Journal of Experimental Botany, Vol. 63, No. 16,
2, pp.
2012
pp.695–709,
5741–5749,
2012
doi:10.1093/jxb/err313
doi:10.1093/jxb/ers267 Advance Access publication 4 November, 2011
This paper is available online free of all access charges (see http://jxb.oxfordjournals.org/open_access.html for further details)
RESEARCH
FloweringPAPER
Newsletter Review
Lessons
fromoceanica
flower colour
evolution
on changes
targets ofinselection
In
Posidonia
cadmium
induces
DNA
methylation and chromatin patterning
Carolyn A. Wessinger and Mark D. Rausher*
Department
of Biology,
Duke
University, Durham,
NC 27708,
Maria
Greco,
Adriana
Chiappetta,
Leonardo
BrunoUSA
and Maria Beatrice Bitonti*
* To whom correspondence should be addressed: E-mail: [email protected]
Department of Ecology, University of Calabria, Laboratory of Plant Cyto-physiology, Ponte Pietro Bucci, I-87036 Arcavacata di Rende,
Cosenza, Italy
Received 29 May 2012; Revised 28 August 2012; Accepted 31 August 2012
* To whom correspondence should be addressed. E-mail: [email protected]
Received 29 May 2011; Revised 8 July 2011; Accepted 18 August 2011
The genetic basis of flower colour evolution provides a useful system to address the debate over the relative contribution of regulatory vs. functional mutations in evolution. The relative importance of these two categories depends on
Abstract
the type of flower colour transition and the genes involved in those transitions. These differences reflect differences
In
mammals,
is widely
considered
as a non-genotoxic
through
a methylation-dependent
in the
degree cadmium
of deleterious
pleiotropy
associated
with functionalcarcinogen
inactivationacting
of various
anthocyanin
pathway genes.
epigenetic
mechanism.
Here,
the
effects
of
Cd
treatment
on
the
DNA
methylation
patten
are
examined
together
with
Our findings illustrate how generalized statements regarding the contributions of regulatory and
functional
mutations
its
effect
on
chromatin
reconfiguration
in
Posidonia
oceanica.
DNA
methylation
level
and
pattern
were
analysed
in
to broad categories of traits, such as morphological vs. physiological, ignore differences among traits within catactively
growing
organs,
under
short(6
h)
and
long(2
d
or
4
d)
term
and
low
(10
mM)
and
high
(50
mM)
doses
of
Cd,
egories and in doing so overlook important factors determining the relative importance of regulatory and functional
through
a Methylation-Sensitive Amplification Polymorphism technique and an immunocytological approach,
mutations.
respectively. The expression of one member of the CHROMOMETHYLASE (CMT) family, a DNA methyltransferase,
was
also assessed
by qRT-PCR.
ultrastructure
investigated by transmission electron
Key words:
Adaptive evolution,
adaptiveNuclear
mutation,chromatin
anthocyanins,
flower colour,was
pleiotropy.
microscopy. Cd treatment induced a DNA hypermethylation, as well as an up-regulation of CMT, indicating that de
novo methylation did indeed occur. Moreover, a high dose of Cd led to a progressive heterochromatinization of
interphase nuclei and apoptotic figures were also observed after long-term treatment. The data demonstrate that Cd
perturbs the DNA methylation status through the involvement of a specific methyltransferase. Such changes are
Introduction
linked to nuclear chromatin reconfiguration likely to establish a new balance of expressed/repressed chromatin.
Overall,
the data show
an epigenetic
the mechanism
underlying
toxicity inoffered
plants.
The
standardCd
explanation
for this pattern is that morNovel phenotypes
are ultimately
derivedbasis
from to
mutations
that
are favoured by natural selection. An important goal of evolu-
phological and physiological traits differ categorically in their
tionary change. Only recently, with the widespread availability
of molecular genetic and genomic tools, has it become possible
to address this goal. Documentation of genetic changes involved
Introduction
with phenotypic change in diverse taxa is accumulating (for
summaries see Hoekstra and Coyne, 2007; Stern and Orgogozo,
In the Mediterranean coastal ecosystem, the endemic
2008). These data have sparked a vigorous scientific dialogue
seagrass Posidonia oceanica (L.) Delile plays a relevant role
regarding the nature of genetic mutations that contribute to
by ensuring primary production, water oxygenation and
evolutionary change (Stern, 2000; Hoekstra and Coyne, 2007;
provides niches for some animals, besides counteracting
Wray, 2007; Carroll, 2008; Stern and Orgogozo, 2008; Wagner
coastal erosion through its widespread meadows (Ott, 1980;
and Lynch, 2008; Streisfeld and Rausher, 2010). This debate
Piazzi et al., 1999; Alcoverro et al., 2001). There is also
centres largely on whether phenotypic changes proceed primarconsiderable evidence that P. oceanica plants are able to
ily through regulatory changes or primarily through changes to
absorb and accumulate metals from sediments (Sanchiz
coding sequences that affect protein function, and whether the
et al., 1990; Pergent-Martini, 1998; Maserti et al., 2005) thus
answer to this question depends on the type of trait involved.
influencing metal bioavailability in the marine ecosystem.
Broad comparisons across many taxa (e.g. Stern and Orgogozo,
For this reason, this seagrass is widely considered to be
2008) have suggested that the evolution of morphological traits
a metal bioindicator species (Maserti et al., 1988; Pergent
proceeds primarily through regulatory mutations, while the evoet al., 1995; Lafabrie et al., 2007). Cd is one of most
lution of physiological traits more often involves functional
widespread heavy metals in both terrestrial and marine
mutations to coding sequences.
environments.
morphological traits are specified by genes that act upstream in
developmental pathways, regulatory changes with subtle effects
on gene action in a restricted set of tissues are favoured because
they minimize adverse disruption to development. Conversely,
since physiological traits are specified by genes near the tips of
Although not essential for plant growth, in terrestrial
developmental pathways, functional mutations that occur more
plants, Cd is readily absorbed by roots and translocated into
frequently than regulatory mutations will by nature have a more
aerial organs while, in acquatic plants, it is directly taken up
limited scope of action and be tolerated. However, Hoekstra
by leaves. In plants, Cd absorption induces complex changes
and Coyne (2007) point out that the categorization of traits as
at the genetic, biochemical and physiological levels which
‘morphological’ vs. ‘physiological’ is somewhat artificial, and
ultimately account for its toxicity (Valle and Ulmer, 1972;
that we need an unbiased framework for understanding when
Sanitz di Toppi and Gabrielli, 1999; Benavides et al., 2005;
we should predict regulatory or functional mutations to preWeber et al., 2006; Liu et al., 2008). The most obvious
dominantly contribute to evolutionary change in a given trait.
