Genes affecting phenotypic plasticity in Arabidopsis: pleiotropic effects and reproductive fitness of photomorphogenic mutants


Massimo Pigliucci Depts. of Botany and of Ecology & Evolutionay Biology, University of Tennessee, Knoxville, TN 37996-1100, USA. Tel: 423 974 6221; fax: 423 974 0978; e-mail:


Many plants exhibit characteristic photomorphogenic shade ’avoidance’ responses to crowding and vegetation shade; this plasticity is often hypothesized to be adaptive. We examined the contribution of specific photomorphogenic loci to plastic shade avoidance responses in the annual crucifer Arabidopsis thaliana by comparing single-gene mutants defective at those loci with wild type plants exhibiting normal photomorphogenesis. The hy1 and hy2 mutants, deficient in all functional phytochromes, were less plastic than the wild type in response to a nearby grass canopy or to a low-red/far-red light ratio characteristic of vegetation shade. These mutants displayed constitutively shade-avoiding phenotypes throughout the life cycle regardless of the treatment: they bolted at an earlier developmental stage and were characterized by reduced branching. In contrast, the hy4 mutant, deficient in blue light reception, exhibited greater plasticity than the wild type in response to vegetation shade after the seedling stage. This mutant produced more leaves before bolting and more basal branches under normal light conditions when compared to the wild type. These results indicate that specific photomorphogenic loci have different and sometimes antagonistic pleiotropic effects on the plastic response to vegetation shade throughout the life cycle of the plant. The fitness of the constitutively shade-avoiding phytochrome-deficient mutants was lower than that of the plastic wild type under normal light, but was not different in the vegetation shade treatments, where all genotypes converged toward similar shade avoidance phenotypes. This outcome supports one key prediction of the adaptive plasticity hypothesis: that inappropriate expression of shade avoidance traits is maladaptive.


Phenotypic plasticity, the expression by a genotype of different phenotypes in response to different environmental conditions, is thought to be a major component of phenotypic evolution (Bradshaw, 1965; Schlichting, 1986; Sultan, 1987; West-Eberhard, 1989; Schlichting & Pigliucci, 1998). Yet, we still know very little about the genetic mechanisms underlying most plastic responses (Scheiner, 1993; Pigliucci, 1996). A better understanding of the genetic basis of phenotypic plasticity is necessary for several reasons. First, it will inform us about genetic constraints on the evolution of plasticity, thereby aiding the reconstruction of past evolutionary histories that led to existing reaction norms (Schlichting & Pigliucci, 1995). Second, it will allow improvement of models of future evolutionary trajectories, which will be based on a mechanistic, rather than statistical, description of current constraints (see Ward, 1994, for an example). Third, it will provide information about patterns of differential gene expression across ecologically relevant environments, hence offering insights into adaptive phenotypic evolution.

An ecologically significant example of plasticity is the suite of photomorphogenic responses elicited in plants by vegetation shade (Casal & Smith, 1989). Many plants display an array of morphological ‘shade-avoidance’ responses to the reduction in red/far-red ratio (R:FR) characteristic of foliage shade due to the selective absorption of red wavelengths by chlorophyll (Smith, 1982). Typical shade avoidance responses include enhanced shoot elongation, reduced branching and accelerated bolting in rosette plants under low R:FR. These responses have been hypothesized to be a form of adaptive plasticity which allows plants to detect neighbours and develop a morphology that enhances light capture in dense stands (Casal & Smith, 1989; Schmitt & Wulff, 1993). To support the adaptive plasticity hypothesis, it is necessary to demonstrate both that shade avoidance phenotypes confer higher fitness in dense vegetation, and that they lower fitness due to physiological or biomechanical costs when plants are grown in isolation. Results consistent with this prediction have emerged from recent experiments (Schmitt et al., 1995; Dudley & Schmitt, 1996).

The genetic and molecular bases of plant responses to light are currently the subject of active investigation, and several classes of genes have been shown to play a role in photomorphogenesis (Bagnall, 1992; Chory, 1993; Quail et al., 1995; Smith, 1995). Many of these studies have employed the annual crucifer Arabidopsis thaliana as a model system. The primary photoreceptors include the family of phytochrome molecules, which convert reversibly between R- and FR-absorbing states and the cryptochrome family of blue light receptors, as well as an unidentified UV-B receptor. In A. thaliana, five phytochrome genes (PHY A–E) have been identified with separate although potentially overlapping functions (Quail et al., 1995; Smith, 1995). In green plants, shade avoidance responses are thought to be mediated largely by light-stable phytochromes, including B. The blue light receptor, on the other hand, is a sensor of light directionality and intensity, but is insensitive to variation in R:FR (normally associated with vegetation shade). Light reception systems in plants are extremely complex and sophisticated, and several other genes have already been uncovered that are involved in the signal transduction pathways leading from the reception of the light signal to the morphogenic effects (Deng et al., 1991).

