Differential genetic variation in adaptive strategies to a common environmental signal in Arabidopsis accessions: phytochrome-mediated shade avoidance


Professor Harry Smith, Division of Plant Science, University of Nottingham, Sutton Bonington Campus, Loughborough, LE12 5 RD, UK. Fax: + 44 (0) 150 9856822; E-mail: harry.smith@nottingham.ac.uk


Shade avoidance is a syndrome of plastic responses to light signals encountered in crowded plant communities and is a crucial component of competitive strategy in higher plants. The responses are mediated via signal perception by specific members of the phytochrome family of photoreceptors, which detect the relative proportions of red (R) and far-red (FR) radiation within dense communities. We analysed two aspects of shade avoidance, the acceleration of flowering and the enhancement of elongation growth, displayed by more than 100 accessions of Arabidopsis thaliana (Heyn.) in response to FR-proximity signals. Both traits showed wide variation between accessions, which was unrelated to the latitude of the location of original collection. Flowering acceleration is a major feature of shade avoidance in rosette plants such as Arabidopsis, and most accessions showed dramatic responses, but several were identified as being recalcitrant to the proximity signal. These accessions are likely to be informative in the analysis of quantitative variation in shade avoidance. Hypocotyl elongation, treated here as an indicator of elongation growth responses, also varied widely amongst accessions. The variations in flowering acceleration and elongation were not correlated, indicating that microevolution in the downstream pathways from signal perception has occurred separately.


the total numbers of days before bolting


far-red radiation (700–800 nm)


the total number of rosette leaves produced before bolting


total phytochrome


far-red-light absorbing form of phytochrome


phytochrome photoequilibrium


red-light absorbing form of phytochrome


red radiation (600–700 nm)


ratio of the photon irradiance between 655 and 665 nm to that between 725 and 735 nm


Light signals perceived by the phytochrome family of photoreceptors are crucial to the appropriate acclimation of plants to their natural environment (Smith 1982, 1995). Such signals have been implicated in the regulation of germination, seedling emergence and establishment, the architecture of the vegetative plant and the induction and rate of flowering (Smith 2000). Recent years have seen major advances in the study of the transduction pathways of responses to light signals, and these advances have come mainly through the application of mutation selection techniques (Hudson 2000). However, the downstream transduction pathways that lead to functional plasticity in response to light signals in the natural environment remain a closed book. In the post-genomic era, microarray studies are beginning to reveal that the expression of hundreds of genes are regulated in phytochrome signal transduction (Tepperman et al. 2001), but it remains difficult to allocate these genes to specific ecophysiological responses. In this study we have used the natural genetic variation that exists between accessions of Arabidopsis thaliana as an alternative approach to begin the elucidation of downstream processes and the eventual identification and characterization of ecologically important genes.

The shade avoidance syndrome is one of the best-studied phenomena of adaptive phenotypic plasticity in plants (Smith 1982, 1995). Shade avoidance is a suite of developmental responses evoked by the proximity of neighbouring plants. Selective absorption by the photosynthetic pigments causes radiation reflected/scattered by plant leaves to be relatively enhanced in the far-red waveband (FR, c. 700–800 nm), providing a unique and unambiguous signal of competitive threat. The intensity of the FR signal depends on neighbour proximity (Ballaréet al. 1990) and population density (Gilbert et al. 1995) and appears to be a function of the total area of reflective leaf material within the perceptive range of the plant (Gilbert et al. 2001). These signals are detected by the phytochrome family of signal transducing photoreceptors, and evoke appropriate competitive or survival reactions, including rapid elevations in elongation growth, reduction in branching and acceleration of flowering. Physiological investigations using mutants that lack individual members of the phytochrome family have demonstrated that all the identifiable components of the shade avoidance syndrome are mediated principally by phytochrome B (phyB), with redundant action also by phyD and phyE (Smith & Whitelam 1997). Thus, a single environmental signal, perceived by closely related members of a photoreceptor family, evokes branched pathways leading to different elements of adaptive response, a classic single-input multiple-output system (Smith 2000).