symptom of Cd toxicity is a reduction in plant growth due to
In particular, we need to statistically determine whether the
an inhibition of photosynthesis, respiration, and nitrogen
evolution of a trait occurs through preferential fixation of cermetabolism, as well as a reduction in water and mineral
tain mutations over alternatives. This endeavour requires the
uptake (Ouzonidou et al., 1997; Perfus-Barbeoch et al., 2000;
intensive study of genetic changes involved in multiple evoluShukla et al., 2003; Sobkowiak and Deckert, 2003).
tionary occurrences of individual traits (Kopp, 2009; Streisfeld
At the genetic level, in both animals and plants, Cd
and Rausher, 2010). Yet in the broad surveys that have been
can induce chromosomal aberrations, abnormalities in
Key
words:
5-Methylcytosine-antibody,
cadmium-stress
condition,
chromatin
reconfiguration,
position
within
developmentalCHROMOMETHYLASE,
pathways (Hoekstra and Coyne,
tionary
genetics
is to understand, and possibly
predict, whether
DNA-methylation,
MethylationSensitive
Amplification
Polymorphism
(MSAP),
Posidonia
oceanica
(L.) Delile.
2008).
It is argued that because
certain types of mutations are preferentially involved in evolu- 2007; Stern and Orgogozo,
© 2011
The Author
[2012]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved.
ª
The Author(s).
For Permissions,
please article
e-mail:distributed
[email protected]
This
is an Open Access
under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/bync/3.0), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Downloaded from http://jxb.oxfordjournals.org/ at Duke University on April 2, 2013
Abstract
5742 | Wessinger and Rausher
evolutionary change in flower colour. More direct evidence indicating that changes in flower colour are adaptive is provided by a
growing body of investigations directly demonstrating the operation of natural selection on flower colour variation (reviewed by
Rausher, 2008; Waser and Price, 1981; Jones and Reithel, 2001;
Schemske and Bierzychudek, 2001, 2007; Irwin and Strauss,
2005; Streisfeld and Kohn, 2007; Hopkins and Rausher, 2012).
Here we review recent progress using the anthocyanin pathway as a model system for understanding when and why we can
predict regulatory or functional genetic changes to predominate
in the evolution of flower colour.
Mutation rate and pleiotropy
For any given trait, two factors are likely to influence the relative
contribution of regulatory and functional mutations to adaptive
evolution: the relative rates of different mutations and their relative fitness effects. In the absence of fitness differences among
various mutations, mutations that arise more frequently should
predominantly contribute to adaptive divergence. The relative mutation rates for regulatory vs. functional mutations will
depend on the evolutionarily relevant mutational ‘target size’ of
each. This target size will reflect: (1) the number of genes in
each category; and (2) the number of nucleotide sites per gene
that, if mutated, can produce a particular phenotype, which may
be correlated with both the gene length and functional structure
of the gene. Target size will also likely depend on whether the
new phenotype involves a gain or loss of molecular function.
Fig. 1. Examples of commonly observed evolutionary shifts in flower colour. (A, B) Transition from pigmented Ipomoea cordatotriloba to
white I. lacunosa. (C, D) Transition from purple Penstemon neomexicanus to red P. barbatus. (A, B) With permission from Tanya Duncan;
(C, D) Taken by Carolyn A. Wessinger.
Downloaded from http://jxb.oxfordjournals.org/ at Duke University on April 2, 2013
conducted to date, few traits are represented by more than one
mutation.
Flower colour is a trait that affords an unusual opportunity to
adopt this approach. Flower colour is evolutionarily labile, with
examples of evolutionary divergence between closely related
species from diverse genera and families (Fig. 1). This offers the
opportunity to examine ‘replicate’ cases of the same phenotypic
change, allowing a rigorous statistical assessment of patterns. In
addition, the structure of the anthocyanin biosynthetic pathway,
which produces the most important and widespread floral pigments, is conserved across angiosperms and is well-characterized genetically. Because this pathway consists of relatively few
enzyme-coding genes and associated regulatory genes (Fig. 2), a
candidate-gene approach to dissecting the genetic underpinnings
of flower colour divergence has been successful.
Although flower colour divergence may occur through neutral drift or linkage to pleiotropic traits experiencing selection,
in many cases such change is likely adaptive. Indirect evidence
for a common role of natural selection in flower colour change
is provided by the widespread convergent evolution of pollination syndromes (Fenster et al., 2004). Evolutionary shifts from
one pollination syndrome to another often involve particular
flower colour transitions. For example, shifts from bee to hummingbird pollination are typically accompanied by shifts from
blue/purple to red flowers, while shifts from either bee or bird
to hawkmoth pollination are often accompanied by shifts from
pigmented to pale or white flowers (reviewed by Rausher, 2008).
This regularity is inconsistent with genetic drift as major cause of
Lessons from flower colour evolution | 5743
Loss-of-function phenotypes may have a larger mutational target
size than gain-of-function phenotypes since there are normally
many more nucleotide sites that, if mutated, eliminate a molecular function vs. those that create a novel function.
Any deleterious pleiotropic effects on fitness will negatively
influence a mutation’s relative contribution to adaptation. Thus,
even if regulatory and functional mutations produce the same
advantageous phenotype and enjoy the same fitness benefit,
they may have unequal net fitnesses if one negatively perturbs
other pleiotropic traits specified by that gene (Otto, 2004). Since
the fixation probability for an advantageous mutation is proportional to its net fitness, mutations with minimal deleterious
pleiotropic effects will disproportionately contribute to adaptive
divergence.
For most traits, little information is available on either relative mutation rates or the relative magnitudes of deleterious
pleiotropy for regulatory vs. functional mutations. However, we
can obtain an approximate estimate of the rates for regulatory
and functional mutations causing particular flower colour shifts
because spontaneous flower colour mutants are highly visible and
are often preserved by horticulturalists. Although there is little
empirical evidence on differential pleiotropy, we can make strong
conjectures about the relative pleiotropic effects of different types
of mutations based on our understanding of the structure and
function of the anthocyanin pathway. This information permits a
preliminary evaluation of how relative mutation rates and differential pleiotropy affect the relative contribution of regulatory and
functional mutations to evolutionary changes in flower colour.
