Download The Integrated Phenotype

Integrative and Comparative Biology
Integrative and Comparative Biology, volume 52, number 1, pp. 64–76
doi:10.1093/icb/ics043
Society for Integrative and Comparative Biology
SYMPOSIUM
The Integrated Phenotype
Courtney J. Murren1
Department of Biology, College of Charleston, 66 George Street, Charleston, SC 29424, USA
From the symposium ‘‘The Impacts of Developmental Plasticity on Evolutionary Innovation and Diversification’’
presented at the annual meeting of the Society for Integrative and Comparative Biology, January 3–7, 2012 at
Charleston, South Carolina.
1
E-mail: [email protected]
Synopsis Proper functioning of complex phenotypes requires that multiple traits work together. Examination of relationships among traits within and between complex characters and how they interact to function as a whole organism is
critical to advancing our understanding of evolutionary developmental plasticity. Phenotypic integration refers to the
relationships among multiple characters of a complex phenotype, and their relationships with other functional units
(modules) in an organism. In this review, I summarize a brief history of the concept of phenotypic integration in plant
and animal biology. Following an introduction of concepts, including modularity, I use an empirical case-study approach
to highlight recent advance in clarifying the developmental and genomic basis of integration. I end by highlighting some
novel approaches to genomic and epigenetic perturbations that offer promise in further addressing the role of phenotypic
integration in evolutionary diversification. In the age of the phenotype, studies that examine the genomic and developmental changes in relationships of traits across environments will shape the next chapter in our quest for understanding
the evolution of complex characters.
Introduction
The rapid expansion of publicly available genomic
data over the last decade has provided ample genetic
information about model and nonmodel organisms,
as well as about replicate populations within species
(e.g., 1000 genomes project in humans and 1001 genomes project in Arabidopsis thaliana). The expansion of these data will continue as costs continue to
decrease and efficiency of methods increase (Lister
et al. 2009). In addition, information from these
data sets is contributing to advances in broad areas
of evolutionary biology, from species’ phylogenetic
relationships to the evolution of genomes (Alföldi
et al. 2011; Rutter et al. 2012). Whereas genetic
and genomic data have become abundant in the
last decade, synthesis of accompanying phenotypic
data lag behind.
The era of the phenotype or phenomics (Houle
et al. 2010; Furbank and Tester 2011) is upon us.
Classical approaches to collection of phenotypic
data are joined by new technologies (e.g., mircroarray, RNAseq, and imaging analyses) and innovative
statistical methodologies, which through time may
contribute to enhanced capacity for collection of
phenotypic data (Cheng et al. 2011; Furbank and
Tester 2011; Sozzani and Benfey 2011). The rapid
expansion of genomic data in the last decade was
accompanied by conceptual and analytical growth
in bioinformatics. An analogous conceptual and
methodological set of advances is needed to pull together vast historic and contemporary phenotypic
data sets and to interpret these large phenotypic
data sets on their own, through development, and
in light of genomic data. A general question in evolutionary biology is the relationship between variation at the genomic level and in variation in complex
phenotypes. One place to begin to build an understanding of complex phenotypes and the influences
of genomic, developmental, and environmental variation is through the examination of phenotypic
integration.
History and origins
Phenotypic integration refers to the biological relationships among multiple characters of a complex
phenotype, and their relationships with other
Advanced Access publication May 15, 2012
ß The Author 2012. Published by Oxford University Press on behalf of the Society for Integrative and Comparative Biology. All rights reserved.
For permissions please email: [email protected].
Phenotypic integration
functional units (or modules) in the organism
(Schlichting 1989; Pigliucci and Preston 2004).
Proper functioning of complex phenotypes requires
that traits work together within a module.
Modularity describes the relative independence of
complex traits. The interactions among characters
may change through development and across environments (Murren 2002). Formalization of the concept of phenotypic integration began more than half
a century ago, inspired by research dating back to
descriptive work in the early 1900s by Thompson
(1917) (Pigliucci and Preston 2004). Olson and
Miller (1958) defined the term morphological integration as the interdependence of morphological
traits to produce an organized functional organism.
Their book emphasized the role of development in
trait function, and argued the importance of using
statistical approaches to study morphological integration. Their perspectives on complex morphological
phenotypes inspired much subsequent research on
morphology and evolution of form (see Afterword
by Chernoff and Magwene 1999 in Olson and
Millers 1958; also see Cheverud 1982, 1996;
Klingenberg 2010). From the early formal discussions
of phenotypic integration, the joint importance of
development, environment, and differentiation in
patterns of integration among species or populations
has been apparent. The history of the concept has
shaped the modern study of integration such that the
study of ecological, genetic, and developmental influences on complex traits are woven together.
About the time that Olson and Miller were examining animal morphologies, Clausen et al. (1940) and
Clausen and Heisey (1958, 1960) were conducting
long-term investigations on a suite of plant species
planted across an altitudinal gradient in the Sierra
Nevada. These botanists and cytogeneticists hypothesized that species’ differences are largely due to differences in patterns of what they termed character
coherence (patterns in the relationships among
traits). Their work emphasized the role of the environment in both the maintenance of character coherence and the plasticity of phenotypes (although it
was not termed plasticity at that time). These studies
of patterns of variation of traits’ relationships stimulated a shift in research focus in plant biology toward investigations that examined the influence of
the environment on traits and their functional relationships on integration. Berg (1960) developed and
tested hypotheses of whether patterns of traits’ relationships (that she termed correlation pleiades)
65
differed between taxa pollinated by a specialized or
generalized pollinator systems. She found evidence
for greater integration within flowers for plants
with species-specific pollination systems than for
those with generalist pollination systems. Tests of
Berg’s hypotheses in various species and communities are now numerous, yet have mixed support
(Armbruster et al. 1999; Hansen et al. 2007;
Edwards and Weinig 2011).
A small but active group of researchers worked on
aspects of integration in the 1980s and 1990s building evidence of the importance of development and
environmental variation (Cheverud 1982; Schlichting
1989; Cheverud 1996; Wagner and Altenberg 1996;
Callahan and Waller 2000). Following publication of
the volume Phenotypic Integration edited by Pigliucci
and Preston (2004), the conceptual framework of
phenotypic integration moved into mainstream evolutionary biology. At the time of preparation of this
article, over 100 citations with the key word phenotypic integration were indexed in Web of Science
(ISI) since 2004; with plants and vertebrate animals
the most widely examined organisms. Differing perspectives on the evolution of complex traits and phenotypic integration from these two major lineages of
eukaryotes is, in part, due to differences in body
plans, and to the influence of the research questions
framed by Berg and colleagues on the plant side and
Olson and Miller on the animal side.