Plant species are known to differ in the magnitude of their responses to proximity signals, and in a limited sense this has been shown to be a function of the ecological habitat in which the species grow. For example, herbaceous species that tolerate woodland shade generally respond much less than others that are aggressive competitors in more open environments (Morgan & Smith 1979). The sensitivity of early successional tree species to FR signals in crowded environments appears to be considerably greater than late-successional trees, providing an explanation for the ecological relationships between species in developing forest canopies (Gilbert et al. 2001). There is strong evidence that, within a single species, populations derived from open or shaded habitats have different phenotypic responses to the FR signal. Bain & Attridge (1988) showed that field and hedgerow populations of Gallium aparine exhibited different responses to FR signals. In recent years major efforts have concentrated on approaches using transplantation. Extensive investigations into the population biology of Impatiens capensis by Johanna Schmitt and colleagues (e.g. Dudley & Schmitt 1995, 1996; Donohue & Schmitt 1999; Donohue et al. 2000a, b, 2001; Schmitt et al. 1999) have elegantly and convincingly demonstrated the crucial role of phytochrome-mediated shade avoidance in the natural environment. These conclusions are confirmed by related studies using Abutilon theophrasti (Weinig 2000a, b). A small number of contrary observations have been reported; for example, van Hinsberg & van Tienderen (1997) showed that shade populations of Plantago lanceolata were more responsive to a FR signal than were sun populations. Even in this case, however, phenotypic differences were observed between the two populations, although they appeared to be opposite in direction to those more commonly reported.

Direct evidence for the adaptive nature of plasticity in shade avoidance has been adduced, both through transgenic and mutational disablement of the phytochrome perception system (Schmitt et al. 1995; Ballaré & Scopel 1997; Pigliucci & Schmitt 1999). However, a fuller understanding will require detailed knowledge of the multiple-branched transduction pathways that emanate from the photoreceptors. Plasticity observed as variation in physiological output must presumably result from the selective expression of genetic variation within these transduction pathways. This means that as an individual plant responds to fluctuating environmental light signals, there must be selection within the branched transduction pathways emanating from the initial signal perception event. Furthermore, in ecotypes, varieties, populations or individuals that show varying capacities for phenotypic acclimation, there must on this basis be differential flux of information along the downstream pathways specific for individual plastic responses. Presumably therefore these variations in physiological output have a genotypic basis. The current challenge is to identify the varying components within downstream transduction pathways that are responsible for phenotypic plasticity (Schlichting & Pigliucci 1993, 1995). Once such genes are known it should be possible to establish the molecular basis of plastic variations and to approach an understanding of the mechanisms that maintain variation for quantitative traits in natural populations.

Clearly, the downstream events leading to enhanced extension growth in response to a FR-proximity signal must be different from those leading to accelerated flowering in response to the same signal. In rosette plants such as Arabidopsis thaliana, an acceleration of flowering is the most dramatic response to FR-proximity signals (Halliday, Koornneef & Whitelam 1994; Bagnall et al. 1995). Floral acceleration has obvious fitness implications, increasing the probability of survival to reproduction under conditions of extreme competition. For such rosette plants the enhancement of elongation in response to a proximity signal would on the face of it appear to have little ecological benefit, but such elongation enhancement nevertheless does occur. It can be seen in the elongation of leaves and petioles, which in some cases can be very marked. Arabidopsis is a ‘model’ species, but by selecting accessions (ecotypes) that were originally collected from a wide range of different locations we have been able to sample the variation that occurs in collections of natural populations. We examined the responses of more than 100 accessions to FR-proximity signals provided in controlled environment cabinets.

Materials and method

Plant materials and growth conditions

Seed of Arabidopsis thaliana was obtained from the Nottingham Arabidopsis Stock Centre (NASC). Table S1 provides a full list of the accessions used; note that the Stock Centres refer to each accession as an ‘ecotype’, but we prefer the term ‘accession’ as it carries no functional implications. We began with more than 300 accessions, but incomplete or non-synchronous germination and poor long-term growth of many forced us to concentrate on a somewhat smaller group. We were able to obtain data from 157 accessions for flowering time, but the more stringent requirement for synchronous germination in the hypocotyl studies reduced that list to 105.