Downloaded from http://jxb.oxfordjournals.org/ at Duke University on April 2, 2013
Fig. 2. The anthocyanin pathway. The pathway has three branches. The anthocyanins produced, and hence flower colour, are
determined by which branches are functional. Anthocyanins derived from pelargonidin are red. Those derived from delphinidin tend to
be blue/purple. Those derived from cyanidin are variable, from red to magenta to blue/purple, depending on plant taxon. Flux down
the different branches is determined largely by the activity of enzymes F3’h and F3’5’h. Downregulation or functional inactivation
of these enzymes causes flux to be redirected down a different branch, resulting in a different flower colour, as indicated by flower
symbols. Production of non-pigmented flowers can be accomplished by downregulation or functional inactivation of other enzymes
or transcription factors (marked with ∅). Side branches leading to other flavonoids are in grey. ANS, Anthocyanidin synthase; bHLH,
basic-helix-loop-helix; CHI, chalcone isomerase; CHS, chalcone synthase; DFR, dihydroflavonol-4-reductase; DHK, dihydrokaempferol;
DHM, dihydromyricetin; DHQ, dihydroquercetin; F3H, flavanone 3-hydroxylase; F3’H, flavonoid 3’-hydroxylase; F3’5’H, flavonoid
3’,5’-hydroxylase; FLS, flavonol synthase; LAR, leucoanthocyanidin reductase; WDR, WD-repeat.
5744 | Wessinger and Rausher
Evolutionary loss of pigmentation
Evolutionary transitions from blue to red
flowers
A second common evolutionary transition is from blue or purple to red flowers (Fig. 1). This type of shift often accompanies
shifts in adaptation from bee to hummingbird, co-occurs with
other floral adaptations that facilitate hummingbird pollination,
and can arise repeatedly within a plant genus or family (Rausher,
2008; Thomson and Wilson, 2008). Shifts from bluer to redder
flowers generally involve a change in the biochemical class of
anthocyanin that is produced. Each type of flavonoid, including anthocyanins, can occur in three different forms that vary
in the number of hydroxyl groups attached to the B-ring of the
molecule. Anthocyanins derived from pelargonidin have one
hydroxyl group, those derived from cyanidin have two, and those
derived from delphinidin have three. In general, an increase in
hydroxyl groups shifts the hue of the molecule from redder to
bluer: pelargonidin-based anthocyanins tend to be red or orange,
cyanidin-based anthocyanins range from reddish-magenta to
Downloaded from http://jxb.oxfordjournals.org/ at Duke University on April 2, 2013
A flower colour transition common to many angiosperm taxa
is loss of floral anthocyanins, which typically produces white,
or sometimes yellow, flowers (Fig. 1). Such evolutionary shifts
often accompany a switch in pollinators from bees or hummingbirds to moths or bats (Baker, 1963; Stebbins, 1970; Grant,
1994). In theory, shifts from pigmented to white flowers could
involve any mutation that blocks one or more steps of the anthocyanin pathway (Fig. 2). These include: (1) loss-of-function
(LOF) mutations to any pathway enzyme; (2) cis-regulatory
mutations that downregulate any pathway enzyme; (3) LOF to
any of the three proteins (R2R3 MYB, bHLH, and WDR) that
regulate expression of the enzyme-coding genes; and (4) cis-regulatory mutations that downregulate any of these regulators. For
our purposes, we consider mutations in category 1 to be functional and those in categories 2–4 to be regulatory.
Characterization of spontaneous white flower mutants in
Petunia, Antirrhinum, Ipomoea, and other plant taxa have demonstrated roughly equal frequencies of both regulatory and functional mutations: 29 of 69 tabulated cases involve spontaneous
functional mutations to core pathway enzymes (Fig. 2); 8 of 69
cases involved spontaneous cis-regulatory mutations to core
pathways enzymes; and 32 of 69 cases involved spontaneous
transcription factor mutations (for details, see Streisfeld and
Rausher, 2010). By contrast, the evolutionary fixation of white
flowers within natural populations or species all involve inactivation of anthocyanin transcription factors (8–12 cases; Streisfeld
and Rausher, 2010). This pattern is strikingly and statistically
different from that seen in the spontaneous mutations, indicating that mutations affecting transcription regulatory proteins are
preferentially fixed in the evolutionary loss of pigmentation.
Moreover, in all but one case (where the identity of the regulatory gene is unknown), the mutation fixed by natural selection
inactivated the floral R2R3MYB transcription factor. By contrast, most spontaneous regulatory mutations eliminating floral
anthocyanins inactivated the bHLH and WDR regulatory proteins. Thus, despite larger mutational target sizes for bHLH and
WDR, mutations to R2R3Myb are preferentially fixed (Streisfeld
and Rausher, 2010).
This pattern can be explained by differences among these
genes in the inferred magnitude of their associated pleiotropy,
which we can deduce from properties of the anthocyanin pathway. This pathway, including enzymes and regulatory proteins,
is also responsible for producing related classes of flavonoid
compounds that perform diverse functions in vegetative plant
tissues (Winkel-Shirley, 2002). Most plants deploy a number of
R2R3MYB transcription factors to regulate anthocyanin production, each with a tissue-specific or context-dependent expression profile (Ramsay and Glover, 2005; Albert et al., 2011).
Consequently, inactivation of the R2R3MYB responsible for
regulating anthocyanin production in floral tissues eliminates
anthocyanins in that tissue only. By contrast, enzymes of the
anthocyanin pathway, as well as the bHLH and WDR regulators, are typically expressed in vegetative tissues as well as floral
tissue. LOF mutations to these proteins would not only produce white flowers, but also would block production of other
flavonoid compounds that enhance fitness. This is expected to
produce deleterious pleiotropic effects, while few such effects
are expected for the inactivation of the R2R3MYB. The apparent exclusive involvement of the floral R2R3Myb in evolutionary
shifts to white flowers is thus consistent with the expectation that
mutations incurring the least deleterious pleiotropy will be preferentially fixed by selection.
We lack extensive empirical data demonstrating the importance of this pleiotropy on plant fitness. However, our limited
data are consistent with this interpretation. In field experiments,
Rausher and Fry (2003) detected no deleterious pleiotropic fitness effects of an LOF mutation in the floral R2R3Myb gene
in Ipomoea purpurea. By contrast, an LOF mutation to Chs
in I. purpurea reduces survival and fecundity substantially
(Coberly and Rausher, 2003, 2008), as does an LOF mutation to
Ans in Phlox drummondii (Levin and Brack, 1995; R Hopkins
and MD Rausher, unpublished). These results support the contention that inactivation of the R2R3MYB protein has a much
smaller detrimental pleiotropic effect on fitness than inactivation of enzyme-coding genes. However more experiments of this
type with other pathway genes, including the other regulatory
genes, are needed before definitively concluding that differential
pleiotropy explains the highly biased involvement of the Myb
gene in evolutionary transitions to white or yellow flowers.
