Genetics and evolution of phenotypic plasticity to nutrient stress inArabidopsis: drift, constraints or selection?

Genetics and evolution of phenotypic plasticity to nutrient stress inArabidopsis: drift, constraints or selection?

Biological Journal of the Linnean Society (1998), 64: 17–40. With 4 figures Article ID: bj970193 Genetics and evolution of phenotypic plasticity to n...

413KB Sizes 0 Downloads 60 Views

Biological Journal of the Linnean Society (1998), 64: 17–40. With 4 figures Article ID: bj970193

Genetics and evolution of phenotypic plasticity to nutrient stress in Arabidopsis: drift, constraints or selection? MASSIMO PIGLIUCCI∗ AND NOAH BYRD Departments of Botany and of Ecology & Evolutionary Biology, University of Tennessee, Knoxville, TN 37996–1100, U.S.A. Received 19 May 1997; accepted for publication 18 September 1997

To better understand the genetic basis and evolution of phenotypic plasticity, we have investigated how the model plant Arabidopsis thaliana (Brassicaceae) responds to nutrient stress. A preliminary experiment showed that two populations that are very closely related genetically tended to respond in a similar fashion to a variety of nutrient stresses. We then asked if there is a general relationship between the degree of genetic differentiation of 16 natural populations of A. thaliana and the similarity in the way they cope with a fundamental nutrient stress, nitrogen limitation. We also grew plants from four mutant lines known to be affected in nitrogen uptake and metabolism, using their background isogenic line as a control. This last experiment tested whether or not defects in major genes involved in nitrogen bioprocessing affect the intensity or pattern of phenotypic plasticity. We found a high degree of genetic differentiation among populations for the ability to respond to nitrogen stress. However, we detected no significant correlation between the genetic distance among natural populations and the similarity of their response to low nitrogen availability. Since the genetic distances among populations were measured using neutral molecular markers, this suggests that random genetic drift and other non-deterministic evolutionary phenomena were not the driving force shaping differences among populations in the response to stress. On the other hand, several characters were highly correlated in their responses to nitrogen limitation, suggesting either that they were modified by natural selection in a like manner, or that they are influenced by similar genetic constraints (due to either pleiotropy or tight linkage). Finally, the mutants did not differ from the parental wild type strain in their pattern of nitrogeninduced stress response. Therefore, although the genes defective in the mutants are part of the biochemical pathway that uptakes and metabolizes nitrates, we conclude that they are not involved in the control of phenotypic plasticity to nitrogen limitation in this species.  1998 The Linnean Society of London

ADDITIONAL KEY WORDS:—nitrogen stress – nitrate reductase – genetic distance – plasticity genes. CONTENTS

Introduction . . . . . . . . . . . . . . . . . . . . . . . Material and methods . . . . . . . . . . . . . . . . . . .

18 20

∗ Correspondence to M. Pigliucci. Email: [email protected] 0024–4066/98/050017+24 $25.00/0

17

 1998 The Linnean Society of London

18

M. PIGLIUCCI AND N. BYRD

Plant material and experimental setup . . . . . . . . . . . . Statistical analyses . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . Experiment I: identical plasticities of Landsberg and Columbia . . . Genetic distances and geography . . . . . . . . . . . . . Experiment II: variation for plasticity to nitrogen stress among natural populations . . . . . . . . . . . . . . . . . . . . Experiment II: matrix comparisons . . . . . . . . . . . . . Experiment III: effects of mutations affecting nitrogen uptake and metabolism on phenotypic plasticity to nitrogen stress . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . Is there a correlation between differences in phenotypic plasticity among natural populations and their genetic similarity? . . . . . . . . Are the plasticities of different characters correlated in natural populations? Are known loci involved in nitrogen uptake and metabolism capable of affecting the intensity or pattern of phenotypic plasticity? . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . .

. . . . .

20 22 24 24 25

. .

27 28

. .

28 32

. .

32 34

. . .

