Download Using Artificial Selection to Understand Plastic Plant Phenotypes1

INTEGR. COMP. BIOL., 45:475–485 (2005)
Using Artificial Selection to Understand Plastic Plant Phenotypes1
HILARY S. CALLAHAN2
Barnard College, Columbia University, Department of Biological Sciences, 3009 Broadway, New York, New York 10027
INTRODUCTION
Plants are sessile, autotrophic, and grow in an indeterminate and modular manner. Because of this,
physiological and ecological research has often focused on responses to shifts in resource availability
(e.g., availability of light) or to changes in strongly
correlated non-resource factors (e.g., photoperiod, R:
FR ratio). Such factors can elicit physiological or morphological responses within individual plants: phenotypic plasticity. Such factors can also exert selective
pressures, shaping evolved responses: shifts in the distribution of phenotypic variation within populations
(i.e., changes in mean, variance, covariance, etc.)
(Lande and Arnold, 1983; Rausher, 1992). The plasticity of any given trait, which has a genetic basis and
which may or may not be adaptive, can intensify or
attenuate evolved responses, and can itself evolve in
response to selection depending on the scale of spatial
or temporal heterogeneity (Via and Lande, 1985; Via
et al., 1995; Pigliucci, 1996, 2001). Since selection can
act on both variation for traits and variation for the
plasticity of traits, the evolution of plastic traits can
only be understood by addressing the complex interplay between the plastic responses of individuals and
the evolved responses of populations.
The tools of ecological and quantitative genetics are
therefore crucial for studying the function and evolution of plastic plant traits (Ackerly et al., 2000; Ackerly and Monson, 2003), and here I will address one
of the most powerful: assessment of direct and indirect
responses to artificial selection. While selection studies
have the potential to provide important insights, their
design, execution, and interpretation can be extremely
challenging. Beyond the complexities sketched
above—well-researched in plants but certainly not
unique to them—another challenge emerges from a
similarity between plants and other organisms: they do
not respond to just a single environmental variable at
a time. Any given environmental factor can elicit a
plastic response, exert selection, or both. This can create constraints—short- or long-term, antagonistic or
synergistic—that can affect whether and how individuals respond via plasticity to other environmental factors, as well as the future evolutionary trajectory of
the population (Bell and Lechowicz, 1994; Van Tienderen and Van Hinsberg, 1996).
Understanding such constraints requires sophisticated examination of quantitative genetic architecture, including (1) how a single trait may relate to one or more
other traits within a single environment (i.e., genetic
correlations, pleiotropy, or linkage: Fuller et al., 2005),
(2) how a single trait is expressed in multiple environments (i.e., plasticity: Scheiner, 2002; Brakefield,
2003) and sometimes (3) a merger of the first two
considerations: relationships among multiple traits expressed in multiple environments. This third issue is
likely to be important in wild populations.
Artificial selection experiments addressing the complex evolution of phenotypically plastic traits require
subjecting large experimental populations (consisting
of many different plant genotypes) to variation in multiple environmental factors, often in novel factorial
combinations. Such studies can simultaneously quantify genetic variance for traits and genetic variance for
plasticity, allowing insights into whether and how
traits and their associated plasticities evolve, potentially but not necessarily in tandem. By concentrating
on artificial selection studies that investigate phenotypically plastic traits in plants, this review aims to
1 From the Symposium Selection Experiments as a Tool in Evolutionary and Comparative Physiology: Insights into Complex Traits
presented at the Annual Meeting of the Society for Integrative and
Comparative Biology, 5–9 January 2004, at New Orleans, Louisiana.
2 E-mail: [email protected]
475
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
SYNOPSIS.
The plasticity of any given trait, which has a genetic basis and which may or may not be
adaptive, can intensify or attenuate evolved responses, and can itself evolve in response to selection depending
on the scale of spatial or temporal heterogeneity. To investigate the complex function and evolution of plastic
traits, an appealing yet challenging approach is assessing responses to artificial selection. Here, I review how
artificial selection has been employed to explore four botanical research themes: (1) relationships between
plastic and evolved responses to multiple stresses, (2) integration of cellular, leaf-level, and whole-plant
responses to altered CO2 concentrations, (3) photomorphogenic and photoperiodic development, both mediated by phytochrome photoreceptors, and (4) the evolution of the pest-induced myrosinase-glucosinolate
system in cruciferous plants. These diverse topics are unified not only because they have been studied using
artificial selection experiments, but also because they have considered variability in multiple traits affected
by multiple factors in the external environment. Limitations of such research include a dearth of long-term
studies; a surprising but often logistically necessary omission of control or replicate lines; and numerous
issues relating to assessing impacts of inbreeding and drift. In addition to discussing options for circumventing such limitations, I draw attention to strategies for integrating the results of artificial selection studies
with progress in functional and evolutionary genomics.
476
HILARY S. CALLAHAN
PREDICTABLE ADAPTATIONS TO MULTIPLE STRESSES?
Stanton et al. (2000) have used artificial selection
to investigate how both evolved responses and phenotypic plasticity are influenced by stress. Their work
confronts Grime’s (1977, 1979) controversial C-S-R
system, which classifies plant life histories based on
assumed trade-offs among three basic selective pressures: competitive ability (C selection), stress tolerance
(S selection) or dispersal and colonizing ability (R se-
lection, referring to ruderals, species associated with
highly disturbed habitats). Since Grime defines stress
very generally as ‘‘any external constraint which limits
the rate of dry-matter production of all or parts of the
vegetation,’’ different forms (e.g., aridity, salinity, nutrient deficiency) are predicted to favor a common
suite of adaptations to stress. Such adaptations include
slow growth, delayed reproduction, and adaptive homeostastis, and are at odds with competitive strategies
such as faster growth or adaptive plasticity or the ruderal habit of reproducing very early.
The ‘‘artificial evolution’’ strategy employed by
Stanton and her colleagues (2000, 2004) is a departure
from past comparative studies addressing this issue.
Rather than making comparisons among species (e.g.,
Westoby et al., 2002), it examines adaptive differentiation within a species, specifically among selection
lines shaped by different stress histories. This allows
examining genetic correlations for fitness across environments, looking for positive correlations across
different types of stressful environments (Stanton et
al., 2000) or negative correlations across contrasting
environments believed to select for ‘‘C’’ as compared
with ‘‘S’’ life-history strategies (Stanton et al., 2004).
Their initial selection study, conducted in a greenhouse, used eight replicate sets of seeds, all sampled
originally from a single natural population of the wild
mustard Sinapis arvensis, a weedy outcrossing annual.
Each replicate consisted of six sets of 48 seeds, with
each set assigned to one of six environments: a ‘‘no
stress’’ control, high salt, high boron, low light, low
water, and low nutrients. In each of three successive
generations, 48 progeny (seeds) were randomly sampled from each stress-by-replicate combination. In this
short-term experiment, individual fecundity was quantified and used as the target for selection (‘‘quasi-natural’’ selection). Fecundity was presumably influenced
by combinations of genotype and environment. One
limitation of this greenhouse study is that plants were
grown in individual pots at low densities, minimizing
belowground plant-plant interactions that would occur
in nature. Also, the only correlated physiological traits
quantified were integrative performance (i.e., size) estimates. Despite these drawbacks, this ambitious study
has four notable strengths. First, fecundity data were
used to estimate relative fitness and to conduct formal
phenotypic selection analyses. Second, the study examined whether stress affects the opportunity for selection (i.e., variation in relative fitness). Third, samples from replicate lines were exposed to multiple environments in a follow-up study, allowing assessment
of not only evolved responses to selection but also
phenotypic plasticity and its adaptive significance.