For all studies, seed was sown on 2% agar containing half-strength Murashige-Skoog salts. The dishes were maintained at 4 °C for 4–5 d, after which they were exposed to white light (WL) for a period of c. 1 h to stimulate and synchronize germination. For the floral acceleration experiments, seedlings were grown in a WL cabinet on short days (SD = 8 h light, 16 h dark; 21  ±  1 °C) for 10 d, before transfer to the spectral quality cabinets (see below). Before the beginning of the light treatments, seedlings were transferred to a 50/50 mixture of potting compost and sand. For measurements of hypocotyl extension, seedlings remained on the agar plates throughout. After exposure to WL to stimulate germination, the plates were held at 20 °C in the dark for 24 h before being transferred to the spectral quality cabinets.

Flowering time was measured both as days to bolting (i.e. days after transfer to the experimental conditions) or total rosette leaves to flowering. Between three and nine plants were evaluated per accession. In all cases, the standard errors of the calculated means were less than 10% of the mean. Hypocotyl length was measured after 5 d in the experimental conditions by digital photography followed by computer-assisted scanning. A minimum of 10 hypocotyls were measured for each accession and condition, and the data presented are the means of these measurements. Error bars are not presented because the graphs then become too complicated, but accessions for which the standard errors were greater than 5% of the mean were discarded, and the experiment repeated.

Spectral quality cabinets

In order to evoke pronounced shade avoidance responses we used controlled environment growth cabinets in which the spectral distribution of the incident radiation could be manipulated. These cabinets have been used for many years at Leicester to generate R:FR ratios that establish widely differing proportions of the inactive (Pr) and active (Pfr) forms of phytochrome. The spectral distributions of the radiation in these cabinets are shown in Fig. 1. Using a mixture of white fluorescent tubes and incandescent lamps filtered through FR-transmitting plastic, we were able to provide plants with identical amounts of photosynthetically active radiation (PAR = 400–700 nm) but different amounts of FR (700–800 nm). We supplied sufficient supplementary FR to reduce the phytochrome photoequilibrium (Pfr/P) from the c. 80% in the WL, to 50% in the intermediate and 35% in the strong FR-signal environment. In the figures, the radiation environments are denoted by the R:FR ratio (i.e. [ΣN655−665nm]/[ΣN725−735nm] where N is spectral photon fluence rate). The values of R:FR ratio for the three environments were: WL, 7·2; intermediate FR-signal, 0·32; strong FR-signal, 0·20. Although these spectra do not simulate the natural environment in open or shaded communities, they allow us to distinguish between purely phytochrome-mediated processes and those caused by differences in the amounts of photosynthetic radiation. Seedlings and plants were grown in these cabinets using 24 h illumination at 21  ±  1 °C.

Figure 1.

Spectral photon distributions of radiation provided in controlled environment growth cabinets.


Shade avoidance in Arabidopsis

The most dramatic shade avoidance response in Arabidopsis is the acceleration of flowering. In many cases this response is spectacular, especially when using very low R:FR. As variation in floral acceleration was clearly evident we chose to use this parameter in the accession survey.

The most commonly observed, and usually the most striking, shade avoidance response is enhancement of the elongation of internodes and petioles. In a rosette plant such as Arabidopsis, however, internodes do not normally become apparent until the bolting of the floral (or cauline) stem, which occurs at the onset of flowering. In the presence of a FR proximity signal, enhancement of cauline stem elongation can be observed, but it is usually a relatively small response. Petiole elongation is often enhanced in low R:FR environments, but from a practical viewpoint this response can be difficult to quantify, especially when making comparisons between accessions of differing morphology. Indeed, in some accessions, it is difficult to establish precisely where the petiole ends and the lamina begins. In some cases, enhanced petiole extension is accompanied by reduced lamina extension, and consequently it is not advisable to rely on measurements of total leaf length. A strong and readily quantifiable elongation response can be observed at the seedling stage, when hypocotyl elongation is occurring. As there was clearly considerable variation in the hypocotyl shade avoidance response, we chose to use this as a parameter in the accession survey.

Accession survey reveals wide variation in shade avoidance responses

Floral acceleration

We measured days to bolting and the number of rosette leaves produced before bolting in 157 accessions exposed either to long-day conditions (24 h continuous light) with a high R:FR ratio, or with two levels of additional FR radiation providing intermediate and strong proximity signals (low R:FR ratio) (see Fig. 1 for radiation spectra). Initially, 300 accessions were included, but incomplete data forced us to reduce the number used in the comparisons to the core of 157, for which statistical validity could be justified. The flowering time data for all the accessions tested, together with the names of the accessions and the geographical location and latitude/longitude of their original collection point, are collected in Table S1. The names and location data were obtained from the NASC website.