35 36 37

INTRODUCTION

One of the most interesting and controversial research efforts in the study of phenotypic plasticity is the investigation of its genetic basis and how it relates to the evolution of plastic responses (Blows & Sokolowski, 1995; Coupland, 1995; Gloeckner & Beck, 1995; Brakefield et al., 1996). There are fundamentally three types of genes that can underlie a given type of phenotypic plasticity. (1) Allelic sensitivity (Schlichting & Pigliucci, 1993). In this case, the same set of genes controls the expression of a given character in any number of environmental conditions, but the products of the alleles present at those loci respond differently to at least some of these environments. For example, the reactivity of an enzyme could be altered by pH or temperature. (2) Plasticity genes (sensu Pigliucci, 1996a). Here the gene(s) codify for a specific environmental receptor, which is sensitive to a limited number (often only one particular type) of external stimuli, and which triggers the action of several other genes thereby channelling the developmental program down one of two or more possible alternatives. (3) Transduction genes. This class is conceptually introduced in this paper in the context of plasticity studies, and it represents a subset of the original definition of plasticity genes (Schlichting & Pigliucci, 1995). Transduction plasticity genes are regulatory genes which are turned on or off by specific environmental stimuli. However, these genes respond to the action of the plasticity genes described in (2), not directly to the environmental signal(s). Of course, all three types of genes can be involved in the same plastic response, but it is possible that if one or two classes are missing, this may dramatically affect the possible evolutionary trajectories. Differences of opinion in this regard are not difficult to find. While some authors argue that a knowledge of the genetic machinery underlying plasticity is inconsequential to modelling phenotypic evolution (Via, 1993a,b), others concede that it is important at least under certain conditions (van Tienderen & Koelewijn, 1994; de Jong, 1995; Via et al., 1995). Still more researchers maintain the view that elucidation of the genetic basis of plasticity is a sine qua non for a biologically meaningful understanding of organismal evolution in heterogeneous environments (Scheiner, 1993b; Schlichting & Pigliucci, 1993, 1995; Pigliucci, 1996b).

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

19

We think that although it may be possible to predict short term evolution of phenotypic plasticity without considering the details of its genetic control, its longterm evolution depends on such details (Pigliucci, 1996b). Several authors have advanced theoretical and mathematical arguments to support this view (Turelli, 1988; Houle, 1991; Gromko, 1995). Furthermore, the investigation of the genetics of any evolutionarily relevant trait is a worthwhile goal in and of itself, and an increasing volume of classical and molecular genetic literature has focused on the study of the genetic control of various types of phenotypic plasticity (Aufsatz & Grimm, 1994; Breto et al., 1994; Chandler & Robertson, 1994; Dubcovsky et al., 1994; Hubel & Schoffl, 1994; Kiyosue et al., 1994; Short & Briggs, 1994; Smith, 1995; Brakefield et al., 1996). The experiments presented here address the general question of the genetics and evolution of phenotypic plasticity by looking at one of the most crucial and basic responses to environmental stress in plants: how they cope with limited nutrients in the soil. In this study we ask the following specific questions: (I) Is there a correlation between similarities in phenotypic plasticity among natural populations and their overall genetic similarity as assessed using neutral molecular markers? If so, this would suggest that plasticity evolves largely by non-directional forces such as random genetic drift and mutation accumulation. Notice that this is a comparison between genetic and phenotypic degrees of divergence among populations, not a direct correlation between molecular and phenotypic markers. If the latter were attempted, we would not expect any significant correlation regardless of the evolutionary forces in action, unless the molecular markers were numerous enough to effectively saturate the genome and therefore include (or closely flank) at least some of the genes controlling the plasticities. (II) Are the plasticities in response to nutrient stress of different characters correlated among natural populations? If so, this would imply either similar genetic constraints (due to pleiotropy or linkage), or the action of concerted selection. In other words, such a correlation would be supportive of a directed (i.e. nonrandom) mechanism underlying the evolution of plasticity to nutrients. (III) Are loci known to affect the uptake and metabolism of nitrogen capable of altering the intensity or pattern of phenotypic plasticity to limited nutrient availability? If so, then these genes would be another example of ‘plasticity genes’; if not, then we have to hypothesize the existence of distinct genes involved in sensing and reacting to nutrient deficiency, which are independent of some of the known genetic elements controlling nitrogen uptake and metabolism. To address these questions, we performed three experiments designed to study phenotypic plasticity to nutrient stress in natural populations and laboratory lines of the model system Arabidopsis thaliana (Brassicaceae), recently adopted for studies in ecological genetics (Aarsen & Clauss, 1992; Clauss & Aarsen, 1994a,b; Zhang & Lechowicz, 1994, 1995; Pigliucci & Schlichting, 1995, 1996; Pigliucci et al., 1995a, b; van Tienderen et al., 1996), as well as for more classical molecular research (Meyerowitz, 1989; Pyke, 1994; Kunkel, 1996). First, we examined the phenotypic plasticity to a wide variety of nutrient stresses of two very closely related inbred lines, raised under constant (and therefore selectively identical) laboratory conditions for decades. This would tell us if plasticity can diverge by random processes in