Fourth, the study used a self-incompatible species, so
that all pollinations (performed within a line) resulted
in outcrossing, minimizing inbreeding.
Stanton and colleagues (2000) found consistent directional selection for stress-avoiding rather than
stress-tolerating strategies, with phenotypic selection
for earlier flowering occurring in all five stress envi-
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
complement recent treatments by Conner (2003), Pigliucci (2003), Brakefield (2003), Fry (2003) and Scheiner (2002).
I discuss how useful this approach has been not only
in disentangling complexity, but also in unifying at
least four rather disparate topics in plant eco- and evolutionary physiology. The first topic, how populations
respond to diverse stresses, has been of enduring interest for plant ecologists and is increasingly important
in the face of anthropogenic global change and habitat
degradation (Bradshaw and McNeilly, 1991; Chapin et
al., 1993). The second topic, the potential response of
plants to changes in CO2 availability, is a pressing concern given past plant responses to historical fluctuations and the dramatic changes anticipated during the
21st century (Bradshaw and McNeilly, 1991; Geber
and Dawson, 1993; Sage and Coleman, 2001). The
third topic, the evolution of plant photomorphogenic
and photoperiodic responses, illustrates some of the
challenges of translating progress in functional genomics into an evolutionary ecology framework (Callahan et al., 1997; Schlichting and Smith, 2002). A
fourth topic, chemical defenses induced by herbivores
and pathogens, illustrates how results that are puzzling,
and even in conflict with theory, may become understandable via a more sophisticated view of interspecific
ecological interactions (Mitchell-Olds and Bergelson,
2000; Agrawal, 2001, 2003).
In selection studies of plastic traits, multiple environments often come into play, either during selection,
during evaluation of responses to selection, or both. In
addition to making such studies especially cost- and
labor-intensive, this reality may strongly influence the
choice of a model organism. Another important decision is whether to use ‘‘true’’ artificial selection for a
relevant non-fitness trait, to quantify a proxy of fitness
and impose ‘‘quasi-natural’’ selection, or to instead target viability (Scheiner, 2002; Fry, 2003). Such decisions will affect four other important choices that are
not unique to botanical studies: whether selection is
short- or longer-term, whether and how to replicate
lines, whether to include unselected control lines, and
how to manipulate or account for changes in mating
system, level of inbreeding, inbreeding depression, or
genetic drift. My review of several artificial selection
studies will point out choices made by plant researchers. A concluding section will discuss how these
choices can limit the interpretation of studies, yet
sometimes lead to opportunities for complementary research.
SELECTION
ON
PLASTIC PLANT PHENOTYPES
matter: perhaps it experienced frequent disturbance, biasing all of the selection lines in the study toward rapid
evolution of ruderal strategies. Second, only a shortlived annual was studied. A comparable study with a
perennial could examine whether longer-lived species
evolve ‘‘S’’ rather than ‘‘R’’ strategies in the face of
diverse stresses. The feasibility of such a study is questionable, however, given the arduous nature of both
natural and artificial selection studies with annuals
(Geber and Griffen, 2003).
CO2 LEVELS AND THE EVOLUTION OF
PLANT PRODUCTIVITY
Because net carbon assimilation is an absolute requirement for an increase in plant biomass, artificial
selection targeting increased photosynthetic capacity is
an attractive tool for improving crop productivity.
Some reviewers have questioned the validity of this
improvement technique, because scaling from leaf-level CO2 assimilation to whole-plant yield is too complex and whole plant photosynthetic capacity cannot
be properly estimated using instantaneous leaf-level
measurements. It has also been suggested that genetic
variation may be limited for performance traits so critical to plant fitness, particularly in modern crops (reviewed in Medrano et al., 1995). To confront these
criticisms, a novel study of tobacco, Nicotiana tabacum, combined mutagenesis with artificial selection.
Selection targeted variation in CO2 compensation
point, G, a measure of the balance between carbon
gained through photosynthesis and lost through photoand maintenance respiration. Rather than estimating
instantaneous, leaf-level G and using estimates to rank
and select genotypes, the study imposed viability selection against plants with high compensation points
by reducing CO2 concentrations to 60–70 ppm (as
compared to ambient levels of ;350–360 ppm). Only
;5% of individuals survived to reproductive maturity,
after which selected genotypes and unselected control
genotypes were clonally propagated for a series of
comparative studies.
Recognizing that G is a complex and phenotypically
plastic trait, comparisons were made for G and for
many different biochemical, physiological and anatomical traits (Medrano and Primomillo, 1985; Medrano et al., 1995), taking into account how traits vary
diurnally or during leaf ontogeny (Delgado et al.,
1994), and sometimes in glasshouse vs. field conditions. Despite this sophistication, results are difficult
to interpret because they did not characterize trait plasticities to ambient vs. novel CO2 concentrations. For
example, the study failed to detect differences between
selected and control genotypes in photosynthetic or
photorespiratory machinery (e.g., increases in Rubisco
enzyme content or activity: Delgado et al., 1993; Medrano et al., 1995). Also, selected genotypes did not
show a difference in instantaneous G, despite experiencing strong selection on life-time G. In part, this latter result validates criticisms of instantaneous methods
for estimating G. More generally, it calls into question
VARIATION
IN
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
ronments but not in the control environment. Directional phenotypic selection for faster growth in height
occurred in three of five stressful environments. Variation in relative fitness tended to increase in stressful
as compared to control conditions. In other words,
stress increased the opportunity for selection.
In the face of selection, however, there was a mismatch between evolved and plastic responses. Evolutionary responses toward earlier flowering and greater
height matched selection gradients, but plastic responses were in the opposite direction (e.g., flowering delayed, seedling height reduced). In other words, populations evolved trait means that enhanced their fitness
in stressful habitats, but their plastic responses to
stressful vs. non-stressful conditions were not consistent with adaptive phenotypic plasticity (e.g., Schmitt
et al., 1995; Donohue et al., 2000). Other researchers
have pointed out that specialist taxa may exhibit maladaptive plasticity (e.g., Taylor and Aarssen, 1988;
Lortie and Aarssen, 1996).
Additional evidence for maladaptive plasticity
emerged when ‘‘low light stress’’ selection line were
compared to control lines in a factorial field experiment involving high and low density plantings combined with either full sun or partial shading (Stanton
et al., 2004). In stressful full sun treatments, drought
and heat negatively impacted growth and fecundity;
less stressful ‘‘partial shade’’ conditions were more
productive. Control lines had superior fitness in partial
shade at both high and low density, but control and
‘‘low light’’ selection lines attained similar fitness in
both types of full sun treatment. Trade-offs associated
with adapting to tolerate limited light availability apparently compromised the ability of selected lines to
take advantage of favorable conditions.
This factorial field study also provided estimates of
across-environment genetic correlations for fitness,
predicted to be negative across contrasting environments: e.g., the presumably most stressful and least
competitive (full sun, low density) as compared with
the least stressful and most competitive treatment (partial shade, high density). As in the previous ‘‘multiple
stress’’ study, across-environment genetic correlations
for fitness were always positive, and almost always
significantly so.
Results of these related studies suggest that Grime’s
model, designed to explain patterns of ecological specialization, may extend only partially to patterns of
within-species differentiation. While evolutionary responses to diverse stresses did favor a similar suite of
traits, ‘‘stress specialists’’ did not show the expected
fitness trade-offs across contrasting stressful vs. productive habitats, and also resembled ruderals rather
than stress-tolerators. Stanton et al. (2000) suggest that
short-lived species may respond to stress by improving
the ability to avoid or escape it, while longer-lived
species instead improve stress tolerance. This plausible
generalization should be viewed cautiously for at least
two reasons. First, as in any selection study, it is possible that the history of the progenitor population may
477
478
HILARY S. CALLAHAN
within small populations, potentially limiting standing
genetic variation and responses to selection. Because
the study lacked replicate lines or unselected controls,
they were careful to track each selection line’s effective population size, and the probability of mating
among relatives within lines. With these data in hand,
they were able to convincingly argue that inbreeding
depression, known to be environmentally dependent
(Schmitt and Ehrhardt, 1990; Wolfe, 1993), limited responses to selection (Kelly, 1999), especially in the
predicted high CO2 environment.