It is known that the physiological time to flowering, as measured by the number of rosette leaves produced before bolting, and the actual time to flowering, measured in days, tend to be related in a simple linear manner. Figure 2 shows the data for LTF (leaves to flowering) and DTF (days to flowering) plotted against each other for each of the three radiation environments. In all cases the data could be plotted onto single regression lines, although there was more scatter in the WL environment, and the r2 values were only significant for the two WL + FR environments.

Figure 2.

Shade avoidance effects on the relationship between the number of leaves to flowering and the number of days to flowering.

Figure 3 shows the distribution of the two flowering time parameters, DTF and LTF, among the accessions surveyed ranked according to the data for the WL, high R:FR environment. For the majority of accessions, whether they flower early or late under continuous WL, transfer to low R:FR conditions accelerated flowering markedly. This was especially obvious for the LTF parameter (Fig. 3a–c), but is also quite clear for the DTF parameter (Fig. 3c–d). However, in both cases, a number of individual accessions were less accelerated by the FR signal, and some of them were hardly accelerated at all. These data indicate that there is considerable genetic variation in the shade avoidance floral acceleration response. The variation is more noticeable under the intermediate FR-signal conditions (i.e. R:FR = 0·32) than under the strong FR-signal conditions (R:FR = 0·2).

Figure 3.

Figure 3.

Shade avoidance acceleration of flowering in 157 accessions of Arabidopsis. (a) & (b) leaves to flowering; (c) & (d) days to flowering.

Figure 3.

Figure 3.

Shade avoidance acceleration of flowering in 157 accessions of Arabidopsis. (a) & (b) leaves to flowering; (c) & (d) days to flowering.

Frequency plots of the flowering data emphasize the magnitude of the shade avoidance acceleration of flowering. In Fig. 4 we show the LTF data for all three radiation environments. In WL (i.e. zero FR-signal) the median number of leaves produced before flowering was c. 40, whereas under the WL + FR environments this value was reduced markedly to c. 14 with an intermediate signal, and to c. 10 with the strong FR-proximity signal. In the latter two cases, however, there was a marked tail towards higher numbers of leaves to flower. This tail comprises those accessions that have limited sensitivity to the FR-proximity signals.

Figure 4.

Frequency distributions of data for leaves to flowering in three radiation environments differing in R:FR ratio.

To identify those accessions that were least responsive to the FR-signal, we converted the flowering time data to an acceleration index. Floral acceleration index (FAI) was calculated as follows:


By plotting FLAR:FR = 0·32 against FLAR:FR = 0·2 we generated Fig. 5. A small group of accessions clustered around the co-ordinates X = 1·0–1·6:Y = 1·0–1·6 on this plot, indicating there was little acceleration of flowering in these accessions at either the intermediate or strong FR signals. These accessions are identified in the inset of Fig. 5. At the other extreme were accessions with very high values of FLA under both environments, indicating that those accessions were very sensitive to the signal.

Figure 5.

Accession floral acceleration responses to the far-red proximity signal. This figure plots the floral acceleration indices for two levels of far-red proximity signal against each other, allowing the identification of accessions with extreme responses. The inset enlarges the portion nearest the origin of the graph and shows eight accessions with minimal floral acceleration responses to the proximity signal.

There was no discernible latitudinal relationship between the sensitivity of the accessions to the shade avoidance signal. Figure 6 shows the LTF data plotted against the latitude of origin of the accession. To highlight this lack of latitudinal relationship Fig. 7 shows a selected group of accessions, originally collected from the same locality in Germany, that have markedly different flowering time characteristics. The 14 accessions shown here required from 24 to 73 leaves to flower in the continuous WL environment, yet all of them were markedly accelerated by the strong FR-proximity signal, flowering at between 10 and 13 leaves. By contrast, of the eight accessions identified in Fig. 6 as being least sensitive to the proximity signals, five were from the Iberian Peninsula (Fig. 8).

Figure 6.

Latitudinal distribution of floral acceleration responses.

Figure 7.