One caveat to this argument is that cis-regulatory mutations
that reduce the expression of an enzyme-coding gene or regulatory gene only in flowers would also be expected to incur minimal pleiotropy, yet these have not been seen in evolutionary
transitions. In fact, such cis-regulatory mutations that inactivate
only the promoter module responsible for expression in flowers
likely have a very small ‘target size’ and are rare. Of 69 spontaneous mutations reducing floral pigmentation tabulated by
Streisfeld and Rausher (2010), only eight were cis-regulatory. Of
these, most involved either transposon insertions or large deletions in the promoter region, which likely inactivate the gene in
all tissues. Although technically these are cis-regulatory mutations, they would experience the same magnitude of deleterious
pleiotropy as knockouts of enzyme-coding genes.
Lessons from flower colour evolution | 5745
transcriptional activator (Smith and Rausher, 2011; the identity
of this transcription factor is currently unknown). In contrast,
for three of four cases where adaptation involved inactivation of
F3’5’h, this has been accomplished through functional inactivation of coding sequences (Supplementary Table S1). The exception is Phlox drummondii, in which a cis-regulatory mutation
downregulates F3’5’h, although the functionality of this gene
in red-flowered individuals has not been tested (Hopkins and
Rausher, 2011).
Mutations to F3’h that contribute to evolutionary divergence
in flower colour do not reflect the spontaneous mutation rate –
despite the higher rate of functional mutations, only regulatory
mutations with effects restricted to floral tissue have been fixed
by natural selection. On the other hand, mutations to F3’5’h
involved in flower colour divergence do largely reflect the spontaneous mutation rate – most involve functional inactivation.
These patterns suggest that functional mutations to F3’h, but
not F3’5’h, have low fitness relative to regulatory mutations due
to associated pleiotropic effects. The major pleiotropic roles of
F3’h and F3’5’h are the production of di- and trihydroxylated
flavonoids in vegetative tissues, respectively. We have compiled
several lines of evidence that the production of dihydroxylated
flavonoids, but not trihydroxylated flavonoids, in vegetative tissue may be important for plant fitness. This evidence is consistent with the interpretation that LOF mutations in F3’h, but not
F3’5’h, incur substantial deleterious pleiotropy.
First, we compared the relative abundance of the different
anthocyanins found in flowers vs. vegetative tissue reported by
biochemical surveys of anthocyanin production (Table 2). Floral
anthocyanidin production is quite variable among species surveyed: 20% produce pelargonidin, 44% produce cyanidin, and
45% produce delphinidin. Despite this variability in floral tissue,
most plant species (nearly 90%) produce only the dihydroxylated cyanidin in vegetative tissue. This pattern suggests that
most species that produce pelargonidin or delphinidin in flowers
produce cyanidin in their foliage, consistent with natural selection favouring production of dihydroxylated flavonoids in vegetative tissue.
Second, differences in the taxonomic distributions of F3’h and
F3’5’h also suggest that a loss of F3’h activity is substantially
more costly than a loss of F3’5’h activity. The incorporation of
both F3’h and F3’5’h into the anthocyanin pathway occurred
Table 1. Mutations responsible for evolutionary transitions from blue/purple to red flowers
Blue species
Red species
Gene action
Reference
Ipomoea ternifolia: cy
Ipomoea purpurea: cy
Ipomoea purpurea: cy
Ipomoea coccinea: pg
Ipomoea horsefalliae: pg
Ipomoea quamoclit: pg
F3’H – regulatory
F3’H – regulatory
F3’H – regulatory
Iochroma cyaneum: de
Iochroma cyaneum: de
Penstemon neomexicanus: de
Antirrhinum kelloggii: de
Phlox drummondii (purple subsp.): de
Iochroma gesnerioides: pg
Iochroma gesnerioides: pg
Penstemon barbatus: pg
Antirrhinum majus: cy
Phlox drummondii (pink subsp.): cy
F3’H – regulatory
F3’5’H – coding
F3’5’H – coding (and regulatory)
F3’5’H – coding
F3’5’H – regulatory
Des Marais and Rausher, 2010
Des Marais and Rausher, 2010
Zufall and Rausher, 2004; Des Marais
and Rausher, 2010
Smith and Rausher, 2011
Smith and Rausher, 2011
Unpublished data
Ishiguro et al., 2011
Hopkins and Rausher, 2011
Comparisons are between anthocyanins produced in flowers of blue-flowered species and a closely related red-flowered species. Gene action
denotes whether the mutation is a regulatory mutation causing downregulation of indicated gene, or a loss-of-function (functional) mutation. cy,
cyanidin; de, delphinidin; pg, pelargonidin.
Downloaded from http://jxb.oxfordjournals.org/ at Duke University on April 2, 2013
blue/purple (depending on the presence of co-pigments), and
delphinidin-based anthocyanins tend to be blue/purple. Not
surprisingly, in both spontaneous mutations and evolutionary
substitutions, shifts from bluer to redder flowers almost always
involve a decrease in hydroxylation state, although there are
much less common types of mutations that involve elimination
of co-pigments (Scott-Moncrieff, 1936; Takahashi et al., 2006).
The production of cyanidin and other dihydroxylated flavonoids requires the activity of flavonoid 3’-hydroxylase (F3’H).
The production of delphinidin and other trihydroxylated flavonoids requires the activity of flavonoid 3’,5’-hydroxylase
(F3’5’H). Shifts from more- to less-hydroxylated anthocyanins typically involve the inactivation of one or more of these
hydroxylating enzymes, which redirect flux through the appropriate branch of the pathway. Inactivation of F3’5’h results in
a shift from delphinidin to cyanidin if F3’h is also expressed
or a shift from delphinidin to pelargonidin if not. Inactivation
of F3’h can cause a shift from cyanidin to pelargonidin in the
absence of F3’5’h expression. Inactivation of either gene could
be accomplished by: (1) a functional mutation to the enzyme;
(2) a cis-regulatory mutation that downregulates the enzyme; or
(3) a functional mutation to an anthocyanin transcription factor
that specifically affects the expression of the enzyme. Again, we
consider mutations in category 1 to be functional and those in
categories 2 and 3 to be regulatory.
Although the sample size is still small, the predominance of
regulatory vs. functional mutations involved in the evolution of
red flowers displays intriguing patterns. Spontaneous mutations
causing red flowers occur through LOF mutations in coding
sequences of both F3’h and F3’5’h, with no cases of spontaneous regulatory mutations (Supplementary Table S1, available at
JXB online). This pattern indicates that the target size for mutations that specifically inactivate these hydroxylating enzymes
is much larger for functional than for regulatory mutations.
