxford, U.K.
Klingenberg, C. P. 2009. Morphometric integration and modularity in config-urations of landmarks: tools for evaluating a priori hypotheses. Evol.Dev. 11:405–421.
Lande, R. 1979. Quantitative genetic analysis of multivariate evolution appliedto brain:body size allometry. Evolution 33:402–416.
Lessels, C. M., and P. T. Boag, 1987. Unrepeatable repeatabilities: a commonmistake. Auk 104:116–121.
Lindenfors, P., J. L. Gittleman, and K. E. Jones. 2007. Sexual size dimorphismin mammals. Pp. 16–26 in D. J. Fairbairn, W. U. Blanckenhorn, and T.Szekely, eds. Sex, size and gender roles: evolutionary studies of sexualsize dimorphism. Oxford Univ. Press, Oxford, U.K.
De Magalhaes, J. P., and J. Costa. 2009. A database of vertebrate longevityrecords and their relation to other life-history traits. J. Evol. Biol.22:1770–1774.
Marroig, G., and J. M. Cheverud. 2001. A comparison of phenotypic vari-ation and covariation patterns and the role of phylogeny, ecology, andontogeny during cranial evolution of New World monkeys. Evolution55:2576–2600.
———. 2004a. Cranial evolution in sakis (Pithecia, Platyrrhini) I: Inter-specific differentiation and allometric patterns. Am. J. Phys. Anthrop.125:266–278.
———. 2004b. Did natural selection or genetic drift produce the cranialdiversification of neotropical monkeys? Am. Nat. 163:417–428.
———. 2005. Size as a line of least evolutionary resistance: diet and adaptivemorphological radiation in New World Monkeys. Evolution 59:1128–1142.
———. 2010. Size as a line of least resistance II: direct selection on size orcorrelated response due to constraints? Evolution 64:1470–1488.
Marroig, G., M. de Vivo, and J. M. Cheverud. 2004. Cranial evolution in sakis(Pithecia, Platyrrhini) II: evolutionary processes and morphological in-tegration. J . Evol . Biol. 17:144–155.
Marroig, G., L. T. Shirai, A. Porto, F. B. Oliveira, and V. De Conto. 2009.The evolution of modularity in the mammalian skull II: evolutionaryconsequences. Evol. Biol. 36:136–148.
Marroig, G., D. A. R. Melo, and G. Garcia. 2012. Modularity, noise andnatural selection. Evolution 66:1506–1524.
Martınez-Abadıas, N., M. Esparza, T. Sjøvold, R. Gonzalez-Jose, M. Santos,M. Hernandez, and C. P. Klingenberg. 2011. Pervasive genetic integra-tion directs the evolution of human skull shape. Evolution 66:1010–1023.
Maynard Smith, J., R. Burian, S. Kauffman, P. Alberch, J. Campbell, B.Goodwin, R. Lande, D. Raup, and L. Wolpert. 1985. Developmen-tal constraints and evolution: a perspective from the Mountain Lake
Conference on Development and Evolution. Quart Rev. Biol. 60:265–287.
Mitteroecker, P., and F. Bookstein. 2007. The evolutionary role of modularityand integration in the hominoid cranium. Evolution 62:943–958.
Monteiro, L. R., V. Bonato, and S. F. dos Reis. 2005. Evolutionary integrationand morphological diversification in complex morphological structures:mandible shape divergence in spiny rats (Rodentia, Echimyidae). Evol.Dev. 7:429–39.
Moore, W. J. 1981. The mammalian skull. Cambridge Univ. Press, Cambridge,U.K.
Nijhout, H. F. 2011. Dependence of morphometric allometries on the growthkinetics of body parts. J. Theor. Biol. 288:35–43.
Oliveira, F. B., A. Porto, and G. Marroig. 2009. Covariance structure in theskull of Catarrhini: a case of pattern stasis and magnitude evolution.J. Human Evol. 56:417–430.
Olson, E. C., and R. L. Miller. 1958. Morphological integration. Universityof Chicago Press, Chicago, IL.