A selection of accessions from a single location in Germany exhibiting a wide range of floral acceleration responses to the far-red proximity signal.

Figure 8.

Floral acceleration responses of common laboratory lines and of the accessions selected as being most recalcitrant to the far-red signal.

Hypocotyl elongation

To assess natural variation in extension growth responses to the FR-proximity signal we attempted to measure hypocotyl extension in the set of accessions for which flowering time had been measured. Unfortunately, a significant proportion of these accessions failed to germinate entirely synchronously, even with a WL pre-treatment. This left a smaller subset of 105 for which reliable data on hypocotyl elongation could be acquired. The actual data for the 105 accessions are given in Table S1 and Fig. 9. Figure 9a shows the data ranked according to the hypocotyl length in WL, and Fig. 9b presents the frequency distributions of hypocotyl length in the two environments.

Figure 9.

Hypocotyl elongation responses of 105 Arabidopsis accessions to the far-red proximity signal. (a) Data ranked according to hypocotyl length in the absence of the far-red signal. (b) Frequency distribution of hypocotyl length data.

Variation in the two shade avoidance responses is independent

Having established the extent of variation in the two shade avoidance responses, we wished to determine whether there was any relationship between the distributions. If accessions that were resistant to the signal for the floral acceleration response were similarly resistant for the hypocotyl elongation response, then we could assume that common, presumably early, steps in the transduction pathways were exerting control. We consequently calculated a hypocotyl elongation index (HEI) as:


where HL is hypocotyl length. Although numerator and denominator are reversed in this index compared to the floral acceleration index, it generates a positive value, which for practical purposes is preferred.

Interestingly, the frequency distributions of the two indices are almost identical (Fig. 10). The bulk of the accessions showed what appeared to be a normal distribution around a median of c. 4, but there were outliers with both very strong and very weak responses. We then plotted FAI against HEI, for all accessions for which both data were available, and this is presented in Fig. 11. There was no relationship between the distributions of the two indices across the range of more than 100 accessions analysed, although accessions that were highly responsive in both responses were lacking (i.e. no accessions having a FAI > 5 and a HEI > 5).

Figure 10.

Similarity of frequency distributions of floral acceleration and hypocotyl elongation responses to the far-red signal.

Figure 11.

Lack of relationship between the floral acceleration and hypocotyl elongation responses of Arabidopsis accessions to the far-red signal.


These data present an overview of the variation in response magnitude, and presumably therefore of signal perception sensitivity, to FR proximity signals in Arabidopsis. The populations used for these experiments were obtained from the Arabidopsis Stock Centres and were not all single-seed descent stock. Therefore, we cannot be certain that each of the accessions is genotypically homogeneous. Furthermore, the accessions were originally collected many years ago, and some have been maintained in collections in diverse locations before being donated to the Stock Centre. It is thus not correct to describe them as natural populations, but nevertheless the observation of marked variation between accessions must indicate at least the possibility that populations adapted to certain environments or localities have evolved differential capacities to respond phenotypically to the environmental signals. If this tenet is accepted then the data demonstrate very wide phenotypic variation in two principal shade avoidance responses between genotypically different populations of a single species.

This species is commonly found in open and disturbed ground, its trivial names in the UK being not only Thale Cress (named after the German physicist Johann Thal, 1542–83) but also Common Wall Cress, or Rock Cress. Whether shade avoidance responses are ecologically important for such plants is a moot point, although a strong argument can be made that acceleration of flowering when plants become shaded is of significant advantage and may enhance fitness by ensuring that reproduction occurs before the plant succumbs to low light levels. The ecological significance of phyB-mediated hypocotyl elongation responses to FR proximity signals is as yet unknown, but the absence of phyA, in the phyA mutant, leads to maladaptive and ultimately fatal elongation in seedlings exposed to canopy radiation (Yanovsky et al. 1995). Thus, the photoregulation of hypocotyl elongation is clearly a very important strategy in Arabidopsis.