However, the evolutionary fixation of red flowers can involve
predominantly functional or predominantly regulatory mutations, depending on whether the shift requires the inactivation
of F3’h or F3’5’h (Table 1). All examined cases where the evolution of red flowers occurred through the inactivation of F3’h
involve regulatory mutations, with effects restricted to floral tissue. This can involve either a cis-regulatory change to F3’h (e.g.
Des Marais and Rausher, 2010) or a change to an F3’h-specific
5746 | Wessinger and Rausher
Table 2. Proportions of species containing major classes of
anthocyanins in floral and vegetative tissue
Anthocyanin class
Floral tissue
(n = 860)
Vegetative tissue
(n = 376)
Derived from pelargonidin
Derived from cyanidin
Derived from delphinidin
0.199
0.442
0.448
0.016
0.891
0.160
Data compiled from Robinson and Robinson (1931, 1932, 1933,
1934) Lawrence et al. (1938) Price and Sturgess (1938) Lawrence
et al. (1939) Taylor (1940) Beale et al. (1941) Harborne (1967)
Harborne (1988), and Harborne and Williams (1998).
et al., 2005) than the monohydroxylated kaempferol. This provides a convincing explanation for why quercetin in particular
is induced by environmental stresses, and it has been suggested
that this protective function is so essential that it has generated
strong purifying selection on mutations reducing the ability to
produce quercetin throughout higher plants (Pollastri and Tattini,
2011).
Flavonols also act as developmental modulators at multiple
positions within the auxin hormone signalling pathway, enabling them to influence plant architecture, root morphology, gravitropism, and other phenotypes sensitive to auxin (reviewed by
Taylor and Grotewold, 2005). Both quercetin and kaempferol
can bind to auxin transporters and inhibit polar auxin transport
(Jacobs and Rubery, 1988; Murphy et al., 2000; Brown et al.,
2001). Data provided by Jacobs and Rubery (1988) suggest that
quercetin has a stronger affinity for auxin transporter binding
sites compared to kaempferol, a result that has been frequently
cited in subsequent papers, although Jacobs and Rubery did not
report a statistical comparison. If there were a statistically significant difference, it would suggest that quercetin is a more
important inhibitor of polar auxin transport than kaempferol. In
support of this conclusion, Arabidopsis thaliana mutants that
lack F3’h activity (and therefore can produce kaempferol, but not
quercetin) have similar auxin transport phenotypes as mutants
that completely lack flavonols (Lewis et al., 2011). This result
suggests that normal auxin transport depends upon the ability
to produce quercetin. Furthermore, quercetin and kaempferol
have different effects on auxin degradation. Quercetin inhibits
peroxidase-mediated degradation of endogenous auxin (IAA),
while kaempferol acts as a cofactor in this pathway (Furuya
et al., 1962). Interestingly, UV radiation induces the expression of the peroxidases that degrade IAA (Jansen et al., 2001)
in addition to increasing the ratio of quercetin to kaempferol,
as discussed above. UV radiation has known impacts on auxin
metabolism, although it is unclear whether this occurs as a direct response to UV radiation or an indirect response mediated
through UV-responsive flavonols (reviewed by Jansen, 2002).
Combined with the observation that auxin itself stimulates the
flavonoid pathway genes, including F3’h (Lewis et al., 2011),
it seems possible that there is a complicated regulatory pathway
that integrates signals from UV radiation and auxin metabolism
which possibly relies upon the presence of quercetin.
Finally, it is worth noting that a variety of flavonoid compounds also have important roles as antimicrobial and antiherbivory compounds (reviewed by Harborne and Williams, 2000).
Proanthocyanidins are particularly important antiherbivory
compounds (reviewed by Dixon et al., 2004). The majority of
proanthocyanidins found in nature are dihydroxylated (procyanidin), while propelargonidin and prodelphinidin are rare (Tanner,
2004), suggesting procyanidins are more effective, although we
have not found direct documentation of this.
These various lines of evidence are all consistent with the
hypothesis that functional inactivation of F3’h is very costly,
whereas inactivation of F3’5’h is not. If this hypothesis is correct, it would explain why shifts to red flowers are associated
with downregulation of F3’h but functional inactivation of
F3’5’h. Downregulation of F3’h only in floral tissue avoids the
expected deleterious pleiotropic effects that would be associated
Downloaded from http://jxb.oxfordjournals.org/ at Duke University on April 2, 2013
prior to the beginning radiation of angiosperms (Rausher, 2006;
Seitz et al., 2006), implying that absence of either gene from any
flowering plant is due to gene loss. We are unaware of any plants
that lack a functional copy of F3’h; if they exist, they are almost
certainly rare. F3’5’h, however, appears to have been completely lost in a number of taxa including Rosaceae, most of the
Asteraceae, Antirrhinum majus, Arabidopsis thaliana, Ipomoea,
Iochroma gesnerioides, Matthiola, and Tulipa. Note that this
list represents a large proportion of taxa where the presence of
F3’5’h has been explicitly investigated and likely is the tip of
the iceberg for taxa lacking F3’5’h. If one maps these absences
onto the most recent consensus tree for angiosperms (Stevens,
2001, version 9) along with taxa that are known to have a functional F3’5’h because they produce trihydroxylated flavonoids,
it is clear that there have been numerous independent losses
(Supplementary Fig. S1). This difference between F3’5’h and
F3’h in rate at which LOF mutations accumulate is most easily
interpreted as reflecting more stringent constraint, due to greater
deleterious pleiotropy, of LOF mutations in F3’h.