Much of the recent progress in elucidating the function of specific photomorphogenic genes has resulted from experiments with mutants in which the function under study is impaired (Kendrick et al., 1994). Many of these mutants were originally identified because their seedlings exposed to white light display an elongated hypocotyl (embryonic shoot) compared with de-etiolating (i.e. halting elongation and opening the cotyledons under white light) wild type seedlings. In other words, the normal plastic photomorphogenic reaction is disabled in these mutants (Liscum & Hangarter, 1991; Yanovsky et al., 1995). Such photomorphogenic mutants provide a valuable opportunity for evolutionary investigations of shade avoidance responses. By examining mutants in which plasticity to light is impaired, it is possible to study the pleiotropic effects of specific genes on plasticity of a suite of ecologically relevant traits. Moreover, by quantifying the fitness of mutants which constitutively express a given phenotype, relative to plastic wild type plants across a range of environments, it is possible to investigate the adaptive significance of shade avoidance responses (Schmitt et al., 1995).

Evolutionary biologists have historically been skeptical of research conducted on mutants, under the pretext that the resulting phenotypes are too extreme, and that very likely the genetic basis of variation for the traits of interest under natural conditions are to be found in ‘modifier’ genes with minor effects. Both these assumptions have been challenged by recent research merging evolutionary and mechanistic approaches (Pigliucci, 1998). First, the goal of evolutionary studies of mutants is to be able to carry out phenotypic manipulations to explore the relationship between the trait of interest (plasticity in this case) and fitness. This is a crucial step to establish a case for the adaptive meaning of the characters in question (Reznick & Travis, 1996; Rose et al., 1996; Bell, 1997). Phenotypic manipulation does benefit from the availability of ‘extreme’ phenotypes, such as in the case of well-established evolutionary studies manipulating feathers length in birds or flower appearance in plants (Futuyma, 1998). The recent use of transgenic organisms to address evolutionary questions is another example (Linder & Schmitt, 1995; Purrington & Bergelson, 1995; Rogers & Parkes, 1995; Schmitt et al., 1995; Weigel & Nilsson, 1995). Second, recent reports have established that – contrary to popular belief – natural populations do show genetic variation at major regulatory loci, such as the heat shock complex in Drosophila (Krebs & Feder, 1997). Therefore, we feel that this relatively new path of combining molecularly induced phenotypic manipulation and ecological experiments has the potential to enlarge the usual scope of evolutionary questions and further our understanding of adaptive evolution and of the macroevolutionary changes necessary to yield phenotypic novelties (West-Eberhard, 1989).

Here we discuss an experiment in which we used elongated hypocotyl mutants of A. thaliana to investigate the impact of specific photomorphogenic loci on plasticity to vegetation shade. This species displays shade avoidance responses to reduced R:FR, including hypocotyl elongation, accelerated bolting and shoot elongation (Whitelam & Smith, 1991; Robson et al., 1993; Halliday et al., 1994). In the wild, A. thaliana is an opportunistic colonizer of open habitats, which competes for light with other species and is limited in its distribution by the presence of tall grass (M. Pigliucci, H. Callahan and S. Andersson, personal observation). The shade avoidance response may thus play an important role in natural populations and it is therefore of interest to investigate its genetic basis and fitness consequences.

We compared the responses of several phytochrome, blue receptor, and signal transduction mutants with those of wild type A. thaliana plants in order to address the following specific questions:

1 What are the pleiotropic effects (Van Tienderen et al., 1996) of photomorphogenic genes, originally discovered because of their phenotypes at the seedling stage, on the plasticity of ecologically relevant traits expressed later in development? In particular, does the magnitude and nature of pleiotropy through the life cycle differ among genes affecting the phytochrome photoreceptors, the blue light receptor and a transduction gene downstream from these receptors?

2 Are the effects of photomorphogenic genes environment-specific? Namely, how do the effects depend upon the presence or absence of a light cue that triggers a developmental pathway?

3 What is the impact of each photomorphogenic gene on fitness? If there were a cost to expressing the shade avoidance phenotype in the absence of competition, we would predict that mutants expressing constitutive shade-avoidance phenotypes would have a lower fitness when compared to the plastic wild type in a high light environment. On the other hand, the two classes of genotypes should have similar fitness in a low R:FR environment, where the shade avoidance phenotype is also expressed by the wild type.

Throughout the following, it is important to notice that lack of adaptive plasticity may be due to either the inability to detect the environmental change, or to the inability to generate the appropriate morphology once the environmental condition has been assessed. We concentrated mostly on the first aspect of the problem, since the majority of our mutants were defective in photoreception. However, the second question is to a limited extent addressed in this study, since we used a mutant defective in the transduction pathway that starts from the phytochromes and leads to the morphogenic effects.