The variation in flowering response to the FR signal shown here is considerable (Fig. 3), with the acceleration index (FAI) ranging from near unity (no response) to more than 8 (massive response) (Fig. 5). Obviously, this index is to some extent skewed by the triviality that accessions which flower quickly in WL cannot be accelerated as much as those which flower slowly in WL. Another environmental signal that can cause marked acceleration of flowering is vernalization − exposure to a sustained period of low temperature. Johanson et al. (2000) demonstrated that natural variation in response to vernalization is related to the activity of the FRIGIDA (FRI) gene, which operates in consort with other flowering time genes, particularly FLC (Lee et al. 1994). Although we have not examined the accessions studied here for their responses to vernalization, we have shown that FR-treatment will accelerate flowering whether or not active FRI or FLC genes are present (H. Smith, S. D. Michaels and R. Amasino, unpublished results). This may indicate that the vernalization and FR-proximity signals accelerate flowering via separate pathways. Nevertheless, because a large value of FAI depends on slow flowering in WL, whether or not a particular accessions contains an active FRIGIDA gene must be a complicating factor in these experiments. It will be of interest to examine the accessions surveyed here for their responses to a vernalization treatment.

An important point from these data is that the extent of variation in the flowering acceleration response depends on the intensity of the FR proximity signal. In Fig. 3 it is clear that the intermediate signal (R:FR = 0·32) reveals more variation that does the strong signal (R:FR = 0·2). This is in accordance with expectation for a response that is graded with the intensity of the signal, a feature of shade avoidance that has been amply demonstrated in previous research. For example, for all herbaceous plants investigated in the spectral ratio cabinets, we have observed inverse linear relationships between elongation growth rates and the Pfr/P calculated to be established by the incident radiation (see Smith 1982 for review). Even in trees growing in the natural environment, their elongation growth rates have been shown to be precise functions of the intensity of the FR-proximity signal generated within the stands (Gilbert et al. 1995, 2001). Such detailed quantitative relationships have not yet been established for the floral acceleration response, but the data shown here are consistent with a graded response to the signal intensity. Thus, with a strong signal the inherent variation seems to be suppressed, but this is merely a consequence of response saturation.

The ecological basis of the variation remains obscure, however. We failed to find any relationship between response magnitude and latitude of origin of the accessions. Johanson et al. (2000) claimed a latitudinal relationship for vernalization responses, with a north/south distribution significantly different from random. Our data (Fig. 6) show no evidence for such a distribution for the flowering acceleration response to FR proximity signals. Indeed, as shown in Fig. 7, even accessions originally acquired from very close localities can exhibit marked differences in flowering time in continuous WL, and yet all be accelerated to very rapid flowering with the FR signal. These data do not definitely demonstrate variation in floral acceleration within the same locality but do show that time to flowering can vary substantially within the same location. The ecological significance of this is unclear, but we must assume that the response to the proximity signal is of significance even when flowering time itself varies substantially. In a study of nucleotide polymorphism within and among populations of A. thaliana, Bergelson et al. (1998) found no association between geographical and genetic distance between populations. As pointed out by Sharbel et al. (2000), the phylogeography of Arabidopsis has been markedly affected by human action.

Identification of accessions with very weak responses to proximity signals was the principal objective of this set of experiments. Such accessions presumably carry genes whose action, or perhaps inaction, prevents the more usual acceleration of the flowering process. Quantitative genetic analysis of these accessions, involving crosses with mapped accessions, will provide clues to the chromosomal location of the responsible genes and if such data can be linked to surveys of gene expression then identification and characterization of ecologically important genes will be feasible. A number of interesting accessions are highlighted by the data provided here (Fig. 5). The small group of accessions clustered around unity for the two FAI values (see inset in Fig. 5) represent those with very weak responses – these accessions may be regarded as recalcitrant to the environmental signal. Of these eight accessions, five originated from the Iberian Peninsula (Bla-1, Bla-6, Ll-1, Sf-2 from Spain; Co-4 from Portugal), whereas the remaining three are from other parts of Europe (Ge-2, Switzerland; Di-1, France; Lu-1, Sweden). There therefore seems to be no geographical reason for the clumping of these accessions. Furthermore, although these accessions clump near the axis origins of Fig. 5, there is more or less a continuum of co-ordinates up to the extreme responders at FAI values of c. 8. The most extreme responders in this survey were Wil-2 from Russia, Bu-9 from Germany and Pa-2 from Italy. It is worth pointing out that for the eight accessions highlighted here as weak responders, two (Ll-2 and Di-1) flower very quickly in WL and thus may not be true candidates for weak responders (Fig. 8). Thus, as candidates for future quantitative genetic analysis, Bla-1, Bla-6, Co-4, Ge-2, Lu-1 and Sf-2 should provide the most useful data.