Finally, there is substantial evidence that dihydroxylated flavonoids perform various physiological functions better than the
corresponding mono- or trihydroxylated flavonoids. Flavonols
are among the most abundant group of flavonoids (Harborne,
1967) and carry out a variety of functions in plants (reviewed
by Winkel-Shirley, 2002; Taylor and Grotewold, 2005; Pollastri
and Tattini, 2011). One important role is protection against the
harmful effects of environmental stress. Under certain stressful conditions, plants accumulate flavonols, in particular the
dihydroxylated flavonol quercetin, thus dramatically increasing the total flavonol content and also the ratio of quercetin to
the monohydroxylated flavonol kaempferol in tissues exposed
to the stress (reviewed by Pollastri and Tattini, 2011). This has
most commonly been observed in response to high light and
UVB radiation (e.g. Olsson et al., 1998; Ryan et al., 1998, 2001,
2002; Gerhardt et al., 2008; Agati et al., 2009, 2011), but has
also been documented in response to salinity stress (Agati et al.,
2011), drought stress (Tattini et al., 2004), and nitrogen depletion (reviewed by Lillo et al., 2008). This physiological response
appears to be an adaptive mechanism to reduce the harmful
effects of reactive oxygen species produced by these environmental stresses. Flavonols have important antioxidant properties
conferred by their ability to chelate transition metal ions (RiceEvans et al., 1996). Importantly, the dihydroxylated quercetin
is significantly better at chelating both Cu ions (Brown et al.,
1998; Mira et al., 2002) and Fe ions (Mira et al., 2002; Melidou
Lessons from flower colour evolution | 5747
with functional inactivation of F3’h. Functional inactivation
of F3’5’h, on the other hand, would have no effect on the production of dihydroxylated flavonoids in vegetative tissues and
would not be expected to experience greater pleiotropic costs
than mutants that downregulate F3’5’h. Since functional mutations arise at a higher rate, they would predominantly underlie
most evolutionary transitions from delphinidin- to cyanidinbased floral anthocyanins. While the observed pattern is consistent with this explanation, more direct evidence on the magnitude
of deleterious pleiotropy associated with mutations to these two
genes is needed.
Lessons from the genetics of flower colour
evolution
Supplementary material
Supplementary data are available at JXB online.
Supplementary Table S1. Cases identified from the literature
in which a spontaneous mutation to flower colour causes a shift
from more- to less-hydroxylated anthocyanins.
Supplementary Fig. S1. Angiosperm phylogenetic tree showing loss or functional inactivation of F3’5’h.
Acknowledgements
This work was supported by the National Science Foundation
(grant number DEB0841521).
References
Agati G, Biricolti S, Guidi L, Ferrini F, Fini A, Tattini M. 2011. The
biosynthesis of flavonoids is enhanced similarly by UV radiation and
root zone salinity in L. vulgare leaves. Journal of Plant Physiology 168,
204–212.
Agati G, Stefano G, Biricolti S, Tattini M. 2009. Mesophyll
distribution of ‘antioxidant’ flavonoid glycosides in Ligustrum vulgare
leaves under contrasting sunlight irradiance. Annals of Botany 104,
853–861.
Albert NW, Lewis DH, Zhang H, Schwinn KE, Jameson PE,
Davies KM. 2011. Members of an R2R3-MYB transcription factor
family in Petunia are developmentally and environmentally regulated
to control complex floral and vegetative pigmentation patterning. The
Plant Journal 65, 771–784.
Downloaded from http://jxb.oxfordjournals.org/ at Duke University on April 2, 2013
Several lessons flow from patterns revealed by data on the genetic basis of flower colour adaptation. The first is that broad
generalizations, such as ‘morphological traits tend to involve
regulatory change’ or ‘physiological traits tend to involve functional coding-sequence change’ fail to predict differences among
very similar traits. Regardless of whether flower colour is considered a morphological or a physiological trait, adaptation can
involve primarily regulatory or primarily functional genetic
changes, depending on the type of flower colour change and the
pathway position of the genes involved. Transitions to white or
yellow flowers involve regulatory changes mediated by a specific transcription factor (R2R3MYB). Transitions from blue to
red flowers primarily involve functional mutations if the ancestral blue flowers are coloured by delphinidin, but involve primarily cis-regulatory mutations if the ancestral blue flowers are
coloured by cyanidin. These differences indicate that flower colour shifts, even phenotypically identical shifts, may predictably
involve different types of genetic mutations.
The second lesson is that differences in pleiotropy often determine whether regulatory vs. functional mutations are preferentially
involved in adaptive evolution. Minimal deleterious pleiotropy
appears to explain why virtually all cases of transition to unpigmented flowers involve inactivation of the R2R3MYB anthocyanin
transcription factor. It also explains why shifts to red flowers preferentially involve downregulation of F3’h. In the case of F3’5’h,
there is likely little difference among regulatory and functional
mutations in pleiotropy, and the prevalence of functional mutations
appears to be determined largely by mutational target size.
A third lesson is that determining whether we can or cannot
predict the genetic basis of an adaptive trait is facilitated by
examining the genetics of its repeated evolutionary origins (see
also Kopp, 2009). If enough cases are investigated, this approach
allows a powerful statistical assessment of whether different
types of mutation are preferentially involved in different types of
evolutionary transitions and in different traits. In the case of floral
pigment loss, a sufficient number of cases have been examined
to produce a statistically convincing case that R2R3Myb mutations are preferentially fixed. The case for the patterns associated
with blue/purple to red transitions is less secure because it rests
on fewer cases. However, additional cases are being examined,
and it should not be too long before the statistical confidence for
this transition is as high as for loss of pigments.
A final lesson is that evolutionary novelty can be produced
by loss of function at the molecular level. While it is unclear
whether floral pigmentation is an ancestral trait in angiosperms,
it is clearly an ancestral trait of many angiosperm families that
exhibit evolutionary changes in flower colour, as is blue/purple floral pigmentation. Evolutionary transitions to white or
red flowers thus constitute novel phenotypes. Nevertheless, the
mutations underlying these transitions are overwhelmingly lossof-function mutations that either inactivate protein function or
greatly reduce protein expression in floral tissues. We know of
only one gain-of-function mutation responsible evolutionary
change in flower colour: a cis-regulatory mutation in the floral
anthocyanin R2R3Myb gene in Phlox drummondii upregulates
that gene’s expression level and causes an increase in anthocyanin pigment production (Hopkins and Rausher, 2011). It is perhaps not surprising that the magnitude of deleterious pleiotropy
will play a large role in determining which LOF mutations are
preferentially involved in the evolution of flower colour. In principle, we would expect the magnitude of deleterious pleiotropy to
also influence which gain-of-function mutations contribute disproportionately to evolution of plant traits, since the magnitude
of pleiotropy will affect the net selection coefficients of those
mutations. Only future investigations can determine whether this
is the case. We suggest that a profitable approach for such investigations is the genetic dissection of repeated evolutionary transitions to the same novel phenotype.
5748 | Wessinger and Rausher
Baker HG. 1963. Evolutionary mechanisms in pollination biology.
Science 139, 877–883.
Beale GH, Price JR, Sturgess VC. 1941. A survey of anthocyanins.
VII. The natural selection of flower colour. Proceedings of the Royal
Society of London Series B 130, 113–126.
Brown JE, Khodr H, Hider RC, Rice-Evans CA. 1998. Structural
dependence of flavonoid interactions with Cu(II) ions: implications for
their antioxidant properties. The Biochemical Journal 359, 1105–1115.