DO PHOTOMORPHOGENIC AND PHOTOPERIODIC
RESPONSES EVOLVE TOGETHER?
Using fluctuations in the light environment and
elaborate signal transduction mechanisms, plant cells
gather information about the current environment and
then use that information to adaptively modify their
phenotypes (Smith, 2000; Schmitt et al., 2003). One
of the best-studied photoreceptory systems, the phytochromes, uses sensitivity to the ratio of red and farred light (R:FR) in two ways: (1) light-regulation of
growth and development throughout the life cycle
(termed photomorphogenesis), and (2) sensing and responding to cyclic changes in the light environment
(i.e., circadian light/dark rhythms that themselves
change seasonally) (Kevei and Nagy, 2003; Mockler
et al., 2003; Klejnot and Lin, 2004). In serving this
dual function, phytochrome-mediated plasticity is associated with many different morphological and life
history traits, making this study system another example where three complex aspects of genetic architecture are relevant: (1) trait plasticities, which influence whether selection can maintain or change trait
means and variances, and which may themselves
evolve; (2) multiple traits showing plasticity; and (3)
variation in multiple environmental factors co-occurring.
Here, I briefly summarize my own work using artificial selection to study the integrated function and
evolution of photomorphogenic and photoperiod responses. The work focused on an important life history
trait, bolting time, in the model annual plant species
Arabidopsis thaliana (Callahan and Pigliucci, 2005).
In this and many other annual and biennial species,
bolting is the earliest manifestation of the transition to
reproduction. This non-labile trait can be scored by
checking plants daily for the appearance of ;1–2 mm
of an elongating inflorescence at the center of a whorl
of rosette leaves. At that point, either the number of
leaves or the chronological age of plants (in days), or
both, are recorded. In this species, both traits are typically accelerated by nearby vegetation that reduces R:
FR and by sufficiently long photoperiods. Both of
these plasticity responses are, in part, phytochromemediated. While the direction of this plasticity is predictable, its magnitude may vary considerably among
genotypes, and may be affected by daylength. The existence of this variation motivated our efforts to de-
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
the reliability of selection strategies involving extreme
or novel environments. If returning to ambient conditions evokes drastic phenotypic plasticity, among-genotype differences expressed under the conditions of
selection may be diminished, eliminated, or even reversed.
Overlooking these flaws, among this study’s most
relevant findings were differences between selected
and control genotypes in productivity, including 24%
greater above-ground dry matter accumulation, 21%
greater leaf area per plant, and higher leaf nitrogen
(Medrano et al., 1995). The investigators astutely ask
how could this happen without corresponding changes
in physiology? A possible answer: selected genotypes
had more leaf cells per unit area and smaller mean cell
volume, which can reduce maintenance respiration and
increase rates of CO2 transfer from the stomata to carbon-fixation sites in photosynthetic tissues (Delgado et
al., 1992a,b; Medrano et al., 1995). Although plausible, anatomical differences were subtle, and like the
physiological assays described above, failed to consider possible plastic responses triggered by very low
CO2 vs. ambient.
In general, efforts to understand plant productivity
both functionally and evolutionarily must account for
interplay between plastic and evolved responses. This
was strongly emphasized in Bradshaw and McNeilley’s (1991) review of potential plant responses to
increasing atmospheric CO2. Their review points out
that the genetic architecture of a complex phenotype
may itself change as the environment changes and suggests artificial selection a potentially useful approach
for testing the short-term evolutionary effects of global
change.
In this vein, Potvin and Tousignant (1996) used a
rapid-cycling strain of the self-incompatible annual
crop Brassica juncea (canola) to create two artificial
selection lines. Both were selected for an increase in
total fruit biomass, but one was grown in chamber conditions simulating a ‘‘predicted’’ future environment
(i.e., gradual increases in temperature regimes and CO2
concentrations) and the other in a chamber simulating
contemporary conditions. After seven generations of
selection, both lines were exposed to both environments and 14 vegetative and reproductive traits were
measured. Eleven of 14 traits showed significant phenotypic plasticity to contrasting environments, but
only one of those responses was consistent with adaptive phenotypic plasticity (i.e., the direction of the
plastic response matching the direction of selection
gradients). Seven of the 14 traits showed genetic differentiation in response to selection, but only one of
those traits showed a change consistent with adaptation.
Why could these experimental populations not become better adapted to either current or future conditions? In addition to choosing a self-incompatible annual and using a random mating scheme to minimize
low levels of inbreeding, Potvin and Tousignant recognized that some matings among relatives can occur
SELECTION
ON
PLASTIC PLANT PHENOTYPES
days to bolting (r 5 10.71, n 5 68, P , 0.001). It
was therefore expected that the HP and LP lines would
become differentiated not only for the plasticity index
targeted by artificial selection, but also for the plasticity index of this closely related bolting time trait. Parallel evolution of these two plasticity indices did occur
(Fig. 1b), corroborating earlier evidence that the two
traits—and presumably their associated plasticities—
are regulated pleiotropically (Mitchell-Olds, 1996).
The initial screening had also detected a positive
correlation between mean leaf number at bolting and
the plasticity index for this trait, but expected indirect
responses to selection did not occur consistently. Mean
leaf number at bolting did decrease in the LP lines,
but it decreased unexpectedly in the HP lines as well
(Fig. 1c). This suggests that the relationship between
a trait and its plasticity may be summarized rather
poorly by a simple, linear parameter. It also supports
the contention that the plasticity of a trait and the trait
itself can evolve independently rather than in tandem
(Bradshaw, 1965; Pigliucci, 2001).
Mirroring these decreases in mean leaf number in
both LP and HP lines, there were significant increases
in a photoperiod-mediated plasticity index (Fig. 1d).
In other words, response to disruptive selection on a
R:FR-mediated plasticity index resulted, indirectly, in
positive directional selection on the same trait’s photoperiod-mediated plasticity index. Although somewhat surprising, this makes biological sense: genotypes that bolt with very few leaves under inductive
photoperiod conditions would tend to have the greatest
capacity to delay bolting developmentally when those
signals are absent. Again, it seems that linear parameters may not be the most appropriate way to summarize relationships among a trait’s mean and its plasticity to one or more environmental variables.
Other evolutionary ecologists have also found that
the roles played by phytochromes are trait specific. For
example, recognizing that reduced R:FR conditions
not only accelerate bolting, but also alter seedling development, at least two studies have estimated genetic
correlations between bolting- and seedling-stage traits
or trait plasticities (Botto and Smith, 2002; Stenoien
et al., 2002). That such correlations are weak and nonsignificant is entirely consistent with what developmental geneticists have learned about the complex
functions of multiple phytochrome photoreceptors
(e.g., partially redundant and partially antagonistic
function, interactions with other photoreceptory and
hormonal systems, environment-specific gene action,
etc.) (Borevitz et al., 2002; Kevei and Nagy, 2003;
Mockler et al., 2003).
PLASTICITY TRIGGERED BY INTER-SPECIES INTERACTIONS
Many plant ecologists have addressed the function
and evolution of resistance traits that allow individuals
to avoid or reduce damage by herbivores or pathogens
as compared to individuals lacking such traits. Understanding how selection influences resistance traits is
clearly challenging because of three aspects of genetic
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
velop selection lines differentiated for R:FR-mediated
plasticity.