The variation in hypocotyl response to the proximity signal was equally impressive (Fig. 9). The frequency distribution data (Fig. 9b) show that this response is generally quite strong, and the elongation index (HEI) shows variation from values close to unity up to large enhancements of elongation with values of c. 6 (Fig. 11). Hypocotyl elongation is not as straightforward to analyse as is flowering time. Seedling development may vary between accessions such that the overall dynamic of hypocotyl elongation may be different. For example, we had problems in synchronizing germination in many of the accessions, resulting in our having to reduce the number of accessions from which we could reliably obtain data. If the time-course of hypocotyl elongation may also be different between accessions then some of the observed variation in response could be simply a manifestation of the different time-course. However, determining developmental time-courses for such a large number of accessions would be impractical, so we are forced to rely on the data we have, bearing the caveat in mind in interpretation.

Comparing the frequency distributions of the flowering acceleration and hypocotyl elongation indices produced a somewhat surprising similarity (Fig. 10). The distributions were analysed by the two-tailed two sample Kolmogorov–Smirnov test [n1n2D= 173·66 (P < 0·10)], and we concluded that the two samples did not differ significantly in their distributions. The lack of correlation between the two indices in (b) was confirmed using the Spearman coefficient [rs; as N > 50, the statistic z was calculated, where z = rs(N − 1)1/2 = 0·178 (P = 0·4286)] (Sokal & Rolf 1995). The cause of the similarity between the two distributions is unknown, and may simply be chance. It is clear, however, that the outliers are not the same accessions for the two responses.

An important conclusion from the comparisons of the two shade avoidance indices is that the variations in the responses are not correlated (Fig. 11). For example, accessions that respond weakly in flowering acceleration do not necessarily respond weakly in elongation. There is a suggestion of a trade-off between the two responses, however, as no accessions seem to have large values of the two indices – in other words, there are no accessions with FAI above 5 and HEI also above 5. The lack of correlation between two presumably adaptive responses to an identical environmental signal is unexpected, as the general view appears to be that plastic responses of traits are likely to be correlated (Schlichting 1986). For example, a close relationship has been found between flowering time and flowering size (Mitchell-Olds 1996). Furthermore, Zhang & Lechowicz (1994) found a strong correlation between flowering time and overall plasticity of 13 populations of Arabidopsis. There is no indication in our data that plasticity to FR signals is related to time of flowering.

The principal reason for beginning this survey of responses to proximity signals was to select accessions that show extreme responses in order to begin a genetic analysis designed to identify loci that are important in shade avoidance. Our data confirm that genetic information relevant to shade avoidance responses that is available in populations of Arabidopsis is huge with respect to the lines most frequently used in the laboratory. The information obtained by accession comparison and by analyzing segregating populations will allow us to locate and isolate genes modulating the shade avoidance responses and also could reveal important information about the molecular, ecological and evolutionary mechanisms involved in these responses. The data indicate that different loci are important for different facets of the shade avoidance syndrome, which, in itself, may be self-evident. They could also, however, be interpreted as evidence for differential evolution within Arabidopsis of traits evoked by the same environmental signal. Whether this evolution has occurred as a result of selection, or drift, or even has been compounded by human action, is impossible to tell at present. Experiments designed to examine the fitness characteristics of the different genotypes in crowded versus open environments will be needed to answer that question.


This work was funded by awards from the Antorchas Foundation and the British Council to J.F.B. and a NERC Research Grant to H.S. The authors are grateful to Angela Dines, who carried out some of the hypocotyl elongation experiments, and to Malcolm Pratt, who maintained the spectral quality cabinets in working order for a remarkably long period of time to complete these surveys. Thanks are due to Wendy Stoddart for technical assistance and general cheerfulness.


The following material is available from http://www.blackwell-science.com/products/journals/suppmat/PCE/PCE812/PCE812sm.htm

Table S1. This table contains the raw data from which the graphs and figures presented in the written paper have been constructed.

Received 19 July 2001;received inrevised form 21 September 2001;accepted for publication 21 September 2001