Materials and methods

Plant material and growth conditions

Arabidopsis thaliana (Brassicaceae) is a weedy annual plant, originally from the Mediterranean basin, and now cosmopolitan. The plant produces a basal rosette of leaves. It then sends up the apical shoot, which yields the main inflorescence, with flowers opening acropetally and producing dry dehiscing fruits (siliques). Lateral inflorescences then develop basipetally (Hempel & Feldman, 1994) on the main axis. Furthermore, basal offshoots may develop in secondary stems in most ecotypes, if the environmental conditions are favourable.

The isogenic lines used for this work were kindly provided by the Arabidopsis Information Management System (AIMS: We utilized the following lines from the AIMS collection. The Landsberg erecta line represented the genetic background from which the mutants were originally isolated. This ‘wild type’ is actually a mutant, erecta, originally derived from the German natural population Landsberg. This mutation simply causes the production of an erect, as opposed to a more prostrate, stem. While this can obviously influence the effectiveness of shade avoidance plasticity, erecta is still the appropriate reference point for our experiment, since all mutants were originated from that genetic background. The hy1 (elongated hypocotyl) mutant is defective in the production of the chromophore, the component of the phytochrome molecule that is sensitive to light. Therefore, the hy1 mutant produces all five phytochrome apoproteins, but their activities are severely reduced because of a missing step in the biochemical pathway that produces the chromophores. The hy2 mutant is analogous to hy1 in that it produces all phytochromes but lacks chromophores; however, the mutation maps to a distinct genomic region and possibly affects a different biochemical step of the production of chromophores. The hy3 mutant (also known as phyB) is specifically deficient in phytochrome B, one of the major receptors involved in the shade avoidance response. The hy4 mutant is defective for the blue light receptor protein CRY1. The signal generated from the blue receptor, which is relatively insensitive to the red portion of the spectrum but responds to the blue wavelength, is integrated to some extent with the signals from the phytochromes. The signal is then fed into the downstream portion of the pathways leading to photomorphogenic responses, such as the DET and COP genes (Ang & Deng, 1994). The last mutant we considered was hy5, defective for a common transduction pathway element that supposedly links the blue-phytochrome feeding signals just mentioned to the det-cop system. However, this role of intermediate link for hy5 has been demonstrated only for the very early (seed to seedling) stages of the life cycle of A. thaliana. The mutants were originally described by Koornneef et al. (1980) and by Goto et al. (1993). Further references on all these mutants can be found at the AIMS web site, which is regularly updated.

We exposed the five mutant lines and the Landsberg erecta isogenic line to four experimental greenhouse treatments. High light, with sunlight reaching the plants through a transparent vinyl screen (R:FR = 1.05; light intensity at noon on a clear day, ≈250 μM/m2/sec); neutral shade, with neutral shade cloth layered on the vinyl, reducing irradiance to ≈50% of full sun, with R:FR identical to the control; low R:FR, in which plants were shaded with clear vinyl painted with dyes (Lee, 1985), which reduced irradiance to ≈40% of full sun and R:FR to 0.5; and grass canopy, in which A. thaliana was surrounded by perennial ryegrass (Lolium perenne) producing a semidense natural canopy, with irradiance of ≈70% of control and overhead R:FR at 0.8 at the beginning of the experiment (young canopy) and irradiance around 30% with R:FR at 0.4 at the end of the experiment (due to grass growth).

Plants were arranged in racks containing 10 rows of 20 circular cells each, in individual plastic pine cells 2.5 cm in diameter and 15.2 cm deep. This set-up avoided below ground competition among A. thaliana or with the grass. A. thaliana plants were grouped five per row (one per cone), equally spaced and randomly assigned to the available slots. The other cells were empty in all treatments except for the grass canopy, where they were occupied by 3–5 ryegrass seedlings, eventually thinned to one. The ryegrass was seeded 6 weeks before the experimental material, to ensure the presence of a canopy by the time of germination of A. thaliana seeds. The ryegrass was periodically trimmed during the experiment to reduce mechanical interference with the growth of the experimental plants. The racks were arranged in benches in a greenhouse, surrounded by metal grids supporting the screens. Aluminium foil was placed on the sides of the grids to increase internal reflectivity and minimize edge effects. The foil did not extend for the full height of the metal grids, in order to ensure air circulation. The set-up was replicated in 12 blocks, with each block containing five replicates of all lines, and with groups of three blocks assigned to each treatment (for a total of 15 replicates per line/treatment combination). The trays were randomly rotated between blocks at regular intervals (twice a week for the first month, once a week thereafter). Nevertheless, block effects are included in the ANOVA models (see below).