Recognizing that plasticity itself can be a heritable
trait (Scheiner, 1993), we first asked whether a natural
population of A. thaliana could harbor heritable variation for R:FR-mediated plasticity. Second, we examined whether or not there was parallel evolution of
two related bolting time trait plasticities, since the two
traits are often positively correlated and may be subject to pleiotropy (Mitchell-Olds, 1996; Callahan and
Pigliucci, 2002). Third, we asked whether a change in
the plasticity of a target trait would predictably alter
the mean of that same trait (Bradshaw, 1965; Brumpton et al., 1977; Falconer, 1990). Finally, because these
two distinct light signals jointly regulate a common
trait, and because the mechanisms for detecting and
responding to those signals are inter-related at the cellular level (Klejnot and Lin, 2004; Valverde et al.,
2004), we investigated the co-evolution of R:FR- and
photoperiod-mediated plasticities.
This project used a population originally native to
Kendalville, Michigan, that had already been brought
into laboratory cultivation by a commercial supplier.
From a bulk seed sample, a total of 68 maternal families had already been established (Camara et al.,
2000), and all families were screened in a preliminary
study (1) to confirm the general tendency of low R:
FR conditions to accelerate both leaf number at bolting
and days to bolting and (2) to quantify among-family
variation for both traits and for a plasticity index (R:
FR Index 5 1 2 [leaves at bolting in low R:FR/leaves
at bolting in high R:FR]). This dimension-less index
describes the extent to which low R:FR conditions accelerate the developmental stage at which bolting occurs. Another paper provides details of how two high
plasticity (HP1, HP2), two low plasticity (LP1, LP2),
and two control (C1, C2) lines were initially established from this base population and then subject to
two additional episodes of artificial selection targeting
this plasticity index (Callahan and Pigliucci, 2005).
Here, I briefly summarize a follow-up experiment. Genotypes from all six lines were grown in all four combinations of high and low R:FR conditions and long
and short photoperiods.
We expected to find a direct response to short-term
artificial selection, because an initial screening of the
base population had detected not only a significant R:
FR treatment effect and significant among-genotype
variation for leaf number at bolting, but also a significant treatment-by-genotype interaction. In other
words, there was variation among genotypes for the
R:FR-mediated plasticity, the target of our selection
protocol. A direct response to artificial selection was
indeed confirmed, with HP lines becoming significantly differentiated from the LP lines for the R:FRmediated plasticity index (Fig. 1a).
In addition, there were indirect responses to selection. For example, the initial screening study had detected a strong positive correlation between the plasticity index of leaf number and the plasticity index of
479
480
HILARY S. CALLAHAN
architecture: (1) the complex interplay between
evolved changes in basal (i.e., constitutive) resistance
traits and inducible (i.e., plastic) resistance traits combined with the potential for plasticity itself to evolve,
(2) multiple resistance traits and (3) overlapping variation in multiple environmental factors (i.e., multiple
enemies).
A number of conceptual and mathematical models
have tried to predict how constitutive and inducible
defenses should evolve. Briefly, if pests attack infrequently and defense chemicals are energetically costly,
selection should favor low basal levels but high inducible levels. On the other hand, frequent pest attacks
should deflate the energetic costs of producing high
basal levels of defense chemicals while simultaneously
rendering induction less effective. Theory therefore
predicts negative phenotypic and genetic correlations
between induced and basal levels of these resistance
biochemicals, as well as negative correlations between
defensive chemical production and fitness components,
particularly when pests are absent and where plant performance is already compromised (e.g., resource-poor,
competitive, or stressful environments) (Simms, 1992;
Bergelson and Purrington, 1996; Karban and Baldwin,
1997; Agrawal, 1998).
Manipulative field experiments have documented
pest-mediated natural selection for resistance traits, yet
natural populations typically harbor significant additive genetic variation for resistance traits (e.g., Berenbaum et al., 1986; Mauricio and Rausher, 1997).
Moreover, genetic correlation between basal and induced levels of defense chemicals are typically nonsignificant and sometimes positive rather than negative
(Zangerl and Berenbaum, 1990; Karban and Baldwin,
1997). From a mechanistic viewpoint, it is plausible
that constitutive and induced levels of defensive chemicals are regulated by common sets of genes exerting
positive pleiotropy, which would be detectable as a
positive genetic correlation (Siemens and MitchellOlds, 1998).
Given equivocal evidence from past lab and field
experiments, it is unsurprising that artificial selection
has been brought to bear on this challenging topic.
Here, I review studies by Siemens and colleagues
(1998, 2002), who developed the weedy annual Brassica rapa (syn. campestris) an advantageous model
system. Like all members of the mustard family, B.
rapa leaf cells contain myrosinase enzymes. After
pests damage cells, this enzyme hydrolyzes nitrogenand sulfur-containing glucosinolates to produce glucose, sulfate, isothiocynates, and other secondary compounds that can confer resistance to generalist and spe-
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
FIG. 1. Means (6 1 SE) for the selection and control lines for (a) the R:FR-mediated plasticity index for leaf number at bolting, (b) mean
number of leaves in high R:FR and long photoperiods conditions, (c) the R:FR-mediated plasticity index for days to bolting, and (d) the
photoperiod plasticity index for leaf number at bolting.
SELECTION
ON
PLASTIC PLANT PHENOTYPES
FIG. 2. Stronger responses occurred when artificial selection targeted increases (or decreases) in both myrosinase and total glucosinolates. Units are standardized values (zero mean, unit variance).
GHMH, filled circles; GHML, open diamonds; GLMH, open triangles;
GLML, filled squares. From Siemens et al., 2002.
rosinase and glucosinolate levels (Fig. 2). To maximize
their ability to detect costs of defense chemicals, Siemens et al. then grew all four lines in factorial combinations of herbivore induction and competition from
Lolium perenne (a common grass species). This design
aimed to elicit a very wide range of myrosinase and
glucosinolate concentrations and wide variation in performance. Costs were detected in the absence of competition, but not when competitors were present, directly contradicting theoretical predictions. Although
reciprocal effects of B. rapa on L. perenne were not
investigated in this field experiment, Siemens and his
collaborators hypothesized that these surprising results
may have occurred due to an interspecific interaction
other than resource competition, namely allelopathy.
Allelopathic interactions, defined as the production
and release of chemicals by one plant species that inhibit the performance of a co-occurring species, were
explored by conducting a series of lab assays. In one
of the follow-up studies presented by Siemens et al.
(2002), root elongation of L. perenne seedlings was
significantly suppressed when grown in the presence
of myrosinase combined with glucosinolates, but not
when grown in glucosinolates alone, or in the presence
of glucosinolates combined with denatured myrosinase. Their paper also describes a related follow-up
study that revealed significantly greater suppression of
L. perenne root growth when seedlings grew in the
presence of seeds from the MHGH lines as compared
to grown in the presence of MHGL, MLGH, or MLGL
seeds.
This work by Siemens and colleagues provides fascinating insights into the genetic architecture, evolutionary dynamics, and ecological function of a partic-
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
cialist insects (Chew, 1988; Agrawal and Kurashige,
2003). Total glucosinolates and myrosinase can be inferred by quantifying glucose under different conditions, and Seimens et al. (1998) developed rapid, automated assays to accommodate the large sample sizes
required for their artificial selection studies.