20

M. PIGLIUCCI AND N. BYRD

closely related populations in a time span that is very limited from an evolutionary standpoint (question I). Second, we focused on the response to the nutrient stress which emerged as the most marked one from the first experiment, i.e. nitrogen availability. We used 16 natural populations for which we knew the genetic distances from each other based on a battery of 62 Restriction Fragment Length Polymorphisms (RFLPs). The relationships measured by RFLPs are highly correlated with known genetic relationships or pedigrees (King et al., 1993). These markers therefore provide a measure of differentiation among populations mostly due to random evolutionary forces: any directional force acting on a particular RFLP or closely linked marker would be highly unlikely to act on most or all of the other markers randomly scattered across the genome, unless one can envisage some sort of genome-wide selection force. This combination of plasticity data and genetic distances allowed us to further investigate question (I) by broadening the range of genetic differentiation of our sample (though at the cost of focusing on only one type of stress). In this experiment we also compared the patterns of phenotypic plasticity for different biologically important traits to address question (II). Of course, this experiment is unable to discern between the two major directional forces of evolutionary change, genetic constraints and selection; for this we would need more detailed studies of the genetic variation within populations, as well as estimates of selective pressures under field conditions. In the third experiment, we compared the plasticities of specific mutants impaired in either nitrogen uptake or nitrogen processing with their respective wild type to explore question (III). MATERIAL AND METHODS

Plant material and experimental setup This research was conducted using the model system Arabidopsis thaliana (L.) Heynh. (Brassicaceae). This is a cosmopolitan weed commonly used for developmental and molecular studies (Meyerowitz, 1989; Pyke, 1994). Despite its high selfing rate (Abbott & Gomes, 1989), this species harbours substantial genetic variation for quantitative characters and for plasticity to a variety of environmental stimuli (Aarsen & Clauss, 1992; Clauss & Aaarsen, 1994b; Zhang & Lechowicz, 1994; Pigliucci et al., 1995b; Pigliucci & Schlichting, 1996). All plants used in these experiments were obtained through the Arabidopsis Information and Management Service (AIMS - http://aims.cps.msu.edu/aims/). We employed two highly inbred lines, 16 natural populations, and four singlemutant lines. The two inbred lines are Landsberg and Columbia, commonly used for research in molecular biology and physiology. Both lines were derived from a mildly polymorphic natural population living in Poland (Redei, 1992). The two lines are genetically distinct, but the degree of their genetic similarity is very high, and they have been cultivated under standard laboratory conditions (i.e. very similar selective environments) ever since they were collected. The natural populations were: An-1 (Belgium), Bla-12 (Spain), Ct-1 (Italy), Edi-0 (Scotland), Eil-0 (Germany), En-2 (Germany), Lip-0 (Poland), Mt-0 (Libya), No-0 (Germany), Np-0 (Germany), Nw-4 (Germany), Oy-0 (Norway), Rsch-4 (Russia), Sf2 (Spain), Tsu-0 ( Japan), and Wil-2 (Russia). Unfortunately, little is known about the ecology of these populations, except for the distinction between late (Edi-0 and