A. thaliana seeds were incubated for a week in the dark at 4 °C to stimulate and synchronize germination (Pigliucci et al., 1995a,b). The experiment was performed in the greenhouses of Brown University (Providence, Rhode Island, USA), and A. thaliana seeds were planted in mid-February. The soil mix was Metro Mix 350, containing processed bark to improve aeration and wettability. Widespread germination occurred 5 days after planting, and the experiment was continued until senescence of all reproductively active individuals, at the end of the spring. Plants were regularly bottom-watered (to avoid mechanical interference with growth) after seedling establishment, and were fertilized with Peters Professional Fertilizers Water Soluble (20–10–20 N-P-K), given at a rate of 50 p.p.m. every other day (Peat-Lite Formula). Three sets of low-output incandescent lights were used to extend the daylength to 16 h. Plants were harvested on an individual basis, approximately a week after the first siliques had opened on each individual (to allow time for seed ripening). This probably reduced the total reproductive fitness of the plants by limiting late maturation of flowers, especially on the late basal branches; on the other hand, we were in this way able to collect most of the seeds. There is no reason to believe that such a choice dramatically affected genotypic differences in estimated reproductive output (Zhang & Lechowicz, 1994).

The following characters were scored. (i) Hypocotyl length; this represents the aspect of the phenotype originally used to detect the mutants, as well as a very early character in the ontogeny of A. thaliana. (ii) Number of leaves at bolting; this character denotes both the vegetative output of the plant before the reproductive phase is initiated, and the minimum number of vegetative meristems necessary to start reproduction itself. (iii) Bolting time; this is a fundamental aspect of the phenology of A. thaliana, which marks the switch to the reproductive phase and distinguishes different ecotypes under natural conditions (Westerman & Lawrence, 1970; Jones, 1971). A few days after the bolting stem starts elongating, the first flowers appear. (iv) Number of basal offshoots; these are important auxiliary reproductive structures; the meristems from which they emerge are allocated toward the end of leaf production (before bolting: Hempel & Feldman, 1994), but they elongate and produce flowers only under some environmental circumstances, and only after the main inflorescence has started flowering. Since they were scored as meristems here (i.e. dormant buds and incipient branches were counted, not just elongated branches), they will be considered a measure of resource allocation. (v) Total number of fruits produced (by all inflorescences); this is an estimate of overall reproductive fitness. Its reliability in this respect is enhanced by the fact that this plant is a weedy annual that cannot store resources in vegetative structures for use in future seasons. Also, the number of seeds produced per fruit is roughly constant under a wide range of circumstances, and seed germination is usually very high (Westerman & Lawrence, 1970; Pigliucci et al., 1995a).

We would like to point out that our experimental design examined a set of discrete points along the R:FR continuum that would be experienced in nature. In principle it would be very interesting to examine responses to a range of R:FR values, in order to reconstruct the actual shape of the reaction norm (Gavrilets & Scheiner, 1993; Delpuech et al., 1995). The nature of this response may vary among species and depend on the ecology of the particular taxon being considered (Bradshaw & Hardwick, 1989). To gather such an extensive collection of data is, however, very cumbersome in plasticity experiments, and it is therefore rarely achieved. Our data can be used as biologically significant end-points to fine tune our understanding of plasticity to R:FR.

Statistical analyses

Statistical analyses were performed using Procedure GLM in SAS (SAS, 1990). Data were not transformed, since they displayed an approximately normal distribution and the scatter of the residuals did not show any discernible pattern. Notice that we are interested in both the slope and height of the reaction norms of these lines, and therefore no attempt was made to eliminate ‘scale effects’ statistically. Three-way analyses of variance appropriate for a split-plot design were carried out to estimate the magnitude and significance of the following effects: Line, evaluating genetic differences among the Landsberg erecta isogenic line and the mutants; Treatment, gauging the effect of neutral shade, low R:FR, or grass canopy compared with the high-light conditions, and therefore quantifying the plasticity of all lines; Line by Treatment interaction, measuring genetic variation among the lines for plasticity to the treatments used in the experiment; Block, nested within treatment, accounting for residual microenvironmental variation after rotation of the trays; and Line by Block interaction effect, accounting for any microenvironmental variation which could have resulted in line-specific effects. According to the procedure for a split-plot design, the Treatment main effect was tested against block, while the Line and Line by Treatment effects were tested against Line by Block. Type III mean squares are reported for each test, and a sequential Bonferroni correction has been applied to account for multiple simultaneous tests for the five traits (Rice, 1989). Data are presented as reaction norms of genotype by treatment means and their standard errors.