In the first section of this article, I provide a brief
overview to the concepts of modules, modularity,
and morphological integration—a specific aspect of
phenotypic integration. Exhaustive coverage of modularity is beyond the scope of this review; rather, the
goal of this section of the article is to provide an
entrée into the rapidly growing literature and to provide a foundation from which to discuss new empirical advances and future challenges. In the second
half of this article, I use empirical case studies to
highlight recent work on phenotypic integration
that offer promise in our quest for a synthetic understanding of complex characters. Specifically, I
draw attention to recent studies examining (1) how
environmental perturbation influences development
and morphological integration, (2) the influence of
maternal effects on integration within and between
generations, (3) approaches to investigate modules,
and (4) genomic approaches to studying the genetic
basis of integration. I end by highlighting new empirical approaches that hold promise for linking genomics, development, and environment in future
66
C. J. Murren
research on integration. Throughout this review,
I emphasize aspects of inquiry that have support in
both animal and plant systems. I also emphasize perspectives unique to plant or animal research in an
effort to encourage cross-fertilization between
approaches applied in these distinct clades of multicellular eukaryotic organisms.
Phenotypic integration and modularity
A module is a term applied broadly to describe an
aspect of an organism’s phenotype when there are
tight associations among traits within the complex
character, but the links with other character complexes are not as strong (see also Wagner 1996;
Schlichting and Pigliucci 1998; Murren 2002;
Hallgrı́msson and Hall 2005; Klingenberg 2010).
This pattern of associations of traits leads to a semiautonomous nature (termed modularity) in which
certain traits may interact with other modules depending on the function involved. Although
semi-independent, different modules of an organism
must at some level work together as a functional
whole organism.
The construction or maintenance of a phenotypic
module occurs through a combination of nonexclusive processes: shared or closely linked developmental
or genetic components (Schlichting 1989; Conner
2002; Roseman et al. 2009), or trade-offs resulting
from differential patterns of allocation (Tonsor and
Scheiner 2007). Suites of characters that work together as a particular functional unit such as a
flower (Conner and Sterling 1996), or a mandible
(Fig. 1) (Cheverud 1982) can be defined as functional modules. These modules may be composed
of additional submodules with specific subfunctions
(Murren 2002; Bissell and Diggle 2008). Organisms
with strong relationships (high correlations) between
the traits within the module are hypothesized to have
the most efficient functional performance (Brock and
Weinig 2007; Pélabon et al. 2011). Selection may act
on trait relationships, thus integration can be adaptive (Ordano et al. 2008). Assuming that phenotypic
integration has an underlying heritable basis, albeit
complex, the potential evolutionary trajectories of
traits within a phenotypic module will be influenced
by other traits in the module and by the selective
environment.
Modules may have enhanced function when correlations among traits are high; thus selection may
act on integration. In addition, the extent of
Fig. 1 From Roseman et al. (2009). Line drawing depicting
(A) the two regions of the mouse mandible, ascending ramus
and alveolar region, and (B) 22 landmarks used in morphometric
analyses.
evolutionary flexibility of a complex character may
be determined by whether variation is modular or
integrated with other aspects of the organism’s phenotype (Wagner and Altenberg 1996; Roseman et al.
2009; Drake and Klingenberg 2010). Modularity may
enhance potential responses to selection, as
semi-autonomous sets of characters that perform different functions may respond independently to selection (Bissell and Diggle 2010). Alternatively, strong
integration among functional units may limit evolutionary change if selection acts in opposing directions on different aspects of a complex phenotype
with strong internal correlations (Wagner and
Altenberg 1996). Selection on phenotypic interaction
is, therefore, considered to favor modular genetic
and developmental systems (Hallgrı́msson et al.
2009), including hormonal regulation (Ketterson
et al. 2009). Although I have described modules
from a largely phenotypic perspective, modules and
integration may be considered from other biological
viewpoints, including developmental, functional, and
genetic perspectives. In practice, how we define functional modules when the limits of a functional
module occur will influence our interpretations of
their evolution (see approaches to the study of modules below and Fornoni et al. 2009).
67
Phenotypic integration
A synthetic theory of phenotypic integration and
modularity has yet to emerge (Arnold 2005). From a
quantitative genetics perspective, evaluation of the
appropriateness of the G-matrix has formed much
of the theoretical basis of integration. These investigations were inspired in large part by Cheverud
(1982). Quantitative genetic models have been instrumental in developing hypotheses of multivariate
selection on suites of characters, and new tools for
matrix comparison may direct future work from this
perspective (Arnold 2005; Marroig et al. 2011).
However, other aspects of theory of integration are
not as well developed. As eloquently described by
Moczek, a synthetic theory of developmental evolution is yet to emerge. Development and epigenetic
approaches are at the heart of much of the contemporary
work
on
phenotypic
integration
(Hallgrı́msson et al. 2009). The conceptual model
of G!P map advocated by Wagner (1996),
Wagner and Altenberg (1996), and Pigliucci (2010)
describes links between sets of genes and sets of
traits. This conceptual framework is particularly
valuable in light of the rapidly expanding genomic
data (e.g., see discussion of quantitative trait
loci (QTL) approaches below). As an extension of
this model, Klingenberg (2003) included development as an additional conceptual component.
He added that the developmental system (as well as
environmental system) may modulate the path between sets of genes and phenotypic modules.
Empirical investigations of integration currently
are rapidly accumulating in the literature, thus
the majority of this review emphasizes empirical
work.
(Hallgrı́msson et al. 2009). Pleiotropy, therefore,
can be an important mechanism in shaping or constraining
patterns
of
multitrait
evolution.
Alternatively, LD among physically unlinked genes
may arise through natural selection of particular
combinations of alleles that enhance functional efficiency (Conner and Sterling 1995).
Morphological integration
A strength of research on integration in animal biology has been the study of morphological
shape-based phenotypes (Raff 1996; Klingenberg
2010). The emphasis on morphological integration
follows from the tradition of Olson and Miller
(1958). Studies of morphological integration largely
rely on data collected on landmark positions—precise locations of particular morphological features
that can be identified in all organisms examined in
a study (Fig. 1) (Cooper et al. 2011). To examine the
evolution of form, these landmark data sets are often
analyzed through Procrustes superimposition—a
method that employs the landmark information to
extract variation in shape, standardized for size and
orientation (for more detailed information, see
Klingenberg 2010; Viscosi and Cardini 2011). These
tools have been widely employed to examine
morphologies
of
terrestrial
invertebrates
(Klingenberg et al. 2010) and vertebrates (Young
and Badyaev 2010). Recently, this technique has
been implemented to study the shape of a compound
leaf (Klingenberg et al. 2012), suggesting that certain
aspects of plant form, such as flowers (for methodology, see van der Niet et al. 2010; Ordano et al.