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

21

Sf-2) and early (all others) flowering ‘ecotypes’. This difference is a reflection of the environment in which the plants live, with early flowering populations being characteristic of highly disturbed environments (e.g. frequently mowed lawns, or flowering in the spring following germination, to avoid intense competition from other crucifers and grass), and late flowering populations occurring in more stable habitats (e.g. abandoned railroads; or flowering after overwintering as rosettes) ( Jones, 1971; Thompson, 1994). In broad terms, we can expect at least Edi-0 and Sf-2 to have experienced different selective pressures from all other populations, but there might be significant variation in the intensity and pattern of selection also within the early flowering group. Genetic distances based on 62 polymorphic RFLP markers among all populations and between these populations and the inbred control line Landsberg were kindly provided by G. King (King et al., 1993). These authors screened the nuclear genome of A. thaliana, available through 25 k clones generated by E. Meyerowitz. RFLP were scored after labelling the intact phage DNA with 32P, hybridizing to Southern blots and cutting with EcoRI and BglII digestion enzymes. The four mutant lines were: (1) chl1–6, a chlorate-resistant mutant characterized by reduced chlorate (and therefore nitrogen) uptake; this mutant was uncovered because of the insertion of the Tag1 transposon into the 4th intron of the CHL1 gene (Tsay et al., 1993). (2) chl2, a chlorate-resistant mutant displaying a 70–80% reduction in nitrate reductase activity; this mutant is characterized by reduced levels of the nitrate reductase cofactor molybdenum-pterin (MoCo) and was originally isolated by exposure to nitroguanidine (LaBrie et al., 1992). (3) chl4, a chlorateresistant mutant with little or no nitrate reductase activity; this mutant is also characterized by reduced levels of the nitrate reductase cofactor molybdenum-pterin (MoCo) (Braaksma & Feenstra, 1982). (4) chl6, a chlorate-resistant mutant with a 90% reduction in nitrate reductase activity; this mutant also has reduced levels of the nitrate reductase cofactor molybdenum-pterin (MoCo) and was obtained by Ethyl-Methane-Sulphonate mutagenesis (LaBrie et al., 1992). All mutants were originally derived from the Landsberg genetic background, and were backcrossed for several generations after isolation. We set up three experiments, one including the two closely related inbred lines exposed to a variety of nutrient stresses; the second one with all natural populations and the Landsberg inbred line exposed to nitrogen deficiency; and a third one including the four mutants and the corresponding isogenic line. Landsberg therefore provided an internal control for all experiments, as well as a way to directly compare the variation in the natural populations with the one found among the mutants. For all experiments, we put seeds in humidified Petri dishes and stored them in a refrigerator at 4°C for a week to stimulate and synchronize germination. We planted 12 replicates of each population or mutant line in each of a series of treatments (seven for the first experiment, two for the remaining ones) in vermiculite in 24-cell flats with 5 by 5.5 cm cells (5.5 cm depth). We placed the flats in two three-level light racks illuminated by two 40-watt fluorescent tubes (Grow-LuxTM) and two 25-watt incandescent bulbs. The replicates of each line/treatment combination were randomly assigned to racks, levels, and trays. The temperature was kept at about 20°C. Trays were bottom-watered with deionized water throughout the experiment. A complete standard Hoagland nutrient solution was applied to the control plants, at the rate of 2 ml twice per week. The remaining flats (‘stress’ treatments, corresponding to a specific nutrient stress) received a solution containing