We also tested a series of specific a priori hypotheses comparing certain biologically relevant combinations of treatments and lines that were of particular interest for the general questions we were addressing. Among treatments, the comparison of high light with grass canopy contrasted conditions typical of an open site with those in which light is limited by neighbouring vegetation. The high light versus neutral shade comparison was an attempt to isolate the effect, if any, of pure reduction in photosynthetically active radiation, while the light quality was the same as normal sunlight. To further decouple the effects of light availability and spectral light quality, we compared neutral shade to the low R:FR treatment. This last condition was meant to simulate the grass canopy, without the mechanical interference and the changes in temperature and humidity caused by the ryegrass.

The above contrasts of treatments were combined with the following contrasts among lines. The wild type (Landsberg erecta) was compared to the combined chromophore mutants (hy1 and hy2), to ascertain the effect of the loss of all phytochromes on the plasticity to light availability of the wild type. We also contrasted Landsberg erecta with the phytochrome B-deficient mutant hy3, which is defective for the chief receptor controlling the shade avoidance. The comparison between the wild type and the blue receptor-defective mutant hy4 was meant to provide clues about the possible redundant (or complementary) role of this receptor in responding to ecologically meaningful changes in light availability. Finally, we contrasted Landsberg erecta with the transduction mutant hy5, under the hypothesis that a defect at the HY5 locus should compound the phenotypic effects of defects of the blue receptor and of the phytochromes.


Effects of photomorphogenic genes on plastic response to shade

Analysis of variance revealed highly significant line by treatment interaction effects for all traits except bolting time, as well as a treatment main effect indicating significant overall plasticity for hypocotyl length and fruit production, and significant differences among lines in their average response for all characters (Table 1). Plasticity for basal leaves and basal branches almost reached significance after the Bonferroni correction (and were both significant at the nominal α = 0.05 level). Block and block by line effects were not significant, but they were used as error terms for the other effects as specified above. Interaction effects were partitioned into a priori contrasts to investigate the impact of specific mutations on the response to vegetation shade (grass treatment), irradiance (neutral shade treatment), and R:FR (low R:FR treatment). None of the high light versus neutral shade contrasts was significant (Table 2; i.e. all mutants showed the same pattern of plasticity as Landsberg erecta; data not shown – original data and further plots are available upon request).

Table 1.  Two-way analyses of variance estimating genetic differences across mutants and Landsberg-wild type, phenotypic plasticity to different aspects of light availability, and genetic variation for plasticity. Type III MS reported. Bold typeface indicates significance of an effect after a sequential Bonferroni correction. Data were analysed according to a split-plot design. The Treatment effect was therefore tested over the block (treatment) term, while Line and Line by Treatment interaction were tested over line by block (treatment). Thumbnail image of
Table 2.  Contrasts derived from the analyses of variance testing specific a priori hypotheses about the plasticity of mutants when compared to the wild type under certain environmental circumstances. F-ratios are reported, P values in parentheses. Bold typeface indicates significant contrasts after a sequential Bonferroni correction. Thumbnail image of

Phytochrome mutants

Seedlings of chromophore-deficient mutants hy1 and hy2, as well as the phytochrome-B-deficient mutant hy3, displayed hypocotyl elongation relative to the wild type across treatments (Table 1; Fig. 1a). This result was expected, since it is precisely this phenotype that led to the original isolation of these mutants. Interaction contrasts of neutral shade versus low R:FR treatments revealed that the two chromophore mutants, hy1 and hy2, which are deficient in all functional phytochromes, were marginally significantly less plastic than the wild type in their response to R:FR (i.e. the difference was not significant after the Bonferroni correction: Table 2, Fig. 1a). This pattern was not significant for the phytochrome-B-deficient hy3 mutant (Table 2, Fig. 1a), possibly suggesting the involvement of another member of the phytochrome family in the plastic response to overhead R:FR.

Figure 1.

  Reaction norms across all four treatments of seedling and vegetative traits in Landsberg erecta and five mutants affected in light perception. (a) Hypocotyl length; (b) number of leaves; (c) time to bolting.

The chromophore-deficient mutants hy1 and hy2 also displayed constitutively shade-avoiding phenotypes for several traits expressed later in the life cycle. Compared with the wild type, these mutants flowered at a lower number of rosette leaves and produced fewer basal branches across treatments (Table 1; Figs 1b and 2a). The reaction norms of the two chromophore mutants were quite similar for all traits (Table 1; Figs 1 and 2). Interaction contrasts revealed that the mutants were significantly less plastic than the wild type in their response to R:FR for all traits measured after the seedling stage (Table 2; Figs 1b,c and 2): chromophore mutants were relatively nonplastic when the neutral shade and low R:FR treatments were compared, while the wild type displayed a marked increase in the number of basal leaves at flowering, later bolting and enhanced branching when in neutral shade compared with low R:FR (Table 2; Figs 1b,c and 2a). Interaction contrasts for the high light versus grass treatments revealed similar effects of all-phytochrome deficiency on response to vegetation shade for number of leaves, but not branch number or bolting (Table 2; Fig. 1b versus 2a and 1c).