2008; Benitez-Vieyra et al. 2009), may be well
suited for landmark analyses.
Genetics of phenotypic integration
Phenotypic modules may also have associated underlying genetic modularity, primarily through pleiotropy or genetic linkage (Cheverud 1982; Wagner
1996; Cheverud et al. 1997; Juenger et al. 2000;
Cheverud 2004; Murren and Kover 2004). Genetic
basis of covariance or of correlation patterns
among phenotypic traits is attributed to aspects of
genetic architecture such as pleiotropy (or pleiotropic QTL), close linkage, or linkage disequilibrium
(LD) among physically unlinked genes (Cheverud
2004; Murren and Kover 2004; Roseman et al.
2009). A single pleiotropic gene or QTL region simultaneously influences the variation of several traits
that can act through developmental processes
Development and modularity
Development has been considered a key component
of the study of integration since the formalization of
the concept (Olson and Miller 1958). Environmental
or genetic perturbations can shift developmental
pathways, often resulting in phenotypic changes.
Divergence in timing of floral development between
two species of Clarkia—one selfing and one outcrossing—that occupy discrete habitats that vary in
moisture availability suggests that selection acting on
developmental processes contributes to distinct patterns of phenotypic integration and potentially to
diversification (Runions and Geber 2000). Whether
separate aspects such as the timing and the rate of
68
development can be uncoupled (Runions and Geber
2000) will influence the evolutionary flexibility of
developmental traits and the resulting patterns of
integration. Often development and function
(Olson and Miller 1958) are discussed in tandem,
as these biological mechanisms are indistinguishable
in many studies. Two major approaches have attempted to disentangle development from function.
First, studies of developmental instability treat the
variation produced through random events during
development by examining the covariation of the
patterns of asymmetry among traits within individuals. An influence of shared development can be
concluded when high covariances among developmental instabilities are detected, as the environment
(generally) and genomics are held constant. Second,
disruption of genes known to influence development
across the phenotype of the organism may uncover
shared developmental processes when covariance
among multiple traits change with introduction of
variation into the system. Examinations of environmental perturbations on developmental plasticity
draw attention to the influence of epigenetics on
phenotypic integration (Hallgrı́msson et al. 2007,
2009).
Approaches to the study of modules
A large and growing literature exists on modularity
(Schlosser and Wagner 2004; Hallgrı́msson and Hall
2005; Hallgrı́msson et al. 2009; Klingenberg 2010).
Investigations into how complex characters are portioned into modules apply one of two main
approaches: examinations of modules identified
through data sets (data-defined) and those described
a priori (researcher defined).
Modules that are ‘‘data-defined’’ are identified
through a variety of statistical approaches that examine patterns of covariation of traits. Numerous statistical methodologies for delineating membership of
traits in modules as well as patterns of integration
within and between modules have been applied to
phenotypic data. These methods rely on measurable
variation in patterns of covariation and are particularly well described in the literature on morphological integration using landmarks (for example, but
not limited to Mitteroecker and Bookstein 2007;
Marquez 2008; Bissell and Diggle 2008; Magwene
2009; Bissell and Diggle 2010). These data-driven
approaches are valuable in uncovering modules
with specialized functions within a larger complex
C. J. Murren
of characters. For example, two character suites
with distinct functions (one pollinator attraction,
the other pollen transfer) within the flower were detected in two closely related species: Nicotiana alata
and Nicotiana forgetianta (Bissell and Diggle 2010).
Combined phenotypic and quantitative genetic analyses indicated that these modules are stable through
evolutionary time, yet the connections between modules are evolutionarily labile. Co-expression networks
built on vast microarray phenotypes (particularly for
crop and other model organisms) have been built to
elucidate putative gene modules (Ficklin et al. 2010).
These data-driven approaches have likewise been informative in highlighting a mechanistic basis for integration, including shared development, and shared
genetics when observable differences in patterns of
co-variation are observed (see below). Hallgrı́msson
et al. (2009) caution that unless phenotypic variance
exists in the sample, many developmental modules
may not be uncovered via analyses of patterns of
phenotypic (e.g., morphological) covariance data.
‘‘Researcher-defined’’ modules rely on prior research and knowledge of an organism to describe
phenotypic modules. Investigations of phenotypic integration that define functional modules a priori
often address particular hypotheses (e.g., Berg’s hypothesis described above regarding greater correlation among traits in floral vs vegetative modules
and associated with the pollinator-mediated selective
environment) (Fig. 2) (Armbruster et al. 2004;
Ordano et al. 2008; Pélabon et al. 2011), examine
changes through development or across environments in strength of integration or in phenotypic
composition of modules (Murren et al. 2002;
Bossdorf and Pigliucci 2009; Pélabon et al. 2011),
or assess differential functional roles of adjacent
parts (Klingenberg et al. 2010). Pélabon et al.
(2011) examined Berg’s hypothesis of decoupling of
floral and vegetative traits in two populations of
Dalechampia scandens (a neotropical vine) across
gradients of nutrients and of availability of light.
Dalechampia blossoms are in fact inflorescences
that include both a male and a female
subinflorescence (Fig. 2). The showy portion of the
blossom is an involucral bract (a modified leaf).
Examination of developmentally proximate structures, or of homologous structures with different
functions, enables contributions of function and development to be separated. Greater variation was
found in traits not associated with pollination than
in those involved in pollination. This pattern held
Phenotypic integration
69
Fig. 2 From Pelabon et al. Line drawings of Dalechampia scadens in a study of integration within floral and vegetative structures
and homologous structures. (A) The leaf and blossom. (B) Adaxial leaf and stipules (outgrowths underneath the leaf). (C) Blossom.
(D) Abaxial leaf and stipules. Bracts of blossom are modified leaves, and central portion of the blossom includes male and female
flowers making up an inflorescence.
across different environments, and their data were
consistent with Berg’s hypothesis of patterns of variation within and between in pollination traits in
comparison with vegetative traits. Patterns of trait
variation of homologous leaf and bract structures
in Dalachampia suggest that functional differences
between vegetative and pollination-related characters
are of greater importance than is proximity of
aspects of the phenotype or shared elements of a
developmental pathway. Thus, any potential developmental limits were likely broken by natural
selection.
A well-studied system employed to test hypotheses
of phenotypic integration is the mouse mandible.
The mandible has been described to have two functional regions: an alveolar region and an ascending
ramus region (Fig. 1) (Cheverud et al. 1997; Mezey
et al. 2000). The alveolar region is the area that includes the teeth. The ascending remus is the area of
muscle attachment and connection to the rest of the
skull (Cheverud et al. 1997; Roseman et al. 2009).