Two separate artificial selection studies were conducted. In one (Siemens and Mitchell-Olds, 1998), disruptive selection targeted basal glucosinolate only or
basal myrosinase only, creating two types of high and
low lines: GH, MH, GL, ML. In another (Siemens et al.,
2002), all four combinations of high and low glucosinolates were selected (GHMH, GHML, GLMH, GLML).
Although only basal levels of these compounds were
quantified during artificial selection, follow-up experiments to gauge responses to selection assayed not
only basal but also induced levels of defense chemicals, and induction by multiple enemies. Both studies
assumed that using a large number of plants and a selfincompatible annual would minimize the impact of inbreeding and inbreeding depression, and neither study
included tracking of mating between relatives, unselected controls, or replicated selection lines. Both
‘‘up’’ and ‘‘down’’ selection lines were included, presumably to serve as controls for one another (Falconer
and Mackay, 1996).
In the first experiment, selection on basal glucosinolates resulted in a GH selection line that had only
16.4% higher glucosinolate concentrations than the GL
line. In contrast, selection on basal myrosinase generated MH and ML lines a 2-fold difference. To assess
correlations between basal and induced levels of both
defense-related chemicals, follow-up studies were conducted with all four lines. Myrosinase and glucosinolate levels in the leaves of undamaged plants were
compared to levels in plants with leaves damaged by
Peiris xylostella caterpillars or the pathogen fungal
pathogen Leptosphaeria maculans (blackleg disease).
This protocol was also useful for quantifying induced
resistance to these two enemies.
For resistance and for levels of both chemicals, selection line 3 induction treatment interaction terms
were generally non-significant. One exception was the
MH line, which had higher pathogen-induced levels of
glucosinolates than undamaged plants, as well as higher pathogen-induced herbivore resistance. This result
suggests a positive correlation, rather than the expected negative correlation consistent with basal and induced defenses being alternative strategies. Costs of
resistance were also detected, using field conditions
where two types of herbivores occurred: leaf herbivory
by lepidopterans and seedling herbivory by a flea beetle. Even though the MH line had 10.3% greater resistance to flea beetles, an estimate of subsequent seed
production in that line was 15.5% lower.
In the second artificial selection experiment, in
which four divergent lines were created, there was a
stronger response to selection by the MHGH and MLGL
lines as compared to MHGL or MLGH lines. This indicates a positive genetic correlation between basal my-
481
482
HILARY S. CALLAHAN
ular system of defense chemicals. In addition, these
selection lines have proven extremely valuable in exploring other ecological interactions. Strauss et al.
(1999), for example, demonstrated that plants with
weaker defenses against herbivores experience enhanced interactions with pollinators. Although it is important to recognize that these results may be rather
specific to the myrosinase-glucosinolate system, it is
exciting to consider the potential for conducting similar experiments with artificial selection lines that have
been developed in other plant taxa (e.g., Marak et al.,
2000).
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
LIMITATIONS AND OPPORTUNITIES
Because plants are inexpensive and their care subject to few regulations, they resemble insects and other
invertebrates commonly used for artificial selection. In
the evolutionary biology literature, however, artificial
selection studies with plants are seldom maintained for
many generations (cf. Dudley, 1977) and replication is
limited or entirely lacking in many plant studies, including several reviewed here. This may be because
their size and generation times tend to be greater than
insects’, even in semelparous species. Moreover, the
space, light, water, and soil nutrition required by plants
can be costly. Such drawbacks are especially relevant
to studies of plastic plant traits, in which lines must
be maintained in multiple environments, sometimes involving factorial combinations or novel conditions
(e.g., manipulated lighting, altered CO2 concentrations). As in animal studies, logistical challenges escalate when selection protocols require assaying one
or multiple traits in all individuals in every generation.
While such challenges can be surmounted, they can
strongly influence other decisions about study duration, study design, selection protocols, and even the
choice of a model organism.
Of the studies reviewed here, most imposed artificial
selection for only two to four generations; the longest
imposed it for six (Potvin and Tousignant, 1996).
Many studies omitted replication, control lines, or
both. For Siemens and colleagues, for example, chose
to have ‘‘up’’ and ‘‘down’’ lines serving as controls
for each other (Falconer and Mackay, 1996). Their
choice most likely reflects some of the reasons
sketched above, particularly the labor- and cost-intensive scoring of myrosinase and glucosinolate traits.
Given their focus on relating biochemical and ecological aspects of resistance, and fitness consequences
thereof, extra effort in scoring phenotypes was essential and may have justified sacrificing replication.
Work by Medrano and colleagues and by Potvin and
Tousignant was more modest in scope, but also omitted replication, most likely because both studies involved novel manipulations of CO2.
Stanton et al. (2000), in contrast, more rigorously
include replicate lines and unselected controls. Their
work also entailed quantifying multiple fitness components and other performance traits. This was appropriate given their interest in whether cross-environ-
ment correlations for fitness are positive or negative.
Similarly, my own work (Callahan and Pigliucci,
2005) quantified the plasticity of a non-fitness trait,
which required growing lines in novel factorial combinations of light conditions and scoring hundreds of
individuals in every generation for a timing trait. The
decision to work with the diminutive and selfing annual A. thaliana was strongly influenced by these challenges.
Unselected control lines, replicate selection lines, or
both, are crucial for gauging how genetic drift, inbreeding, and inbreeding depression limit responses to
selection (Lynch and Walsh, 1998; Kelly, 1999). To
minimize inbreeding, many researchers work with selfincompatible plant species and impose random mating
(Siemens and Mitchell-Olds, 1998; Stanton et al.,
2000; Siemens et al., 2002), but it can be difficult to
eliminating inbreeding entirely since artificial selection
tends to reduce effective population sizes (Ne). Small
Ne both intensifies the impact of drift and increases the
probability of matings among relatives. These realities
motivated Potvin and Tousignant (1996) to track variation among parental plants in their contributions to
the next generation, to adjust Ne accordingly, to calculate overall rates of inbreeding, and to quantitatively
assess inbreeding depression. In contrast, Siemens and
colleagues simply took steps to maintain large Ne in
the hope of minimizing impacts of drift and inbreeding. Stanton et al. (2000), who rigorously opted for
extensive replication and imposed random mating with
hand-pollinations, often observed that some but not all
replicates showed evolutionary responses to selection.
When a replicated selection study finds limited responses to selection, a helpful technique for distinguishing the impacts of drift vs. inbreeding depression
is inter-mating replicate lines (e.g., Wijngaarden and
Brakefield, 2001), and this option could be pursued by
Stanton and colleagues.
Some plant researchers have instead chosen selfing
species, most prominently A. thaliana, for artificial selection studies. Such a choice raises a completely different set of limitations and opportunities. On the one
hand, selfing species can be problematic because limited recombination means that selection proceeds primarily via line sorting, akin to what happens in clonal
organisms (e.g., Bell, 1997; Kolodynska and Pigliucci,
2003; Van Kleunen and Fischer, 2003). In studies addressing constraints on genetic architecture, it can be
difficult to distinguish when a correlated response to
selection (or a lack of selection response) comes about
due to true pleiotropy or simply due to genetic correlations that came about due to past selection and subsequently limited recombination (i.e., linkage disequilibrium: Lynch and Walsh, 1998). High levels of LD
were clearly worth considering in assessing the artificially selected A. thaliana lines in my own study (Callahan and Pigliucci, 2005). Nonetheless, results were
consistent with other researchers who have more rigorously distinguished pleiotropy from LD using recombinant inbred lines and QTL mapping strategies
SELECTION
ON
PLASTIC PLANT PHENOTYPES
greater plasticity can be achieved by growing successive generations in alternating environments while developing control lines in a single, constant environment (Scheiner, 2002). Employing either of these approaches raises the issue of whether environmental effects carry over from one generation to the next,
obscuring or amplifying responses to selection (Mousseau and Fox, 1998). Carryover effects due to maternal
genotype or maternal environment can be especially
pervasive in plants, with dramatic impacts due to seed
dormancy or longevity (e.g., Shaw et al., 2000; Donohue et al., 2005a) potentially influencing other adult
traits (Donohue et al., 2005b). To minimize, or at least
homogenize, the impact of such maternal effects, some
researchers routinely postpone evaluating responses to
selection until all selected and control lines have been
grown in a uniform environment for at least one generation. Even more desirable is combining this strategy
with control lines grown alongside lines subject to a
particular selection regime (e.g., Stanton et al., 2000),
which permit tracking the impact of environmental effects from generation to generation.