22

M. PIGLIUCCI AND N. BYRD

only 10% of the amount of calcium, magnesium, nitrogen, phosphorus, potassium, or sulphur compared to the control solution, applied at the same rate. In the second and third experiments only low nitrogen stress was applied. Nitrogen was administered as an equal mixture of Ca(NO3)2 and KNO3. For the second and third experiments we measured the actual amount of nitrogen present in a sample of five pots before and after treatment, using a LaMotte⊆ soil kit test (Stegner, 1993). Before treatment all pots displayed a nitrogen concentration of less than 5 ppm. After treatment the pots which received the low nitrogen treatment had an average nitrogen concentration of 5 ppm, while the pots that received the high nitrogen treatment had an average nitrogen concentration of 15 ppm. These levels of nitrogen are representative of those found in natural populations of A. thaliana in Tennessee (M. Pigliucci & H. Callahan, pers. observ.). Flats were rotated every week throughout the experiment to further minimize microenvironmental variation. During experiments two and three we measured the following six characters on each plant: (1) Duration of vegetative growth, the interval between germination and initiation of bolting, during which the vegetative rosette grows. (2) Number of rosette leaves at the end of the vegetative phase, an estimate of meristem allocation to the vegetative phase of the life cycle; these leaves are the primary contributors to photosynthesis in A. thaliana (although small cauline leaves are produced as bracts at the base of each inflorescence). (3) Rosette diameter measured at the end of the vegetative phase, when leaves stop elongating; it quantifies the growth of the leaves, independently of the number of meristems associated to leaf production. (4) Duration of reproductive growth, the time between bolting and senescence (the plants were harvested about a week after their first fruits started to dehisce, in order not to lose seeds while still allowing for the maturation of most flowers). (5) Inflorescence size, the total length of the inflorescence, counting all branches; this is a good estimate of total size of the reproductive structures, since it accounts for the more or less branched architecture of different populations. (6) Fruit production, which is a good proxy for reproductive fitness in this species, since number of fruits and number of seeds are highly correlated, and seed germination is very high under a broad range of conditions (Westerman & Lawrence, 1970; Pigliucci et al., 1995b). Notice that in A. thaliana there is a potential trade-off between leaf production and fruit production, since the same meristems can be allocated to further vegetative growth or to the origination of supplementary inflorescences. This trade-off is particularly evident under stress conditions, when resources are limited (Pigliucci et al., 1995a). During the first experiment, we measured some of the same traits just described, and a few additional ones. In particular, we also measured seedling characters such as hypocotyl length and day of emergence of the first true leaf, as well as number of basal shoots. On the other hand, rosette size was not determined during the first experiment. Statistical analyses After checking for compliance with parametric assumptions, we ran a standard analysis of variance on the full data set for each experiment using Systat version

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

23

6.0 (SYSTAT, 1996). The statistical model for the analysis of the data from the inbred lines and the natural population experiments included the following effects. Population, estimating the mean genetic differences among populations, irrespective of the environment. Treatment, quantifying the amount of variance explained by the difference in nutrient availability (plasticity), regardless of the particular population. Treatment by Population interaction, measuring the degree of genetic differentiation for phenotypic plasticity among populations. A similar model was used to analyse the data from the mutants experiment, where the Population and Treatment by Population effects were substituted by the analogous Line and Treatment by Line effects. Since Population, Line, and Treatment were considered fixed effects, they and their interaction effects were tested over the common error variance. In the case of both experiments, the full model also included a block effect, which we eventually dropped because it did not explain a significant portion of the phenotypic variance. The results are reported as type III Mean Squares and corresponding probabilities of the F-ratios. We applied a table-wide sequential Bonferroni correction to account for multiple tests, given that the analyses were carried out on more than one trait simultaneously (Rice, 1989). We plotted the data as standard reaction norm plots, that is plots in which the availability of nutrients is on the abscissa and the phenotypic value is on the ordinate. Each reaction norm therefore represented the average behaviour of each population (or mutant line) in each treatment. For the second experiment, we obtained a UPGMA dendrogram visualizing the matrix of genetic distances among the natural populations using the NT-SYS⊆ package (NT-SYS, 1996). We constructed analogous dissimilarity matrices quantifying the divergence in the degree of plasticity between all pairwise combinations of populations. To this effect, we calculated the plasticity for each character as the difference between the expression of that character in the control treatment and the expression of the same trait in the low nitrogen treatment. We then calculated the Euclidean distances between the plasticities of each population for that given character. This procedure yields one lower triangular matrix per character (based on population means). These matrices were then compared in two series of tests. All comparisons were carried out by calculating a matrix correlation coefficient (NT-SYS, 1996) whose significance was tested by random permutations of one of the two distance matrices (genetic or phenotypic) while holding the other constant (Mantel, 1967; Manly, 1986; Cheverud et al., 1989). First, we obtained the correlation between each plasticity matrix and the matrix of genetic distances. This tests the hypothesis of an association between the degree of genetic differentiation of our populations and the divergence in their plasticity to nitrogen stress. Second, we compared the plasticity matrix of each trait with the analogous matrices for other traits. This tests the hypothesis that there is a correlation between the plasticities of two different characters. We did not run all pairwise combinations of tests to economize statistical power. Instead, we concentrated on a series of comparisons which are particularly biologically relevant: (1) all pairwise comparisons of the plasticities of vegetative traits (vegetative duration, number of leaves, and rosette diameter), which would be an indication of a genetic or functional integration in the way vegetative traits respond to nitrogen stress; (2) all pairwise comparisons of the plasticities of reproductive traits (reproductive duration, inflorescence size, and fruit production); (3) vegetative vs. reproductive duration, a comparison designed to test the hypothesis that the plasticities of timing of vegetative and reproductive events are related; (4) rosette diameter vs. inflorescence size, to test if the plasticities of plant’s size during the