Roseman et al. (2009) challenged the notion that this
pair of phenotypic regions forms two discrete modules, as various investigations have failed to find patterns of integration and modularity that consistently
fall into one or the other of these two regions.
Employing the perspective that a shared genetic
basis defines modules (in this case shared QTL and
pleiotropy), Roseman et al. (2009) defined two classes of pleiotropic QTL: local and global. If QTL effects are distributed in a more narrow phenotypic
region (or among those traits that are highly correlated) than expected by chance, these are considered
locally pleiotropic. If QTL effects are more widely
distributed throughout the entire phenotype examined, these are considered globally pleiotropic. Using
a ‘‘data-defined’’ approach to examine the coherence
of two ‘‘researcher-defined’’ modules, two classes of
QTL emerged from the mouse-mandible data set.
The first class included a large proportion of pleiotropic QTL that are more highly correlated than due
to chance, and a second set distributed as if by
70
chance. The distribution of QTL effects in their data
set provides evidence favoring a conception of modularity based on physical proximity of traits along a
continuum both for phenotypic correlation of traits
and for QTL expression data sets.
Determining the relative merit of a ‘‘researcherdefined’’ or ‘‘data-defined’’ approach to the study
of modules, in part, depends on the particular hypotheses to be tested. The ‘‘data-defined’’ approach
may be instrumental in identifying QTL that vary in
importance through ontogeny (see also Hallgrı́msson
and Lieberman 2008), identifying ecological and evolutionarily important functional modules previously
undefined (Bissell and Diggle 2008, 2010) or in identifying traits that change functional roles during development and resulting in traits that vary in the
extent they overlap in function at different developmental stages and resulting in a continuum in variation of modularity, such as in the mouse mandible
(Roseman et al. 2009). Bissell and Diggle (2010) suggested that studies of broad functional modules
(such as the flower), defined a priori, may obfuscate
patterns of integration and may have led to the
mixed support for Berg’s hypotheses across systems.
Roseman et al. (2009) concluded that modules defined a priori restrict one’s ability to detect nuances
in patterns of the relationships of traits and the effects of pleiotropic QTL that may vary from patchy
to continuous. However, ‘‘researcher-defined’’ modules are appropriate when particular hypotheses of
locomotion, floral function, or physiology are addressed specifically.
Structural equation modeling (SEM) together with
model selection approaches, combine aspects of both
researcher-defined and data-driven perspectives
(Tonsor and Scheiner 2007; Bissell and Diggle
2010; Santos and Cannatella 2011) and may aid in
the assessment of putative functional modules. This
regression-based approach allows the testing of hypotheses of causal structure as well as discovery of
previously unknown relationships, and the potential
to compare patterns across environmental treatments
or classes of mutant (Tonsor and Scheiner 2007).
Applying SEM to aposematic, diet, growth rate,
and size traits in dendrobatid frogs revealed two phenotypic networks: one associated with aposematism
and the other associated with scale (Santos and
Cannatella 2011). Initial SEM models were built
based on a priori knowledge of aposematism and
body size and then compared against alternative
models that included different traits. Using a novel
C. J. Murren
phylogenetically informed analysis and ancestral reconstruction in this clade, Santos and Cannatella
(2011) were able to narrow down the alternative hypotheses for the evolution of the complex trait for
avoiding becoming prey to two. Phylogenetic
approaches, such as those presented by Santos and
Cannatella (2011) (see also Kolbe et al. 2011), offer a
hope for gaining an understanding of the evolutionary steps that result in complex traits with
behavioral, physiological, ecological, and size components. Combining the strengths of ‘‘researcher-defined’’ and ‘‘data-defined’’ modules in analyses may
be particularly important as we move beyond model
systems in evo–devo studies, and aim to evaluate
whether patterns of phenotypic covariance, and
their developmental and genetic bases, are
species-specific.
Developmental plasticity and
integration
Phenotypic plasticity is a term typically used to describe different phenotypes of a single trait, expressed
in response to the environment (Pigliucci and
Marlow 2001). Multiple traits may respond simultaneously to an environmental gradient; thus, patterns
of integration themselves may be plastic, uncovering
different suites of phenotypic variation (Schlichting
and Pigliucci 1998). Developmental processes may
contribute to these differential patterns of plasticity
and ultimately to diversification (Pfennig et al.
2010). Here I focus on two fresh perspectives that
link development and environment: (1) the effect of
short-term environmental perturbation during development and (2) the effect of longer-term maternal
effects on patterns of integration.
Environmental perturbations during development
Environmental perturbations in early development
can influence future development and ultimately
adult form. During pregnancy, an environmental
perturbation of reduced nutrient quality in the diet
influenced the patterns of integration of the rat skull
through ontogeny (Gonzalez et al. 2011a). These results emphasize how plastic responses may come
about through alternative developmental pathways
and compensatory mechanisms. In a companion
study, phenotypic integration through development
was examined following reduction in nutrients at
several points in development (Gonzalez et al.
2011b). In rats, the timing and duration of the
Phenotypic integration
nutritional environmental perturbation impacts the
changes in covariance structure. Although all three
main structures of the skull were reduced in size in
the low-nutrient treatment, whether the vault, face,
and base had the greatest change depended on the
time during development at which the environmental perturbation occurred. In both studies, environmental perturbation influenced variance in cranial
morphology, lending evidence to developmental plasticity as a critical influence on patterns of phenotypic
integration.
Maternal effects
Maternal effects influence patterns of integration of
traits across generations. In a study of a polymorphic
lizard species, Uta stansburiana, Landcaster et al.
(2010) provided evidence that maternal effects influence phenotypic integration of behavior with coloration of throat and dorsum, egg size, and shape of
the body. Moreover, female lizards influence offspring’s traits through levels of hormones adjusted
through environmental social cues of the throat
colors of neighbors and mate and level of stress.
The result is an appropriate pairing of dorsal and
throat colors, egg size, and anti-predator behavioral
and patterning traits in the offspring. Adaptive maternal effects in this system maintain phenotypic integration as beneficial combinations of traits would
otherwise be broken by recombination.
In a plant system, Galloway and Burgess (2009)
experimentally manipulated the timing of flowering
in Campanulastrum americana to examine the influence of phenology on reproductive integration and
life-history characters in the offspring generation. A
shift in a set of reproductive traits (including flowering, maturation of fruit, and dispersal) occurred
together regardless of whether flowering was experimentally shifted earlier or later. However, timing of
maternal reproduction influenced the frequency of
offspring expressing an annual life history.