Figure 2.

  Reaction norms across all four treatments of reproductive traits in Landsberg erecta and five mutants affected in light perception. (a) Number of basal branches; (b) number of fruits (a proxy for reproductive fitness).

The hy3 mutant, specifically deficient in phytochrome B, displayed a pattern of shade avoidance in response to both reduced R:FR and grass canopy shade similar to the chromophore mutants but the phenotype was less extreme. In fact, interaction contrasts detected no significant difference from the wild type in plasticity, except for a marginally significant (P = 0.04) reduction in plastic response of leaf number to grass shade. Although phytochrome B may contribute to plastic shade avoidance to an extent that our experiment lacked sufficient power to detect, the observed difference between the chromophore and phytochrome-B mutants suggests that phytochromes other than B are important in mediating shade avoidance traits expressed throughout the life cycle of A. thaliana.

Changes in irradiance (controlling for light quality) had little effect on the measured traits. None of the phytochrome mutants differed from the wild type in their plastic response to irradiance (Table 2).

Blue light receptor

At the seedling stage, the blue light receptor mutant, hy4, displayed an elongated hypocotyl, but did not differ from the wild type in the plastic response to R:FR, irradiance or grass shade (Table 2). However, later in development dramatic differences from both the wild type and the phytochrome-deficient mutants became apparent. Plants in which the blue light receptor was disabled were significantly more plastic than the wild type in the response of leaf and branch number to R:FR and grass shade (marginally in the case of number of leaves in the shade to low R:FR contrast: Table 2; Figs 1b and 2a). Under normal R:FR, in both the full light and neutral shade treatments, hy4 mutant plants produced more leaves before bolting and more basal branches than the wild type, but they converged toward the wild type phenotype in the low R:FR and grass canopy treatments (Figs 1b and 2a). This overall pattern contrasted with the reduced plasticity (relative to the wild type) of all phytochrome mutants. Thus, the blue light receptor appeared to act in opposition to the phytochromes in mediating the plastic response of post-seedling traits to R:FR and vegetation shade. The blue light receptor mutants did not differ from the wild type in plastic response to irradiance (high light versus neutral shade in Table 2).

Signal transduction mutant

hy5, the signal transduction mutant, displayed slightly longer hypocotyl, fewer leaves and fewer branches than the wild type, but did not differ significantly in plastic response to R:FR, irradiance or grass shade, except for a marginally significant (= 0.0351) reduction in plastic response of basal branches to R:FR (Table 2).

Effects on reproductive fitness

Reaction norms of fruit production for each genotype were very similar to the reaction norms observed for leaf number and branch number (compare Fig. 2b to Figs 1b and 2a). The low R:FR and grass canopy treatments resulted in a dramatically reduced fruit production relative to the neutral shade and high light treatments. However, the magnitude of this plastic response differed significantly among genotypes (Table 1). The chromophore mutants were significantly less plastic than the wild type in the response of fruit number to R:FR and grass canopy shade (Table 2; Fig. 2b). In contrast, plants deficient in the blue light receptor were more plastic than the wild type, displaying a significant increase in the number of fruits produced in neutral shade compared with low R:FR and a marginally significant increase in plastic response to grass canopy shade (= 0.0525: Table 2; Fig. 2b). The PHYB and signal transduction mutants did not differ significantly from the wild type in plasticity of reproductive fitness (Table 2).


This experiment demonstrated that mutants in which known photoreceptor genes were disabled displayed dramatic changes in plastic response to specific light cues for a variety of traits expressed throughout the life cycle. Furthermore, the effect of these genes was environment-specific: phenotypes tended to converge under low R:FR conditions, while large differences among genotypes emerged under R:FR characteristic of normal daylight. The suppression of plastic responses in the phytochrome-deficient mutants also enabled us to examine the fitness of shade avoidance phenotypes expressed under ‘inappropriate’ conditions (i.e. high light). Our results suggest that shade avoidance can incur a fitness cost for plants at low density (i.e. under high R:FR), which would not be experienced when the R:FR ratio is altered by the presence of a canopy. Although these mutants might conceivably have low fitness relative to the wild type owing to pleiotropic effects unrelated to the shade avoidance syndrome, the environmental specificity of most of our results supports an interpretation based on maladaptive lack of plasticity. If there is any cost intrinsic in these mutations, this does not manifest itself under low R:FR. Furthermore, it is clear from the data on the phytochrome-B and blue receptor mutants that mutations in photoreceptors can occur without any appreciable loss of reproductive fitness.