Finally, it should be mentioned that the studies reviewed here, and many other plant studies, favor just
a few taxa such as A. thaliana or Brassica species in
the mustard family (Brassicaceae). Such taxonomic parochialism mirrors the intense focus on Drosophila, E.
coli, and models in other kingdoms. Despite some limitations on generalizing results from model species to
a broader range of taxa, there are tremendous opportunities for integrating artificial selection studies with
post-genomic approaches. In plants, this can be done
with Arabidopsis and relatives in the Brassicaceae,
with rice and other grasses, or with tobacco and other
members of the Solanaceae. Just as within-species
QTL mapping data have been helpful in interpreting
the results of A. thaliana studies (e.g., Mitchell-Olds,
1996; Callahan and Pigliucci, 2005), comparative QTL
mapping techniques within the Brassicaceae (Osborn
et al., 1997) have the potential to illuminate artificial
selection studies conducted with B. rapa or S. arvense.
Myrosinase, glucosinolate, and insect resistance QTLs
mapped in A. thaliana (Kliebenstein et al., 2002), for
example, may to be directly relevant to the results of
Siemens and colleagues. Given this potential, and
many other similar opportunities, there is little doubt
that artificial selection will continue playing a major
role in addressing eco- and evolutionary plant physiology questions, particularly questions about phenotypically plastic traits.
ACKNOWLEDGMENTS
I thank Ted Garland and John Swallow for their interest in botanical points of view. Diane Byers, Nile
Kurashige, Sam Scheiner, and an anonymous reviewer
provided constructive comments on earlier drafts. A
Barnard College Special Assistant Professor Leave
provided teaching relief and NSF IBN 97-97552 and
IBN 03-44518 have supported my research activities.
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
(Mitchell-Olds, 1996). Other researchers have attempted to circumvent this limitation of A. thaliana by artificially imposing recombination. Ward et al. (2000),
for example, conducted random crosses among diverse
ecotypes and then used F2 seeds as the base population
for artificial selection. Yet Ward’s more artificial approach and results may be biased relative to natural A.
thaliana populations if high levels of LD are in fact
the norm in the wild (Pigliucci, 2003). Two types of
studies will be helpful in illuminating the role of LD
as a constraint to selection: (1) combining development of artificial selection lines with mapping techniques (Ungerer et al., 2003; Ungerer and Rieseberg,
2003) and (2) developing and/or assaying artificial selection lines in field as well as lab conditions.
These observations point toward some of the advantages of working with a selfing species. First, researchers are interested in pursuing QTL mapping projects can inter-mate contrasting selection lines to create
populations of recombinant inbred lines. This task may
be more easily accomplished with selfing species
(Conner, 2003). Second, researchers can avoid laborintensive artificial pollination procedures, whilst simultaneously gaining confidence that reproductive
output (i.e., fitness) is not dependent on the hand-pollination procedures employed in the lab or glasshouse.
Third, selfing is hardly atypical among natural or crop
plants; a predominantly selfing mating system is found
in ;20% of annuals (Barrett, 2002). Finally, an enormous wealth of functional genomics research is available for A. thaliana and for rice (Oryza sativa)—both
premier model organism for genomics, both selfing
species, and one of them the world’s leading crop economically (Kikuchi et al., 2003). Studies with these
and other crop species are likely to provide important
insights, aid interpretation of other eco- and evolutionary and physiology research, and possibly make results
more relevant to the applied plant science community.
Choosing to work with crops, or simply drawing the
base population for selection from a single location,
can introduce a constraint deriving from past selection
responses. This is, of course, a general problem of
quantitative genetic studies, rather than a weakness
specific to artificial selection strategies. Moreover,
such constraints can exist even in species with obligate
outcrossing or mixed mating systems. Medrano and
colleagues, who used an inbred strain of tobacco, were
well aware of this possibility and chose to augment
the strain’s limited genetic variability with mutagenesis. In general, comparison of selected and control
lines derived from mutagenized vs. unmutagenized
base populations would be useful, potentially providing important insights into constraints due to past selection (Pigliucci, 2003).
Before concluding, two additional strategies employed in studies of plastic plant traits should be mentioned, even though they were not featured in the research reviewed. First, plant researchers sometimes
take advantage of the extended viability of seeds, basically ‘‘storing’’ a control line. Second, selection for
483
484
HILARY S. CALLAHAN
REFERENCES
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
Ackerly, D. D., S. A. Dudley, S. E. Sultan, J. Schmitt, J. S. Coleman,
C. R. Linder, D. R. Sandquist, M. A. Geber, A. S. Evans, T. E.
Dawson, and M. J. Lechowicz. 2000. The evolution of plant
ecophysiological traits: Recent advances and future directions.
Bioscience 50:979–995.
Ackerly, D. D. and R. K. Monson. 2003. Waking the sleeping giant:
The evolutionary foundations of plant function. Internat. J. Plant
Sci. 164:S1–S6.
Agrawal, A. A. 1998. Induced responses to herbivory and increased
plant performance. Science 279:1201–1202.
Agrawal, A. A. 2001. Phenotypic plasticity in the interactions and
evolution of species. Science 294:321–326.
Agrawal, A. A. 2003. Community genetics: New insights into community ecology by integrating population genetics. Ecology 84:
543–544.
Agrawal, A. A. and N. S. Kurashige. 2003. A role for isothiocyanates in plant resistance against the specialist herbivore Pieris
rapae. Journal of Chemical Ecology 29:1403–1415.
Barrett, S. C. H. 2002. The evolution of plant sexual diversity. Nature Reviews Genetics 3:274–284.
Bell, G. and M. J. Lechowicz. 1994. Spatial heterogeneity at small
scales and how plants respond to it. In M. M. Caldwell and R.
W. Pearcy (eds.), Exploitation of environmental heterogeneity
by plants, pp. 391–414. Academic Press, San Diego.
Bell, G. A. 1997. Experimental evolution in Chlamydomonas. I.
Short-term selection in uniform and diverse environments. Heredity 78:490–497.
Berenbaum, M. R., A. R. Zangerl, and J. K. Nitao. 1986. Constraints
on chemical coevolution—wild parsnips and the parsnip webworm. Evolution 40:1215–1228.
Bergelson, J. and C. B. Purrington. 1996. Surveying patterns in the
cost of resistance in plants. American Naturalist 148:536–558.
Borevitz, J. O., J. N. Maloof, J. Lutes, T. Dabi, J. L. Redfern, G. T.
Trainer, J. D. Werner, T. Asami, C. C. Berry, D. Weigel, and J.
Chory. 2002. Quantitative trait loci controlling light and hormone response in two accessions of Arabidopsis thaliana. Genetics 160:683–696.
Botto, J. F. and H. Smith. 2002. Differential genetic variation in
adaptive strategies to a common environmental signal in Arabidopsis accessions: Phytochrome-mediated shade avoidance.
Plant Cell and Environment 25:53–63.