Although plastic, the tight integration of reproductive characters suggests that responses of flowering to
climatic change would include shifts in the entire
suite of reproductive characters, and substantially
impact development and life history in the offspring
generation.
Genomic analyses via QTL approaches
QTL studies in both plants and animals point to
pleiotropy (or close linkage) as the primary
71
component of the genetic basis of integration
(Cheverud 2004). Co-localization of QTL for different traits is a metric for identifying genetic modules.
In the agricultural crop, oil rapeseed (Brassica rapa),
analyses of floral and vegetative traits in a set of RILs
(recombinant inbred lines) revealed high levels of
integration within the floral module and little differentiation between floral and vegetative modules
(Edwards and Weinig 2011). Edwards and Weinig
(2011) suggested pleiotropy or close physical linkage
as the likely genetic mechanisms behind these patterns of integration, given two lines of evidence.
Recombination across a large collection of RILs
reduces LD among physically distant genes.
Colocalization of QTL for vegetative and floral
traits is consistent with pleiotropy or close linkage,
two genetic mechanisms that are indistinguishable
when using a QTL approach. Cooper et al. (2011)
presented evidence for co-localized QTL for traits
with shared feeding function and skull shape in cichlids. In addition, there was evidence for QTL for
correlations among traits themselves. Two alternative
perspectives have been used to describe the evolutionary influence of pleiotropy. Pleiotropy may constrain the independent evolution of highly correlated
traits (Edwards and Weinig 2011), or contribute to
adaptive diversifications via phenotypic integration
(Cooper et al. 2011).
Genomic approaches to examining the genetic
basis of phenotypic integration over the past
decade have largely (although not entirely) focused
on QTL analyses. Future studies that include fine
mapping (Murren and Kover 2004; Edwards and
Weinig 2011) will help to disentangle the relative
importance of close linkage and pleiotropy.
Distinguishing these two effects are not possible
with current QTL methods. In combination with
the information from whole-genome projects in
model organisms and agricultural organisms (including crops, stock fish, livestock, and pest species),
combined QTL and fine-mapping studies may lead
to identification of candidate genes that open doors
to investigation of function and of timing of specific
genes and gene combinations.
A path ahead: Genomic, ecological, and
developmental perturbations
New tools on the horizon and novel application of
classical tools offer potentially novel avenues for examining genome-wide, ecological, and developmental
72
influences on patterns of phenotypic integration. For
example, using classical phenotypic approaches,
hybrid species provide a view into the influence of
broad changes in genomic composition on the evolution of integration, through the reshuffling of naturally occurring variation and through changes in
large components of the genome. In systems where
the parental and hybrids are extant, the contrast of
the genomic combinations can paint an evolutionary
picture of integration. In a study of seven species of
Brassica and closely related Raphanus that included
hybrids and their diploid relatives, hybrids were
more integrated both within and among modules
than was either of their parental species (Murren
et al. 2002). Yet, neither cytogenetics nor ecological
similarity explained differences among species in patterns of phenotypic correlations of morphological
and life-history traits. In part, these differences between parental species and hybrids may reflect the
long stable history of this hybrid complex. At the
other end of the hybrid spectrum, cichlid fishes in
East African rift lakes are known for their phylogenetic diversity and their propensity to hybridize in
ecological timescales (Parsons et al. 2011). Two species of cichlid fish from Lake Malawi that differ in
foraging mode and jaw width were experimentally
hybridized to examine the capacity of this genomic
perturbation for altering patterns of integration
(Parsons et al. 2011). Analyses of head shape (from
lateral and ventral views) uncovered greatly reduced
patterns of integration in hybrids compared with
either parent. Some hybrid individuals showed evidence of transgressive segregation—extreme phenotypes in the hybrids in comparison with parental
species. Those individuals that showed transgressive
segregation had further reductions in phenotypic integration. These patterns offer suggestion of how
hybridization can promote evolution of novel phenotypes, break up patterns of integration, and contribute to diversification in a rapidly evolving group
(Parsons et al. 2011). Further studies of hybrids are
needed to elucidate the contributions of introduction
of novel phenotypic variation compared with the selective environment. These studies will be particularly fruitful in groups such as Brassica and cichlids
for which genomic and ecological data are plentiful
and growing, and for which experimental manipulations are possible.
Another successful tool in the study of phenotypic
integration is the introduction of novel phenotypic
variation by way of new mutations that disrupt
C. J. Murren
developmental pathways. One such approach is
EMS mutagenesis—a method that induces new mutations in an experimentally tractable time frame,
introduces novel genetic variation, and potentially
alters trait means and their patterns of phenotypic
integration. Camara et al. (2000) demonstrated that
EMS mutagenesis in A. thaliana altered patterns of
integration by changing an array of particular
pair-wise correlations of traits. Interestingly, an especially strong and well-studied correlation between
flowering time and number of rosette leaves was
less labile in the mutagenized lines than were other
relationships of traits. Other more recent approaches
to genetic perturbations, such as the introduction of
introgression of mutations that influence wing shape
in various genetic backgrounds of Drosophila melanogaster (Dworkin et al. 2009), and comparison of
mutants in Mus musculus known to influence aspects
of the complex craniofacial phenotype with wild-type
M. musculus and its close relatives (Jamniczky and
Hallgrı́msson 2009), may offer insights into the influence of gene–gene interactions into patterns of
phenotypic integration.
Model multicellular organisms such as Arabidopsis,
Drosophila, and Mus are especially well suited for
examination of the effects of genomic perturbation
on patterns of phenotypic integration, as many genomic data are publicly available, and genetic lines
are available at stock centers (e.g., ABRC,
Bloomington and DSSC Drosophila species stock
center). One such resource includes the T-DNA
knockouts in A. thaliana (e.g., SALK T-DNA)
(O’Malley and Ecker 2010). This set of early 17,000
T-DNA knockout lines includes single-gene knockouts all in the same genetic background. Although
the genomic resource is available, phenotypic data
are rare, largely lacking, or rarely detected
(Kuromori et al. 2006; Hanada et al. 2009; see
Hillenmeyer et al. 2008 for a combined data set of
phenotypic and genomic data in yeast). unPAK (undergraduates phenotyping Arabidopsis knockouts
www.arabidopsisunpak.org) is a network of laboratories investigating the influence of single-gene knockouts on complex phenotypic characters such as
germination success, size of rosettes and inflorescences, fruit production, overall plant biomass, and
ultimately patterns of integration. Phenotyping with
the ultimate goal of full genome coverage is intensive
both of time and labor. The unPAK approach is
much like a randomized complete block design
(Gotelli and Ellison 2004) with plants grown and
73
Phenotypic integration
phenotyped through time, across experimentally implemented treatments, and across multiple laboratories. Numerous undergraduate researchers at each
institution employ shared protocols, deposit data
into a shared database, and contribute novel insights
and observations. In this age of the phenotype, collecting phenotypic data such as those planned by
unPAK, is the next frontier in investigation of questions of phenotypic integration and of the role of
developmental plasticity.