Brown DE, Rashotte AM, Murphy AS, Normanly J, Tague BW,
Peer WA, Taiz L, Muday GK. 2001. Flavonoids act as negative
regulators of auxin transport in vivo in Arabidopsis. Plant Physiology
126, 524–535.
Carroll SB. 2008. Evo-devo and an expanding evolutionary synthesis:
a genetic theory of morphological evolution. Cell 134, 25–36.
Coberly LC, Rausher MD. 2003. Analysis of a chalcone synthase
mutant in Ipomoea purpurea reveals a novel function of flavonoids:
amelioration of heat stress. Molecular Ecology 12, 1113–1124.
Des Marais DL, Rausher MD. 2010. Parallel evolution at multiple
levels in the origin of hummingbird pollinated flowers in Ipomoea.
Evolution 64, 2044–2054.
Dixon RA, Xie DY, Sharma SB. 2004. Proanthocyanidins – a final
frontier in flavonoid research? The New Phytologist 165, 9–28.
Fenster CB, Armbruster WS, Wilson P, Dudash MR, Thomson
JD. 2004. Pollination syndromes and floral specialization. Annual
Review of Ecology, Evolution, and Systematics 35, 375–403.
Furuya M, Galston AW, Stowe BB. 1962. Isolation from peas of
co-factors and inhibitors of indolyl-3-acetic acid oxidase. Nature 193,
456–457.
Gerhardt KE, Lampi MA, Greenberg BM. 2008. The effects of
far-red light on plant growth and flavonoid accumulation in Brassica
napus in the presence of ultraviolet B radiation. Photochemistry and
Photobiology 84, 1445–1454.
Grant V. 1994. Modes and origins of mechanical and ethological
isolation in angiosperms. Proceedings of the National Academy of
Sciences, USA 91, 3–10.
Harborne JB. 1967. Comparative biochemistry of the flavonoids .
London: Academic Press.
Harborne JB. 1988. The flavonoids: advances in research since
1980 . London: Chapman and Hall.
Harborne JB, Williams CA, 1998. Anthocyanins and other
flavonoids. Natural Product Reports 15, 631–652.
Harborne JB, Williams CA. 2000. Advances in flavonoid research
since 1992. Phytochemistry 55, 481–504.
Hoekstra HE, Coyne JA. 2007. The locus of evolution: evo devo and
the genetics of adaptation. Evolution 61, 995–1016.
Ishiguro K, Taniguchi M, Tanaka Y. 2011. Functional analysis
of Antirrhinum kelloggii flavonoid 3’-hydroxylase and flavonoid
3’,5’-hydroxylase genes; critical role in flower color and evolution in
the genus Antirrhinum. Journal of Plant Research 125, 451–456.
Jacobs M, Rubery PH. 1988. Naturally occurring auxin transport
regulators. Science 241, 346–349.
Jansen MAK. 2002. Ultraviolet-B radiation effects on plants:
induction of morphogenic responses. Physiologia Plantarum 116,
423–429.
Jansen MAK, van der Noort RA, Tan MYA, Prinsen E, Lagrimini
LM, Thorneley RNF. 2001. Phenol-oxidizing peroxidases contribute
to the protection of plants from ultraviolet radiation stress. Plant
Physiology 126, 1012–1023.
Jones KN, Reithel JS. 2000. Pollinator-mediated selection on a
flower color polymorphism in experimental populations of Antirrhinum
(Scrophulariaceae). American Journal of Botany 88, 447–454.
Kamsteeg J, van Brederode J, van Nigtevecht G. 1979. Genetics
of anthocyanin formation in petals of the red campion (Silene dioica (L)
Clairv.). Genetica 51, 5–13.
Kamsteeg J, van Brederode J, van Nigtevecht G. 1980. Genetical
and biochemical evidence that the hydroxylation pattern of the
anthocyanin B-ring in Silene dioica is determined at the p-coumaroylcoenzyme a stage. Phytochemistry 19, 1459–1462.
Kopp A. 2009. Metamodels and phylogenetic replication: a
systematic approach to the evolution of developmental pathways.
Evolution 63, 2771–2789.
Lawrence WJC, Price JR, Robinson GM, Robinson R. 1938. A
survey of anthocyanins. V. The Biochemical Journal 32, 1661–1667.
Lawrence WJC, Price JR, Robinson GM, Robinson R.1939. The
distribution of anthocyanins in flowers, fruits and leaves. Philosophical
Transactions of the Royal Society of London Series B 230, 149–178.
Levin DA, Brack ET. 1995. Natural selection against white petals in
Phlox. Evolution 49, 1017–1022.
Lewis DR, Ramirez MV, Miller ND, Vallabhaneni P, Ray WK,
Helm RF, Winkel BSJ, Muday GK. 2011. Auxin and ethylene induce
flavonol accumulation through distinct transcriptional networks. Plant
Physiology 156, 144–164.
Lillo C, Lea US, Ruoff P. 2008. Nutrient depletion as a key factor
for manipulating gene expression and product formation in different
branches of the flavonoid pathway. Plant, Cell and Environment 31,
587–601.
Melidou M, Riganakos K, Galaris D. 2005. Protection against DNA
damage offered by flavonoids in cells to hydrogen peroxide: the role of
iron chelation. Free Radical Biology and Medicine 39, 1591–1600.
Hopkins R, Rausher MD. 2011. Identification of two genes causing
reinforcement in the Texas wildflower Phlox drummondii. Nature 469,
411–414.
Mira L, Fernandez MT, Santos M, Rocha R, Florencio MH,
Jenning KR. 2002. Interactions of flavonoids with iron and copper
ions: a mechanism for their antioxidant activity. Free Radical Research
36, 1199–1208.
Hopkins R, Rausher MD. 2012. Pollinator-mediated selection on
flower color allele drives reinforcement. Science 335, 1090–1092.
Murphy A, Peer WA, Taiz L. 2000. Regulation of auxin transport by
aminopeptidases and endogenous flavonoids. Planta 211, 315–324.
Downloaded from http://jxb.oxfordjournals.org/ at Duke University on April 2, 2013
Coberly LC, Rausher MD. 2008. Pleiotropic effects of an allele
producing white flowers in Ipomoea purpurea. Evolution 62,
1076–1085.
Irwin RE, Strauss SY. 2005. Flower color microevolution in wild
radish: evolutionary response to pollinator-mediated selection. The
American Naturalist 165, 225–237.
Lessons from flower colour evolution | 5749
Olsson LC, Veit M, Weissenbock G, Bornman JF. 1998.