Pavlicev, M., J. P. Kenney-Hunt, E. A. Norgard, C. C. Roseman, J. B. Wolf,and J. M. Cheverud. 2007. Genetic variation in pleiotropy: differentialepistasis as a source of variation in the allometric relationship betweenlong bone lengths and body weight. Evolution 62:199–213.
Pavlicev, M., E. A. Norgard, G. L. Fawcett, and J. M. Cheverud. 2011. Evo-lution of pleiotropy: epistatic interaction pattern supports a mechanisticmodel underlying variation in genotype–phenotype map. J. Exp. Zool.316B:371–385.
Porto, A., F. B. Oliveira, L. T. Shirai, V. De Conto, and G. Marroig. 2009.The evolution of modularity in the mammalian skull I: morphologicalintegration patterns and magnitudes. Evol. Biol. 36:118–135.
Promislow, D. E. L., and P. H. Harvey. 1990. Living fast and dying young: acomparative analysis of life-history variation among mammals. J. Zool.220:417–437.
Read, A. F., and P. H. Harvey. 1989. Life history differences among theEutherian radiations. J. Zool. 219:329–353.
Schlosser, G., and G. P. Wagner, eds. 2004. Modularity in development andevolution. University of Chicago Press, Chicago, IL.
Schluter, D. 1996. Adaptive radiation along genetic lines of least resistance.Evolution 50:1766–1774.
Shirai, L. T., and G. Marroig. 2010. Skull modularity in neotropical marsupialsand monkeys: size variation and evolutionary constraint and flexibility.J. Exp. Zool. 314B:663–683.
Smith, K. K. 1996. Integration of craniofacial structures during developmentin mammals. Am. Zool. 36:70–79.
———. 1997. Comparative patterns of craniofacial development in Eutherianand Metatherian mammals. Evolution 51:1663–1678.
———. 2001. Early development of the neural plate, neural crest and facialregion of marsupials. J. Anat. 199:121–131.
Steppan, S. J., P. C. Phillips, and D. Houle. 2002. Comparative quantitativegenetics: evolution of the G matrix. Trends Ecol. Evol. 17(7):320–327.
Wagner, G. P., and L. Altenberg. 1996. Complex adaptations and the evolutionof evolvability. Evolution 50:967–976.
West, G. B., J. H. Brown, and B. J. Enquist. 2001. A general model forontogenetic growth. Nature 413:628–631.
Wilson, D. E., and D. M. Reeder, eds. 2005. Mammal species of the world: ataxonomic and geographic reference. Johns Hopkins Univ. Press, Balti-more, MD.
Young, N. M., G. P. Wagner, and B. Hallgrımsson. 2010. Developmentand the evolvability of human limbs. Proc. Nat. Acad. Sci. 107:3400–3405.
Associate Editor: P. Lindenfors
EVOLUTION NOVEMBER 2013 3 3 2 1
ARTHUR PORTO ET AL.
Supporting InformationAdditional Supporting Information may be found in the online version of this article at the publisher’s website:
Figure S1. (A) Plot of the distribution of resampled ICV values for each mammalian family. Range of the distribution is
proportional to the degree of uncertainty in the ICV estimate. Results are shown for 100 resamplings. (B) Plot of the distribution
of resampled ICV values for each mammalian family after the adjustment suggested by Young et al. (2010).
Figure S2. Plot of the distribution of resampled Flexibility and Constraints indexes for each mammalian family.
Figure S3. Distribution of pairwise vector correlations among the pmax of the different mammalian families.
Figure S4. (A) Average vector correlation between each family pmax and that of other families plotted against the deviations of the
observed flexibility values from theoretical expectations. (B) Flexibility index calculated from residual matrices plotted against
the deviations of the observed flexibility values from theoretical expectations.
Table S1. Skull measurements (see Fig. 2 for acronyms) and their association with the theoretical modularity hypotheses for each
order.
Table S2. Variables measured for the simulation involving the V/CV matrix of each taxon with altered proportions of size variation.