Evaluating epigenetics as a contributor to heritable
phenotypic diversity (West-Eberhard 2003; Richards
et al. 2010), and trans-generational plasticity are key
to translating environmental influences on integration. Advances in detection of differential patterns
of DNA methylation among genotypes (Lister et al.
2009; Becker et al. 2011) and experimental alteration
of DNA methylation (Bossdorf et al. 2010) put
within reach a greater understanding of the influence
of heritable epigenetic polymorphisms on patterns of
phenotypic integration.
It must be emphasized that additional environmental (Pélabon et al. 2011) and developmental perturbation studies (Gonzalez et al. 2011a, 2011b) are
much needed across diverse systems. These types of
studies aid in the understanding of these influences
on changes in patterns of integration and modular
structure and would allow synthesis across multicellular eukaryotes. Short- and long-term perturbation
studies are complementary and would provide data
on buffering and changes in developmental modularity (sensu Klingenberg 2003). Investigation of
nonmorphological modules, such as defense traits
(Huang et al. 2010), is also needed to evaluate
whether common themes exist across classes of phenotypes. Comparisons between existing pairs of
strains, for example, wild types and agricultural lineages specifically selected for defense traits (e.g.,
transgenic plants such as Bt corn), provide additional
systems that have been perturbed genetically, uncovering variation, and appropriate for phenotypic
approaches. Model approaches (such as SEM) will
aid in the assessment of module composition.
Further joining theoretical advances that formally include developmental, genomic, and environmental
variation into our empirical understanding of modules and integration will be an important next step.
The age of genomics made possible the tools to
experimentally introduce genetic variation in myriad
ways and, more recently, to modify epigenetics
through changes in DNA methylation. In the age
of the phenotype, characterizing phenotypes across
environments and through development, while employing tools that perturb genomic, ecological, or
developmental systems will shape the next chapter
in the quest for understanding how complex characters and their relationships are built and maintained
through evolutionary time.
Acknowledgments
I thank Matthew Wund and Armin Moczek for organizing this symposium at Society for Integrative
and Comparative Biology, 2012 and for the invitation to participate. I gratefully acknowledge SICB
and the DAB, DCE, DEDB, DEE, DPCB (formerly
DSEB), DIZ, and DVM sections. Inspiring talks by
other symposium speakers on complex phenotypes
and lively conversations with members of both the
Polyphenism working group and the Costs of
Plasticity working group at NESCENT are gratefully
acknowledged. I also acknowledge Matt Rutter, Allan
Strand, and Hilary Callahan for discussion of this
topic and Matt Rutter for critiques of earlier drafts.
Funding
I gratefully acknowledge financial support from the
National Science Foundation and Society for
Integrative and Comparative Biology for the symposium. This work was supported by a NESCent
short-term fellowship and by the National Science
Foundation (IOS 1052262 to Rutter, C.M., and
Strand; IOS 1146977 to C.M.).
References
Alföldi J, Di Palma F, Grabherr M, Williams C, Kong L,
Mauceli E, Russell P, Lowe CB, Glor RE, Jaffe JD, et al.
2011. The genome of the green anole lizard and a comparative analysis with birds and mammals. Nature 477:587–91.
Armbruster WS, Di Stilio VS, Tuxill JD, Flores TC, Velasquez
Runk JL. 1999. Covariance and decoupling of floral and
vegetative traits in nine neotropical plants: a reevaluation
of Berg’s correlation-pleiades concept. Am J Bot 86:39–55.
Armbruster WS, Pélabon C, Hansen TF, Mulder CPH. 2004.
Floral
integration,
modularity
and
accuracy.
In: Pigliucci M, Preston K, editors. Phenotypic integration.
New York: Oxford University Press.
Arnold SJ. 2005. The ultimate causes of phenotypic integration: lost in translation. Evolution 59:2059–61.
Becker C, Hagmann J, Koenig D, Stegle O, Borgwardt K,
Weigel D. 2011. Spontaneous epigenetic variation in the
Arabidopsis thaliana methylome. Nature 480:245–49.
Benitez-Vieyra S, Medina AM, Cocucci AA. 2009. Variable
selection patterns on the labellum shape of Geoblasta
74
pennicillata, a sexually deceptive orchid. J Evol Biol
22:2345–62.
Berg RL. 1960. The ecological significance of correlation
pleiades. Evolution 14:171–80.
Bissell EK, Diggle PK. 2008. Floral morphology in Nicotiana:
architectural and temporal effects on phenotypic integration. Int J Plant Sci 169:225–40.
Bissell EK, Diggle PK. 2010. Modular genetic architecture of
floral morphology in Nicotiana: quantitative genetic and
comparative phenotypic approaches to floral integration. J
Evol Biol 23:1744–58.
Bossdorf O, Acurri D, Richards CL, Pigliucci M. 2010.
Experimental alterations of DNA methylation affects the
phenotypic plasticity of ecologically relevant traits in
Arabidopsis thaliana. Evol Ecol 24:541–53.
Bossdorf O, Pigliucci M. 2009. Plasticity to wind is modular
and genetically variable. Evol Ecol 23:669–85.
Brock MT, Weinig C. 2007. Plasticity and environmentspecific covariances: an investigation of floral-vegetative
and within flower correlations. Evolution 61:2913–24.
Callahan HS, Waller DM. 2000. Phenotypic integration and
the plasticity of integration in an amphicarpic annual. Int J
Plant Sci 161:89–98.
Camara MD, Ancel CA, Pigliucci M. 2000. Induced mutations: a novel tool to study phenotypic integration and
evolutionary constraints in Arabidopsis thaliana. Evol Ecol
Res 2:1009–29.
Cheng KC, Xin X, Clark DP, La Riviere P. 2011. Whole
animal imaging, gene function, and the zebrafish phenome
project. Curr Opin Genet Dev 21:620–29.
Chernoff B, Magwene PM. 1999. Afterward- morphological
integration: forty years later. In: Olson EC, Miller RL, editors. Morphological integration. Chicago: University of
Chicago Press. p. 319–53.
Cheverud JM. 1982. Phenotypic, genetic and environmental
morphological integration in the cranium. Evolution
36:499–16.
Cheverud JM. 1996. Developmental integration and the evolution of pleiotropy. Am Zool 36:44–50.