Differential flavonoid response to enhanced UV-B radiation in Brassica
napus. Phytochemistry 49, 1021–1028.
flavonoid 3′-hydroxylase and flavonoid 3′,5′-hydroxylase cDNAs
reveals independent evolution of flavonoid 3′,5′-hydroxylase in the
Asteraceae family. Plant Molecular Biology 61, 365–381.
Otto SP. 2004. Two steps forward, one step back: the pleiotropic
effects of favoured alleles. Proceedings of the Royal Society Series B
271, 705–714.
Smith SD, Rausher MD. 2011. Gene loss and parallel evolution
contribute to species difference in flower color. Molecular Biology and
Evolution 28, 2799–2810.
Pollastri S, Tattini M. 2011. Flavonols: old compounds for old roles.
Annals of Botany 108, 1225–1233.
Stebbins GL. 1970. Adaptive radiation of reproductive characteristics
in angiosperms. I: pollination mechanisms. Annual Review of Ecology,
Evolution, and Systematics 1, 307–326.
Price JR, Sturgess VC. 1938. A survey of anthocyanins. VI. The
Biochemical Journal 32, 1658–1660.
Ramsay NA, Glover BJ. 2005. MYB-bHLH-WD40 protein complex
and the evolution of cellular diversity. Trends in Plant Science 10,
1360–1385.
Rausher MD. 2006. The evolution of flavonoids and their genes.
In: E Grotewold, ed, The science of flavonoids . New York: Springer,
pp 175–211.
Stern DL. 2000. Evolutionary developmental biology and the problem
of variation. Evolution 54, 1079–1091.
Stern DL, Orgogozo V. 2008. The loci of evolution: how predictable
is genetic evolution? Evolution 62, 2155–2177.
Stevens PF. 2001. Angiosperm phylogeny website. Available at:
www.mobot.org/mobot/research/apweb.
Streisfeld MA, Kohn JR. 2007. Environmental and pollinatormediated selection on parapatric floral races in Mimulus aurantiacus.
Journal of Evolutionary Biology 20, 122–132.
Rausher MD, Fry JD. 1993. Effects of a locus affecting floral
pigmentation in Ipomoea purpurea on female fitness components.
Genetics 134, 1237–1247.
Streisfeld MA, Rausher MD. 2010. Population genetics, pleiotropy,
and the preferential fixation of mutations during adaptive evolution.
Evolution 65, 629–642.
Rice-Evans CA, Miller N, Papanga G. 1996. Structure-antioxidant
relationships of flavonoids and phenolic acids. Free Radical Biology
and Medicine 20, 933–956.
Takahashi R, Githiri SM, Hatayama K, Dubouzet EG, Shimada
N, Aoki T, Ayabe S-I, Iwashina T, Toda K, Matsumura H.
2006. A single-base deletion in soybean flavonol synthase gene is
associated with magenta flower color. Plant Molecular Biology 63,
125–135.
Robinson GM, Robinson R. 1931. A survey of anthocyanins. I.
The Biochemical Journal 25, 1687–1705.
Robinson GM, Robinson R. 1932. A survey of anthocyanins. II. The
Biochemical Journal 32, 1647–1664.
Robinson GM, Robinson R. 1933. A survey of anthocyanins. III.
Notes on the distribution of leuco-anthocyanins. The Biochemical
Journal 27, 206–212.
Robinson GM, Robinson R. 1934. A survey of anthocyanins. IV. The
Biochemical Journal 28, 1712–1720.
Ryan KG, Markham KR, Bloor SJ, Bradley JM, Mitchell
KA, Jordan BR. 1998. UVB radiation induced increase in
quercetin:kaempferol ratio in wild-type and trangenic lines of Petunia.
Photochemistry and Photobiology 68, 323–330.
Tanner GJ. 2004. Condensed tannins. In: KM Davies, ed, Plant
pigments and their manipulation. Annual Plant Reviews , vol 12. Boca
Raton, FL: Blackwell Publishing–CRC Press, pp 150–184.
Tattini M, Galardi C, Pinelli P, Massai R, Remorini D, Agati G.
2004. Differential accumulation of flavonoids and hydroxycinnamates
in leaves of Ligustrum vulgare under excess light and drought stress.
The New Phytologist 163, 547–561.
Taylor LP, Grotewold E. 2005. Flavonoids as developmental
regulators. Current Opinion in Plant Biology 8, 317–323.
Taylor TWJ. 1940. Plant pigments in the Galapagos Islands.
Proceedings of the Royal Society of London Series B 129, 230–237.
Ryan KG, Swinney EE, Winefield C, Markham KR. 2001.
Flavonoids and UV photoprotection in Arabidopsis mutants. Zeitschrift
fur Naturforschung C 56, 745–754.
Thomson JD, Wilson P. 2008. Explaining evolutionary shifts between
bee and hummingbird pollination: convergence, divergence, and
directionality. International Journal of Plant Sciences 169, 23–38.
Ryan KG, Swinny EE, Markham KR, Winefield C. 2002. Flavonoid
gene expression and UV photoprotection in transgenic and mutant
Petunia leaves. Phytochemistry 59, 23–32.
Wagner GP, Lynch VJ. 2008. The gene regulatory logic of
transcription factor evolution. Trends in Ecology and Evolution 23,
377–385.
Schemske DW, Bierzychudek P. 2001. Perspective: evolution of
flower color in the desert annual Linanthus parryae: Wright revisited.
Evolution 55, 1269–1282.
Waser NM, Price MV. 1981. Pollinator choice and stabilizing
selection for flower color in Delphinium nelsonii. Evolution 35,
376–390.
Schemske DW, Bierzychudek P. 2007. Spatial differentiation for
flower color in the desert annual Linanthus parryae: was Wright right?
Evolution 61, 2528–2543.
Winkel-Shirley B. 2002. Biosynthesis of flavonoids and effects of
stress. Current Opinion in Plant Biology 5, 218–223.
Scott-Moncrieff R. 1936. A biochemical survey of some Mendelian
factors for flower colour. Journal of Genetics 32, 117–170.
Seitz C, Eder C, Deiml B, Kellner S, Martens S, Forkmann G.
2006. Cloning, functional identification and sequence analysis of
Wray GA. 2007. The evolutionary significance of cis-regulatory
mutations. Nature Reviews Genetics 8, 206–216.
Zufall RA, Rausher MD. 2004. Genetic changes associated with
floral adaptation restrict future evolutionary potential. Nature 428,
847–850.
Downloaded from http://jxb.oxfordjournals.org/ at Duke University on April 2, 2013
Rausher MD. 2008. Evolutionary transitions in floral color.
International Journal of Plant Sciences 169, 7–21.