24

M. PIGLIUCCI AND N. BYRD

T 1. Analysis of variance on eight traits expressed by the inbred lines Landsberg and Columbia of Arabidopsis thaliana when exposed to six types of nutrient stress (Experiment I). Type III MS are reported; P-values are in parentheses. The numbers immediately below each source of variation indicate the degrees of freedom. Boldface highlights effects significant after a sequential Bonferroni correction Character (df) Hypocotyl length First leaf emergence Duration of vegetative growth Number of rosette leaves Duration of reproductive growth Number of basal branches Inflorescence size Number of fruits

Genotype (1)

Treatment (6)

Genotype by by treatment (6)

0.36 (0.4182) 1.19 (0.3271) 78.06 (0.0001) 95.56 (0.0001) 3.65 (0.2900) 6.56 (0.0001) 8677.64 (0.0001) 4864.14 (0.0001)

3.88 (0.0001) 17.14 (0.0001) 26.38 (0.0001) 0.75 (0.3499) 35.86 (0.0001) 21.41 (0.0001) 416.13 (0.0001) 6732.21 (0.0001)

1.12 (0.0637) 6.28 (0.0001) 1.44 (0.6570) 0.79 (0.3098) 8.83 (0.0140) 1.84 (0.0002) 12.4 (0.2030) 151.18 (0.3880)

Error (258) 0.55 1.24 2.09 0.66 3.25 0.40 8.66 142.73

vegetative and reproductive phases are related; and (5) number of leaves vs. fruit production, to test if the plasticities of meristem allocation during the vegetative and reproductive phases are related (as it would be expected given the constraint referred to above). It has to be noted that Mantel’s tests are very powerful methods for matrix comparisons for two reasons: (1) they are based on the derivation of an empirical distribution of correlations under the null hypothesis, thereby avoiding any restriction caused by parametric assumptions while at the same time retaining the power given by the knowledge of a specific distribution (as opposed to generalized non-parametric tests) (Cheverud et al., 1989). (2) The null hypothesis is that the matrices are not correlated. This is a much more sensitive test than the one based on the complementary hypothesis that the matrices are fully correlated; that is, that they are either proportional or identical (Sokal & Rohlf, 1981).

RESULTS

Experiment I: identical plasticities of Landsberg and Columbia Landsberg and Columbia did not show any significant mean (i.e. across treatment) genetic difference in the expression of their seedling traits, even though both hypocotyl length and time to first leaf emergence were highly significantly plastic (Table 1). Most of the plasticity was due to the marked response to nitrogen stress, where the hypocotyl was about two-thirds the length that it reached under control conditions, and where it took the seedlings about two extra days to achieve emergence of the

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

25

first true leaf (Fig. 1). Time to first leaf emergence showed a significant genotype by environment interaction, clearly attributable to the marked divergence between the two lines in their reaction to low potassium: while Landsberg grew more slowly then its control, Columbia was indistinguishable from it (Table 1; Fig. 1). The situation was very different during the vegetative phase. Here, the two lines were genetically distinct across treatments for both duration of vegetative growth and number of rosette leaves (Table 1), with Columbia growing longer and with more rosette leaves than Landsberg (Fig. 1). Only the duration of the vegetative phase was affected by nutrient availability, with the plants slowing down by about four days under nitrogen deficiency; the leaf number remained unaltered across treatments. There was no significant genotype by environment interaction for vegetative traits (Table 1; Fig. 1). During the reproductive phase, all characters showed statistically significant genetic differences across treatments between Landsberg and Columbia, with the exception of the duration of reproductive growth (Table 1). Columbia tended to have more basal branches (supernumerary inflorescences), to be markedly taller, and therefore to produce more fruits than Landsberg (Fig. 1). In fact, under control conditions, Landsberg produced 48 fruits while Columbia produced 58, a difference in fecundity of about 20%. All these traits, including duration of reproductive growth, were significantly plastic. The strongest response was elicited once again by nitrogen deprivation, but calcium limitation also significantly reduced branching, height, and particularly fruit production (Table 1; Fig. 1). The analysis of variance detected a significant genotype by environment interaction only for basal branching. The cause of this interaction is mostly attributable to the reversal of ranking between the two populations induced by low magnesium (Landsberg branched slightly more than Columbia), and by the wider separation between the means of the two lines under low potassium (Landsberg showed reduced branching, while Columbia was unaltered) (Table 1; Fig. 1). Overall, Experiment I suggested that nitrogen stress elicits by far the strongest plasticity in A. thaliana (at least, when 10% availability of any nutrient is considered), and that the two closely related inbred lines differ very little, if at all, in their phenotypic plasticity to nutrient stress for a variety of characters.