Cheverud J. 2004. Modular pleiotropic effects of quantitative
trait loci on morphological traits. In: Schlosser G,
Wager GP, editors. Modularity in development and evolution. Chicago: University of Chicago Press. p. 132–53.
Cheverud JM, Routman EJ, Irschick DJ. 1997. Pleiotropic
effets of individual gene loci on mandibular morphology.
Evolution 51:2006–16.
Clausen J, Hiesey WM. 1958. Experimental studies in the
nature of species: IV. Genetic structure of the ecological
races. Washington DC: Carnegie Institution of
Washington. Publication 615.
Clausen J, Hiesey WM. 1960. The balance between coherence
and variation in evolution. Proc Natl Acad Sci USA
46:494–506.
Clausen J, Keck DD, Hiesey WM. 1940. Experimental studies
on the nature of species: effect of varied environments on
Western North American plants. Washington: Carnegie
Institute.
C. J. Murren
Conner JK. 2002. Genetic mechanisms of floral trait correlations in a natural population. Nature 420:407–10.
Conner JK, Sterling A. 1995. Testing hypotheses of functional
relationships: a comparative survey of correlation patterns
among floral traits in 5 insect pollinated plants. Am J Bot
82:1399–406.
Conner JK, Sterling A. 1996. Selection for independence of
floral and vegetative traits: evidence from correlation patterns in five species. Can J Bot 74:642–44.
Cooper WJ, Wernle J, Mann K, Albertson RC. 2011.
Functional genetic integration in the skulls of Lake
Malawi cichlids. Evol Biol 38:316–34.
Drake AG, Klingenberg CP. 2010. Large scale diversification
of skull shape in domestic dogs: disparity and modularity.
Am Nat 175:289–301.
Dworkin I, Kennerly E, Tack D, Hutchinson J, Brown J,
Mahaffey J, Gibson G. 2009. Genomic consequences of
background effects on scalloped mutant expressivity in
the wing of Drosophila melanogaster. Genetics 181:1065–76.
Edwards CE, Weinig C. 2011. The quantitative-genetic and
QTL architecture of trait integration and modularity in
Brassica rapa across simulated seasonal settings. Heredity
106:661–77.
Ficklin SP, Luo F, Feltus FA. 2010. The association of multiple interacting genes with specific phenotypes in rice using
gene coexpression networks. Plant Physiol 154:13–24.
Fornoni J, Ordano M, Boerge K, Domı́nguez CA. 2009.
Phenotypic integration: between zero and how much is
too much. New Phytol 183:248–50.
Furbank RT, Tester M. 2011. Phenomics – technologies to
relieve the phenotyping bottleneck. Trends Plant Sci
16:635–44.
Galloway LF, Burgess KS. 2009. Manipulation of flowering
time: phonological integration and maternal effects.
Ecology 90:2139–48.
Gravassa S, Silva AC, Stoddard PK. 2011. Tight hormonal
phenotypic integration ensures honesty of the electric
signal of male and female Brachyhypopomus gauderio.
Horm Behav 60:420–26.
Gonzalez PN, Hallgrı́msson B, Oyhenart EE. 2011a.
Developmental plasticity in covariance structure of the
skull: effects of prenatal stress. J Anat 218:243–57.
Gonzalez PN, Oyhenart EE, Hallgrı́msson B. 2011b. Effects of
environmental perturbations during postnatal development
on the phenotypic integration of the skull. J Exp Zool
316:547–61.
Gotelli N, Ellison AM. 2004. A primer in ecological statistics.
Sunderland, MA: Sinauer.
Hallgrı́msson B, Hall BK. 2005. Variation: a central concept in
biology. New York: Elsevier Academic Press.
Hallgrı́msson B, Jamniczky H, Young NM, Rolian C,
Parsons TE, Boughner JC, Marcucio RS. 2009.
Deciphering the palimpsest: studying the relationship between morphological integration and phenotypic covariation. Evol Biol 36:355–76.
Hallgrı́msson B, Lieberman DE. 2008. Mouse model and the
evolutionary developmental biology of the skull. Int Comp
Biol 48:373–84.
Phenotypic integration
Hallgrı́msson B, Lieberman DE, Liu W, FordHutchinson AF, Jirik FR. 2007. Epigenetic interactions
and the structure of phenotypic variation in the cranium.
Evol Dev 9:76–91.
Hanada K, Kuromori T, Myouga F, Toyoda T, Li W-H,
Shinozaki K. 2009. Evolutionary persistence of functional
compensation by duplicate genes in Arabidopsis. Genome
Biol Evol 1:409–14.
Hansen TF, Pélabon C, Armbruster WS. 2007. Comparing
variational properties of homologous floral and vegetative
characters in Dalechampia scandens: testing the Berg hypothesis. Evol Biol 34:86–98.
Hillenmeyer ME, Fung E, Wildenhain J, Pierce SE, Hoon S,
Lee W, Proctor M, St. Onge MP, Tyers M, Koller D, et al.
2008. The chemical genomic portrait of yeast: uncovering a
phenotype for all genes. Science 320:362–65.
Houle D, Govindaraju DR, Omholt S. 2010. Phenomics: the
next challenge. Nat Rev Genet 11:855–66.
Huang M, Abel C, Sohrabi R, Petri J, Haupt I, Cosimano J,
Gershenzon J, Tholl D. 2010. Variation of herbivoreinduced volatile terpenes among Arabidopsis ecotypes depends on allelic differences and subcellular targeting of two
terpene synthases, TPS02 and TPS03. Plant Physiol
153:1293–301.
Jamniczky HA, Hallgrı́msson B. 2009. A comparison of covariance structure in wild and laboratory muroid crania.
Evolution 63:1540–56.
Juenger T, Purugganan M, Mackay TFC. 2000. Quantitative
trait loci for floral morphology in Arabidopsis thaliana.
Genetics 156:1379–92.
Ketterson ED, Atwell JW, McClothlin JW. 2009. Phenotypic
integration and independence: hormones, performance and
response to environmental change. Integr Comp Biol
49:365–79.
Klingenberg CP. 2003. A developmental perspective on developmental instability: theory, models and mechanisms.
In: Polak M, editor. Developmental instability, causes and
consequences. New York: Oxford University Press.
Klingenberg CP. 2010. Evolution and development of shape:
integrating quantitative approaches. Nat Rev Genet
11:623–35.
Klingenberg CP, Debat V, Roff DA. 2010. Quantitative genetics of shape in cricket wings: developmental integration in a
functional structure. Evolution 64:2935–51.
Klingenberg CP, Duttke S, Whelan S, Kim M. 2012.
Developmental plasticity, morphological variation and
evolvability: a multilevel analysis of morphometric integration in the shape of compound leaves. J Evol Biol
25:115–29.