For the sake of the following discussion, it should be clarified that the term ‘shade avoidance’, originated from the physiological literature at a time when the molecular basis of these phenotypes was not known, is rather confusing. The shade avoidance phenotype is the developmental default (at least in A. thaliana), i.e. plants elongate unless they are exposed to normal sunlight. This implies that the genes that modify this behaviour do so when the R:FR is high (i.e. under ‘normal’ sunlight). Therefore, mutants are not expected to differ from the wild type under a canopy, for the simple reason that the genes that have been mutated do not have a biological effect under those conditions in the wild type. The real advantage or disadvantage of ‘shade avoidance’ is actually manifest in the ability to branch more and flower later under full sunlight. Accordingly, this should more properly be called a ‘light-exploitation’ response.

Pleiotropic effects of photomorphogenic genes

All mutations analysed in this study were originally described because of their common effect on hypocotyl elongation: the maintenance of the ‘default’ elongated hypocotyl under white light, which normally suppresses hypocotyl elongation in the wild type. Accordingly, in our experiment these mutants expressed an elongated hypocotyl relative to wild type in all light environments. However, the pattern of the hypocotyl elongation response to R:FR differed significantly from wild type in the hy1 and hy2 chromophore mutants. These mutants are deficient in all functional phytochromes and, accordingly, hypocotyl plasticity to low R:FR was totally suppressed. Both chromophore-deficient mutants also had strong pleiotropic effects on other aspects of the shade avoidance response throughout the life cycle. Plants lacking functional phytochromes displayed default shade avoidance phenotypes, such as bolting at an early developmental stage and suppressed branching, even under normal R:FR. A similar, but less extreme pattern of plasticity, not statistically different from the wild type, was observed for the phytochrome-B-deficient hy3 mutant plants. Thus, the shade avoidance to low R:FR we observed in A. thaliana appears to be at least partially mediated by a phytochrome other than phytochrome B. This result is consistent with previous observations of photomorphogenic mutants in a late flowering background demonstrating that phytochrome B and at least one other phytochrome mediate early flowering in response to low R:FR in A. thaliana (Halliday et al., 1994).

The pattern of pleiotropy for the blue light receptor mutant, hy4, was dramatically different from that of the phytochrome mutants. At the seedling stage, the effect of this mutation on hypocotyl elongation was similar to the effects of the phytochrome-deficient mutants. After the seedling stage, however, the blue receptor mutation exercised an effect opposite to that of the phytochrome-deficient mutants, increasing the degree of plastic response to high light versus grass shade and/or to neutral shade versus low R:FR of leaf number at flowering, basal branch number and fruit production. This observation suggests that in wild type plants, the HY4 gene interacts synergistically with light-stable phytochromes early in the life cycle, and antagonistically thereafter. A possible ecological interpretation of this unexpected result is in order. We could think of the blue receptor as a basic sensor for currently perceived photosynthetic light, in contrast with future levels of light to which the phytochrome system seems designed to respond. According to this scheme, the ecological function of the blue light receptor is complementary to that of the phytochromes. For example, while the phytochromes would push the plant to take advantage of the absence of canopy shade, the blue receptor may perceive high levels of sunlight as indicators of another threat (e.g. ensuing drought, which commonly occurs in natural populations of Arabidopsis, M. Pigliucci and H. Callahan, personal observation). The observed antagonistic behaviour of the two mutants would then be the result of the conditions of the experiment. High light in the greenhouse was not accompanied by drought (because of regular watering); therefore, the wild type showed a partially maladaptive response in flowering earlier than the blue receptor mutant. In the latter genotype the ‘high light–drought’ cue is ignored, with beneficial results given the absence of drought. Obviously, this would not be the case under natural conditions. A follow-up experiment designed to test this particular ecological hypothesis has provided data consistent with our interpretation (C. Wells et al., in press).

Mohr (1994) has suggested a model for coaction of photoreceptors in which the FR-absorbing form of phytochrome (Pfr) triggers photomorphogenesis, whilst the blue light receptor modulates the degree of responsiveness. In seedlings, blue light has been observed to amplify responsiveness to FR (Mohr, 1994; Casal & Boccalandro, 1995). Impaired hypocotyl response to low R:FR has been observed in the hy4 mutant in other studies (Casal & Boccalandro, 1995; Yanovsky et al., 1995), although it could not be detected under the conditions of our experiment. In contrast with the model for seedlings, our results suggest that later in development, the HY4 gene product may diminish plasticity to high light versus low R:FR: under low R:FR, the default shade avoidance phenotype was maintained in the hy4 mutant; however, under R:FR characteristic of sunlight, hy4 plants flowered with a larger number of rosette leaves and produced more branches than the wild type, suggesting that this gene normally suppresses this plastic response. This finding is consistent with previous observations of later flowering and increased leaf number in the hy4 mutant in response to changes in photoperiod, which led to the suggestion that the blue light receptor promotes flowering, interacting antagonistically with phytochrome B (Mozley & Thomas, 1995).