Genetic distances and geography Our re-analysis of King et al.’s data (King et al., 1993) produced a complex branching of populations grouped in three major clusters and a number of smaller groups (Fig. 2). The figure does not hint at any obvious grouping by geographical location. For example, one cluster spanned the range from Norway (northern Europe) to Libya (North Africa), while another cluster joined together a population from Poland and one from Japan. This lack of congruence is not surprising, given that A. thaliana has achieved cosmopolitan status relatively recently through human dispersal and that man-driven dispersal events need not follow any geographical pattern. Throughout this paper, however, we consider the RFLP markers as selectively neutral or almost so, which makes them useful to trace evolution of gene frequencies by random drift and other non-directional evolutionary processes (e.g. mutation) in this species.

Duration of reproductive growth (days)

44

C

L

C

C C

L C

L C

L

L

L C

C

L

C C

C L

L

L

C

C L C

L

L

9.5 9 8.5

C

L C

L

C

L C

L

L C

C

(D)

7 C

6.5

C

C

C

C C

C

6 5.5

L

L

L

L

L

C

C

L L

3

(E)

C

C

L C C

C L

C

L C

C

40

L

L

L L

2.5

(F)

L

L

C

C

2

L L

1.5 1

C

0.5

60

(G) C

30

C

C

C

L

50

C

(H) C

C

L

C

C

C L

L

L L

L

L

15

No. fruits

L

25

L

Con

Ca

Mg

N

C

40 30

L

L

P

K

L

C L

20 C L

10

10

L

L C

C

0

35

20

L

L

10

4.5

38

Inflorescence size (mm)

C

10.5

5

42

39

(B)

11

7.5

(C)

43

41

12 11.5

8

No. rosette leaves

26 25 24 23 22 21 20 19 18 17

(A) L

No. basal branches

Duration of vegetative growth (days)

3.2 3 2.8 2.6 2.4 2.2 2 1.8 1.6 1.4 1.2

First leaf emergence (days)

M. PIGLIUCCI AND N. BYRD

Hypocotyl length (mm)

26

0 P

K

S

Con

Ca

Mg

N

S

Figure 1. Reaction norms to various nutrient stresses of eight characters in the inbred lines Landsberg and Columbia of Arabidopsis thaliana. C=Columbia, L=Landsberg. Bars indicate standard errors. A, hypocotyl length; B, first leaf emergence; C, duration of vegetative growth; D, number of rosette leaves; E, duration of reproductive growth; F, number of basal branches; G, inflorescence size; H, number of fruits. Abbreviations: Con, control; Ca, calcium; Mg, magnesium; N, nitrogen; P, phosphorus; K, potassium; S, sulphur.

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

27

Genetic distances based on RFLP markers 32

24

16

8

0 An Oy Mt Np Eil En Wil Rsch Nw Ct Lip Tsu Sf Bla Edi No Lan

Figure 2. UPGMA dendrogram visualizing the genetic distances among 16 natural populations of Arabidopsis thaliana and one isogenic line (Landsberg). Data are from an RFLP analysis based on 62 polymorphic markers published in (King et al., 1993) and kindly provided by the authors. The dendrogram was constructed using the NT-SYS software package.