Kolbe JJ, Revell LJ, Szekely B, Brodie ED III, Losos JB. 2011.
Convergent evolution of phenotypic integration and its
alignment with morphological diversification in Caribbean
Anolis ecomorphs. Evolution 65:3608–24.
Kuromori T, Wada T, Kamiya A, Yuguchi M, Yokouchi T,
Imura Y, Takabe H, Sakurai T, Akiyama K, Hirayama T,
et al. 2006. A trial of phenome analysis using 4000
Ds-insertional mutants in gene-coding regions of
Arabidopsis. Plant J 47:640–51.
75
Landcaster LT, McAdam AG, Sinervo B. 2010. Maternal adjustment of egg size organizes alternative escape behaviors,
promoting adaptive phenotypic integration. Evolution
64:1607–21.
Lister R, Gregory BD, Ecker JR. 2009. Next is now: new technologies for sequencing of genomes, transcriptomes, and
beyond. Curr Opin Plant Biol 12:107–18.
Magwene PM. 2009. Statistical methods for studying modularity: a reply to Mitteroecker and Bookstein. Syst Biol
58:146–49.
Marquez EJ. 2008. A statistical framework for testing
modularity in multidimensional data. Evolution 62:
2688–708.
Marroig G, Melo D, Porto A, Sebastiao H, Garcia G. 2011.
Selection response decomposition (SRD): a new tool for
dissecting differences and similarities between matrices.
Evol Biol 38:225–41.
Mezey J, Cheverud J, Wagner G. 2000. Is the genotypephenotype map modular? An approach using quantitative
trait locus (QTL) data. Genetics 156:305–11.
Mitteroecker P, Bookstein F. 2007. The conceptual and statistical relationship between modularity and morphological
integration. Syst Biol 56:818–36.
Murren CJ. 2002. Phenotypic integration in plants. Plant
Species Biol 17:89–99.
Murren CJ, Kover PX. 2004. QTL mapping: a first step
toward an understanding of molecular genetic mechanisms
behind phenotypic complexity/integration. In: Pigliucci M,
Preston K, editors. Phenotypic integration. New York:
Oxford University Press.
Murren CJ, Pendelton N, Pigliucci M. 2002. Phenotypic integration in several closely related species of Brassica. Am J
Bot 89:655–63.
Olson EC, Miller RL. 1958. Morphological integration.
Chicago: University of Chicago Press.
O’Malley RC, Ecker JR. 2010. Linking genotype to phenotype in the Arabidopsis unimutant collection. Plant J
61:928–40.
Ordano M, Fornoni J, Boege K, Dominguez CA. 2008. The
adaptive value of phenotypic floral integration. New Phytol
179:1183–92.
Parsons KJ, Son YH, Albertson RC. 2011. Hybridization promotes evolvability in African cichlids: connections between
transgressive segregation and phenotypic integration. Evol
Biol 38:306–15.
Pélabon C, Armbruster WS, Hansen TF. 2011. Experimental
evidence for the Berg hypothesis: vegetative traits are more
sensitive than pollination traits to environmental variation.
Funct Ecol 25:247–57.
Pfennig DW, Wund MA, Snell-Rood EC, Cruickshank T,
Schlichting CD, Moczek AP. 2010. Phenotypic plasticity’s
impacts on diversification and speciation. Trends Ecol Evol
25:459–67.
Pigliucci M. 2010. Genotype-phenotype mapping and the end
of the ‘genes as blueprint’ metaphor. Philos Trans R Soc
Lond B 365:557–66.
Pigliucci M, Marlow ET. 2001. Differentiation for flowering
time and phenotypic integration in Arabidopsis thaliana
76
response to season length and vernalization. Oecol
127:501–08.
Pigliucci M, Preston K (eds.) 2004. Phenotypic integration:
studying the ecology and evolution of complex phenotypes.
New York: Oxford University Press.
Raff RA. 1996. The shape of life: genes, development and evolution of animal form. Chicago: University of Chicago Press.
Richards CL, Bossdorf O, Pigliucci M. 2010. What role does
heritable epigenetic variation play in phenotypic evolution?
BioScience 60:232–37.
Roseman CC, Kenny-Hunt JP, Cheverud JM. 2009.
Phenotypic integration without modularity: testing hypotheses about the distribution of pleiotropic quantitative trait
loci in a continuous space. Evol Biol 36:282–91.
Runions CJ, Geber MA. 2000. Evolution of the self-pollinating
flower in Clarkia xantiana (Onagraceae). I. Size and development of floral organs. Am J Bot 87:1439–51.
Rutter MT, Cross KV, Van Woert PA. 2012. Birth, death and subfunctionalization in the Arabidopsis genome. Trends Plant Sci.
published online (doi:10.1016/J.tplants.2012.01.006).
Santos JC, Cannatella DC. 2011. Phenotypic integration
emerges from aposematism and scale in poison frogs.
Proc Natl Acad Sci 108:6175–80.
Schlichting CD. 1989. Phenotypic integration and environmental change. BioScience 39:460–64.
Schlichting CD, Pigliucci M. 1998. Phenotypic evolution: a
reaction norm perspective. Sunderland MA: Sinauer
Associates.
C. J. Murren
Schlosser G, Wagner GP. 2004. Modularity in development
and evolution. Chicago: University of Chicago Press.
Sozzani R, Benfey PN. 2011. High-throughput phenotyping of
multicellular organisms: finding the link between genotype
and phenotype. Genome Biol 12:219.
Thompson DW. 1917. On growth and form. Cambridge:
Cambridge University Press.
Tonsor SJ, Scheiner SM. 2007. Plastic trait integration across a
CO2 gradient in Arabidopsis thaliana. Am Nat 169:E119–40.
van der Niet T, Zollikofer CPE, Ponce de Leon MS,
Johnson SD, Linder HP. 2010. Three dimensional geometric morphometrics for studying floral shape variation.
Trends Plant Sci 15:423–26.
Viscosi V, Cardini A. 2011. Leaf morphology, taxonomy and
geometric morphometrics: a simplified protocol for beginners. Plos One 6:e25630.
Wagner GP. 1996. Homologues, natural kinds and the evolution of modularity. Am Zool 36:36–43.
Wagner GP, Altenberg L. 1996. Perspective: complex adaptations and the evolution of evolvability. Evolution
50:967–76.
West-Eberhard MJ. 2003. Developmental plasticity and evolution. New York: Oxford University Press.
Young RL, Badyaev AV. 2010. Developmental plasticity links
local adaptation and evolutionary diversification in foraging
morphology. J Exp Zool 314B:434–44.