Our results also indicate that the wild type blue photoreceptor reduces the production of basal offshoots, whereas the phytochromes increase the number of basal offshoots produced by a plant exposed to full sunlight. The pleiotropic plastic responses on leaf number and basal branch production are probably attributable to their tight developmental linkage (Hempel & Feldman, 1994), as has been suggested for other plant species (van Tienderen, 1990; Schmitt, 1993, 1995). Such correlated plasticity of developmentally and genetically linked traits may be the underlying basis of what has been termed phenotypic integration (Schlichting, 1989; Holloway & Brakefield, 1995), i.e. the correlation among the plasticities of different characters.

Environment-specific effects of photomorphogenic genes

Mutations at specific photomorphogenic loci had large effects on morphology and development in the high light and neutral shade treatments, and much smaller effects in the low R:FR and grass canopy treatments. Under low R:FR, all genotypes tended to converge on a shade avoidance phenotype, bolting at an early developmental stage (number of leaves) and producing few branches. However, under R:FR characteristic of natural sunlight, the phytochrome-deficient mutants retained the shade avoidance phenotype, in contrast to plastic wild type plants, which bolted at a later developmental stage and produced more branches. This finding is consistent with the suggestion that shade avoidance is the default developmental state, and that phytochrome-mediated photomorphogenic pathways are activated by red light in natural sunlight (Chory, 1993). The apparently antagonistic action of the blue light receptor discussed above was also evident only under high R:FR. Thus, the phenotypic impact of this gene depended upon the presence of high levels of light. These results suggest that in natural populations, the genetic variation at the loci regulated by both phytochromes and blue receptor is more likely to be expressed, and thus be available for response to selection, in high R:FR/high irradiance environments.

Effects of photomorphogenic genes on fitness

The hypothesis that shade avoidance responses are adaptive leads to two critical predictions. (1) Shade avoidance phenotypes must have higher relative fitness in dense vegetation. (2) Shade avoidance phenotypes must be disadvantageous in isolated plants (Casal & Smith, 1989; Schmitt & Wulff, 1993). Comparing the fitness of mutants deficient in specific photomorphogenic responses with that of the plastic wild type across a range of ecologically relevant environments provides an important tool for testing both of these predictions (Schmitt et al., 1995). In our experiment, the contrasts of the high light and grass canopy treatments or the contrasts of the neutral shade and low R:FR treatments provided a test of the prediction that expression of shade avoidance traits is costly in the absence of competition. The fitness of the constitutively shade-avoiding phytochrome-deficient mutants when compared to the plastic wild type was lower in the high light than in the grass canopy treatments, and was also lower in the neutral shade than in the low R:FR treatments. This result indicates that inappropriate expression of shade avoidance traits in isolated plants is maladaptive in A. thaliana.

Similarly, constitutively elongated mutants have lower fitness relative to nonelongated wild type at low density, but maintain fitness comparable to the wild type at high density in Brassica rapa (Schmitt et al., 1995). Moreover, constitutively elongated mutants of Cucumis sativus suffer greater mechanical damage and mortality than the wild type under low-density conditions (Casal et al., 1994). Our experiment could not test the prediction that shade avoidance traits are advantageous under competitive conditions in A. thaliana, since all the mutants converged toward a shade-avoidance phenotype in the grass canopy treatment. However, it may be possible to obtain the range of phenotypes necessary to test this prediction by using transgenic A. thaliana plants in which the shade avoidance response is suppressed by over-expression of phytochrome A (Whitelam & Harberd, 1994). A similar experimental approach in tobacco provided strong support for the adaptive plasticity hypothesis (Schmitt et al., 1995). Thus, plants genetically modified at specific photomorphogenic loci not only afford insight into mechanisms of gene function, but are also valuable tools for testing evolutionary and ecological genetics hypotheses. Furthermore, the current state of the accumulated knowledge on systems in which shade avoidance has been studied suggests some obvious strategies in the genetic engineering of more efficient varieties of crop plants (Linder & Schmitt, 1995; Purrington & Bergelson, 1995; Rogers & Parkes, 1995; Weigel & Nilsson, 1995). Marked progress in this area is certain to be forthcoming.


We thank Fred Jackson for greenhouse assistance, Susan Dudley for discussions about this research, and Albrecht von Arnim, Lisa Dorn, George Gilchrist and Ariel Novoplansky for comments on the manuscript. This work was supported by NSF grants DEB-9527551 to M.P. and GER9023389 and DEB9509239 to J.S.