Bradshaw, A. D. 1965. Evolutionary significance of phenotypic plasticity in plants. Advances in Genetics 13:115–155.
Bradshaw, A. D. and T. McNeilly. 1991. Evolutionary response to
global climatic change. Annals of Botany 67:5–14.
Brakefield, P. M. 2003. Artificial selection and the development of
ecologically relevant phenotypes. Ecology 84:1661–1671.
Brumpton, R. J., H. Boughey, and J. L. Jinks. 1977. Joint selection
for both extremes of mean performance and of sensitivity to a
macroenvironmental variable 1. Family selection. Heredity 38:
219–226.
Callahan, H. S. and M. Pigliucci. 2002. Shade-induced plasticity and
its ecological significance in wild populations of Arabidopsis
thaliana. Ecology 83:1965–1980.
Callahan, H. S. and M. Pigliucci. 2005. Indirect consequences of
artificial selection on plasticity to light quality in Arabidopsis
thaliana. J. Evol. Biol.
Callahan, H. S., M. Pigliucci, and C. D. Schlichting. 1997. Developmental phenotypic plasticity: Where ecology and evolution
meet molecular biology. Bioessays 19:519–525.
Camara, M. D., C. A. Ancell, and M. Pigliucci. 2000. Induced mutations: A novel tool to study phenotypic integration and evolutionary constraints in Arabidopsis thaliana. Evol. Ecol. Res.
2:1009–1029.
Chapin, F. S., K. Autumn, and F. Pugnaire. 1993. Evolution of suites
of traits in response to environmental stress. Am. Nat. 142:S78–
S92.
Chew, F. S. 1988. Searching for defensive chemistry in the Cruciferae, or, do glucosinolates always control interactions of Cruciferae with their potential herbivores and symbionts? In K. S.
Spencer (ed.), Chemical mediation of coevolution, pp. 81–112.
Academic Press, New York.
Conner, J. K. 2003. Artificial selection: A powerful tool for ecologists. Ecology 84:1650–1660.
Delgado, E., J. Azconbieto, X. Aranda, J. Palazon, and H. Medrano.
1992a. Leaf photosynthesis and respiration of high CO2-grown
tobacco plants selected for survival under CO2 compensation
point conditions. Plant Physiol. 98:949–954.
Delgado, E., M. A. J. Parry, D. W. Lawlor, A. J. Keys, and H.
Medrano. 1993. Photosynthesis, ribulose-1,5-bisphosphate carboxylase and leaf characteristics of Nicotiana tabacum L. genotypes selected by survival at low CO2 concentrations. J. Exper. Bot. 44:1–7.
Delgado, E., M. A. J. Parry, J. Vadell, D. W. Lawlor, A. J. Keys,
and H. Medrano. 1992b. Effect of water stress on photosynthesis, leaf characteristics and productivity of field-grown Nicotiana tabacum L. genotypes selected for survival at low CO2. J.
Exper. Bot. 43:1001–1008.
Delgado, E., J. Vadell, and H. Medrano. 1994. Photosynthesis during
leaf ontogeny in field-grown Nicotiana tabacum L. lines selected by survival at low CO2 concentrations. J. Exper. Bot. 45:
547–552.
Donohue, K., L. Dorn, C. Griffith, E. S. Kim, A. Aguilera, C. R.
Polisetty, and J. Schmitt. 2005a. Environmental and genetic influences on the germination of Arabidopsis thaliana in the field.
Evolution 59:740–757.
Donohue, K., L. Dorn, C. Griffith, E. S. Kim, A. Aguilera, C. R.
Polisetty, and J. Schmitt. 2005b. Niche construction through
germination cueing: Life history responses to timing of germination in Arabidopsis thaliana. Evolution 59:771–785.
Donohue, K., D. Messiqua, E. H. Pyle, M. S. Heschel, and J.
Schmitt. 2000. Evidence of adaptive divergence in plasticity:
Density- and site-dependent selection on shade-avoidance responses in Impatiens capensis. Evolution 54:1956–1968.
Dudley, J. W. 1977. 76 Generations of selection for oil and protein
percentage in maize. International Conference on Quantitative
Genetics, Iowa State University Press.
Falconer, D. S. 1990. Selection in different environments: Effects
on environmental sensitivity (reaction norm) and on mean performance. Genetical Research 56:57–70.
Falconer, D. S. and T. F. C. Mackay. 1996. Introduction to quantitative genetics. Longman, Essex.
Fry, J. D. 2003. Detecting ecological trade-offs using selection experiments. Ecology 84:1672–1678.
Fuller, R. C., C. F. Baer, and J. Travis. 2005. How and when selection
experiments might actually be useful. Integr. Comp. Biol. 45:
391–404.
Geber, M. A. and T. E. Dawson. 1993. Evolutionary responses to
global change. In P. M. Kareiva, J. G. Kingsolver, and R. B.
Huey (eds.), Biotic interactions and global change, pp. 165–
178. Sinauer Associates, Sunderland, Massachusetts.
Geber, M. A. and L. R. Griffen. 2003. Inheritance and natural selection on functional traits. Internat. J. Plant Sci. 164:S21–S42.
Grime, J. P. 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary
theory. Am. Nat. 111:1169–1194.
Grime, J. P. 1979. Plant strategies and vegetation processes. John
Wiley, Chichester, U.K.
Karban, R. and I. T. Baldwin. 1997. Induced responses to herbivory.
University of Chicago Press, Chicago.
Kelly, J. K. 1999. Response to selection in partially self-fertilizing
populations. I. Selection on a single trait. Evolution 53:336–
349.
Kevei, E. and F. Nagy. 2003. Phytochrome controlled signalling cascades in higher plants. Physiologia Plantarum 117:305–313.
Kikuchi, S., et al. 2003. Collection, mapping, and annotation of over
28,000 cDNA clones from japonica rice. Science 301:376–379.
Klejnot, J. and C. T. Lin. 2004. A CONSTANS experience brought
to light. Science 303:965–966.
Kliebenstein, D., D. Pedersen, B. Barker, and T. Mitchell-Olds. 2002.
Comparative analysis of quantitative trait loci controlling glucosinolates, myrosinase and insect resistance in Arabidopsis
thaliana. Genetics 161:325–332.
SELECTION
ON
PLASTIC PLANT PHENOTYPES
The adaptive evolution of plasticity: Phytochrome-mediated
shade avoidance responses. Integr. Comp. Biol. 43:459–469.
Shaw, R. G., D. L. Byers, and E. Darmo. 2000. Spontaneous mutational effects of reproductive traits of Arabidopsis thaliana.
Genetics 155:369–378.
Siemens, D. H., S. H. Garner, T. Mitchell-Olds, and R. M. Callaway.
2002. Cost of defense in the context of plant competition: Brassica rapa may grow and defend. Ecology 83:505–517.
Siemens, D. H. and T. Mitchell-Olds. 1998. Evolution of pest-induced defenses in Brassica plants: Tests of theory. Ecology 79:
632–646.
Simms, E. L. 1992. Costs of plant resistance to herbivory. In R. S.
Fritz and E. L. Simms (eds.), Plant resistance to herbivores and
pathogens, pp. 392–425. University of Chicago Press, Chicago.
Smith, H. 2000. Phytochromes and light signal perception by
plants—an emerging synthesis. Nature 407:585–91.
Stanton, M. L., B. A. Roy, and D. A. Thiede. 2000. Evolution in
stressful environments. I. Phenotypic variability, phenotypic selection, and response to selection in five distinct environmental
stresses. Evolution 54:93–111.
Stanton, M. L., D. A. Thiede, and B. A. Roy. 2004. Consequences
of intraspecific competition and environmental variation for selection in the mustard Sinapsis arvensis: Contrasting ecological
and evolutionary perspectives. Am. Nat. 164:736–752.
Stenoien, H. K., C. B. Fenster, H. Juittinen, and O. Savolainen. 2002.
Quantifying latitudinal clines to light responses in natural populations of Arabidopsis thaliana (Brassicaceae). Am. J. Bot. 89:
1604–1608.
Strauss, S. Y., D. H. Siemens, M. B. Decher, and T. Mitchell-Olds.
1999. Ecological costs of plant resistance to herbivores in the
currency of pollination. Evolution 53:1105–1113.
Taylor, D. R. and L. W. Aarssen. 1988. An interpretation of phenotypic plasticity in Agropyron repens (Graminae). Am. J. Bot.
75:401–413.
Ungerer, M. C., C. R. Linder, and L. H. Rieseberg. 2003. Effects of
genetic background on response to selection in experimental
populations of Arabidopsis thaliana. Genetics 163:277–286.
Ungerer, M. C. and L. H. Rieseberg. 2003. Genetic architecture of
a selection response in Arabidopsis thaliana. Evolution 57:
2531–2539.
Valverde, F., A. Mouradov, W. Soppe, D. Ravenscroft, A. Samach,
and G. Coupland. 2004. Photoreceptor regulation of CONSTANS protein in photoperiodic flowering. Science 303:1003–
1006.
Van Kleunen, M. and M. Fischer. 2003. Effects of four generations
of density-dependent selection on life history traits and their
plasticity in a clonally propagated plant. J. Evol. Biol. 16:474–
484.
Van Tienderen, P. and A. Van Hinsberg. 1996. Phenotypic plasticity
in growth habit in Plantago lanceolata: How tight is a suite of
correlated characters. Plant Species Biol. 11:87–96.
Via, S., R. Gomulkiewicz, G. D. Jong, S. M. Scheiner, C. D.
Schlichting, and P. H. V. Tienderen. 1995. Adaptive phenotypic
plasticity: Consensus and controversy. Trends Ecol. Evol. 10:
212–216.
Via, S. and R. Lande. 1985. Genotype-environment interaction and
the evolution of phenotypic plasticity. Evolution 39:505–522.
Ward, J. K., J. Antonovics, and R. B. Thomas. 2000. Is atmospheric
CO2 a selective agent on model C3 annuals? Oecologia 123:
330–341.
Westoby, M., D. S. Falster, A. T. Moles, and I. J. Wright. 2002. Plant
ecological strategies: Some leading dimensions of variation between species. Ann. Rev. Ecol. System. 33:125–159.
Wijngaarden, P. J. and P. M. Brakefield. 2001. Lack of response to
artificial selection on the slope of reaction norms for seasonal
polyphenism in the butterfly Bicyclus anynana. Heredity 87:
410–420.
Wolfe, L. M. 1993. Inbreeding depression in Hydrophyllum appendiculatum—role of maternal effects, crowding, and parental
mating history. Evolution 47:374–386.
Zangerl, A. R. and M. R. Berenbaum. 1990. Furanocoumarin induction in wild parsnip—Genetics and populational variation.
Ecology 71:1933–1940.
Downloaded from http://icb.oxfordjournals.org/ at Pennsylvania State University on February 27, 2014
Kolodynska, A. and M. Pigliucci. 2003. Multivariate responses to
flooding in Arabidopsis: An experimental evolutionary investigation. Funct. Ecol. 17:131–140.
Lande, R. and S. J. Arnold. 1983. The measurement of selection on
correlated characters. Evolution 37:1210–1226.
Lortie, C. and L. W. Aarssen. 1996. The specialization hypothesis
for phenotypic plasticity in plants. Internat. J. Plant Sci. 157:
484–487.
Lynch, M. and B. Walsh. 1998. Genetics and analysis of quantitative
traits. Sinauer Associates Inc., Sunderland, Massachusetts.
Marak, H. B., A. Biere, and J. M. M. Van Damme. 2000. Direct
and correlated responses to selection on iridoid glycosides in
Plantago lanceolata L. J. Evol. Biol. 13:985–996.
Mauricio, R. and M. D. Rausher. 1997. Experimental manipulation
of putative selective agents provides evidence for the role of
natural enemies in the evolution of plant defenses. Evolution
51:1435–1444.
Medrano, H., A. J. Keys, D. W. Lawlor, M. A. J. Parry, J. Azconbieto, and E. Delgado. 1995. Improving plant production by
selection for survival at low CO2 concentrations. J. Exper. Bot.
46:1389–1396.
Medrano, H. and E. Primomillo. 1985. Selection of Nicotiana tabacum haploids of high photosynthetic efficiency. Plant Physiol.
79:505–508.
Mitchell-Olds, T. 1996. Genetic constraints on life-history evolution:
Quantitative-trait loci influencing growth and flowering in Arabidopsis thaliana. Evolution 50:140–145.
Mitchell-Olds, T. and J. Bergelson. 2000. Biotic interactions—Genomics and coevolution—Editorial overview. Curr. Opin. Plant
Biol. 3:273–277.
Mockler, T., H. Yang, X. Yu, D. Parikh, Y.-C. Cheng, S. Dolan, and
C. Lin. 2003. Regulation of photoperiodic flowering by Arabidopsis photoreceptors. Proc. Nat. Acad. Sci. 100:2140–2145.
Mousseau, T. A. and C. W. Fox. (eds.) 1998. Maternal effects as
adaptations. Oxford University Press, New York.
Osborn, T. C., C. Kole, I. A. P. Parkin, A. G. Sharpe, M. Kuiper, D.
J. Lydiate, and M. Trick. 1997. Comparison of flowering time
genes in Brassica rapa, B. napus and Arabidopsis thaliana. Genetics 146:1123–1129.
Pigliucci, M. 1996. How organisms respond to environmental changes: From phenotypes to molecules (and vice versa). Trends Ecol.
Evol. 11:168–173.
Pigliucci, M. 2001. Phenotypic plasticity; Beyond nature and nurture. The Johns Hopkins University Press, Baltimore.
Pigliucci, M. 2003. Selection in a model system: Ecological genetics
of flowering time in Arabidopsis thaliana. Ecology 84:1700–
1712.
Potvin, C. and D. Tousignant. 1996. Evolutionary consequences of
simulated global change: Genetic adaptation or adaptive phenotypic plasticity. Oecologia 108:683–693.
Rausher, M. D. 1992. The measurement of selection on quantitative
traits: Biases due to environmental covariances between traits
an fitness. Evolution 46:616–626.
Sage, R. F. and J. R. Coleman. 2001. Effects of low atmospheric
CO2 on plants: More than a thing of the past. Trends Plant Sci.
6:18–24.
Scheiner, S. M. 1993. Genetics and evolution of phenotypic plasticity. Ann. Rev. Ecol. System. 24:35–68.
Scheiner, S. M. 2002. Selection experiments and the study of phenotypic plasticity. J. Evol. Biol. 15:889–898.
Schlichting, C. D. and H. Smith. 2002. Phenotypic plasticity: Linking molecular mechanisms with evolutionary outcomes. Evol.
Ecol. 16:189–211.
Schmitt, J. and D. W. Ehrhardt. 1990. Enhancement of inbreeding
depression by dominance and suppression in Impatiens capensis. Evolution 44:269–278.
Schmitt, J., A. C. McCormac, and H. Smith. 1995. A test of the
adaptive plasticity hypothesis using transgenic and mutant
plants disabled in phytochrome-mediated elongation responses
to neighbors. Am. Nat. 146:937–953.
Schmitt, J., J. R. Stinchcombe, M. S. Heschel, and H. Huber. 2003.
485