T 2. Analysis of variance for the natural populations included in the study (Experiment II). Type III MS are reported. Boldface indicates significant effects according to a sequential Bonferroni correction Character (df)

Population (16)

Treatment (1)

Population by treatment (16)

Duration of vegetative growth

12386.59 (0.0001) 2677.66 (0.0001) 25.82 (0.0001) 18.39 (0.0001) 13395.28 (0.0001) 3297.94 (0.0001)

1309.65 (0.0001) 1134.17 (0.0001) 801.98 (0.0001) 85.98 (0.0001) 511023.03 (0.0001) 222688.67 (0.0001)

997.19 (0.001) 474.73 (0.0001) 9.16 (0.0001) 6.45 (0.1855) 5799.67 (0.0001) 2099.33 (0.0001)

Number of rosette leaves Rosette diameter Duration of reproductive growth Inflorescence size Number of fruits

Error (343–347) 33.46 20.29 0.63 4.91 363.73 214.71

Experiment II: variation for plasticity to nitrogen stress among natural populations The natural populations examined here showed highly significant genetic differentiation (independent of the environment) for all traits measured, as evidenced by the significance of the population effect in all cases (Table 2). All characters were also highly significantly plastic (treatment effects in Table 2), and there was very significant genetic differentiation for phenotypic plasticity in all traits except the duration of reproductive growth (population by treatment effects in Table 2). Upon inspection of the reaction norm plots, two major features of the data become evident

28

M. PIGLIUCCI AND N. BYRD

(Fig. 3). First, for some traits (duration of the vegetative period and number of rosette leaves) most of the genetic variation and of the variation for plasticity was attributable to a few divergent lines (Fig. 3). The most differentiated populations were the two late flowering ones, Sf-2 from Spain and especially Edi-0 from Scotland. Three of the remaining four characters, on the other hand, showed a continuous spread of the reaction norms in a fan-like fashion (Fig. 3), indicating no net distinction among populations (the last character, duration of reproductive growth, did not show any significant variation for plasticity: Table 2; Fig. 3). Second, for four traits (number of leaves, rosette diameter, inflorescence size, and fruit production) there was more genetic variation expressed under control conditions than under low nitrogen. The converse was true for duration of vegetative growth, which showed more variation under conditions of stress; however, it must be emphasized that this reversal was largely attributable to the anomalous behaviour of one population (Edi0). In the case of duration of reproductive growth, the reaction norms were essentially flat, indicating little (albeit statistically significant) plasticity for this character among populations. Experiment II: matrix comparisons The first set of matrix comparisons involved the matrix of RFLP-based distances and the six matrices describing the differences among natural populations in their degree of phenotypic plasticity to nitrogen stress. All six tests were clearly not significant, suggesting independence between the genetic distance as measured by the 62 molecular markers and the diversification of plasticity patterns in these populations (Table 3). We then calculated the correlations between a series of plasticity matrices, selected from all possible comparisons as described in Material and Methods. Three of the nine tests yielded highly significant results (Table 4). The first set of comparisons was among traits expressed during vegetative growth. Of these three correlations, the one between vegetative duration and number of leaves was highly significant; neither of the plasticities of these two traits, however, was correlated to the plasticity of rosette diameter. The second set of comparisons involved characters measured during the reproductive phase. Again, one of the three correlations was highly significant, relating the plasticity of fruit production to the plasticity of inflorescence size; yet, neither of these two plasticities was correlated with the plasticity of reproductive duration. In the remaining comparisons, we found no relation between the plasticities of vegetative and reproductive duration (i.e. the lengths of the two phases of the life cycle), or between the plasticities of number of leaves and fruit production (i.e. the two measures of meristem allocation); there was, on the other hand, a highly significant correlation between the plasticities of rosette diameter and of inflorescence size (the two measures of plant size during the two phases of the life cycle). Experiment III: effects of mutations affecting nitrogen uptake and metabolism on phenotypic plasticity to nitrogen stress The analysis of variance of the mutant data showed a highly significant difference in all phenotypic traits of the mutant lines, independent of the treatment (Table 4,

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

140.00

60.00

(A) No. rosette leaves

Duration of vegetative growth (days)

160.00 120.00 100.00 80.00 60.00 40.00 20.00

(C)

Duration of reproductive growth (days)

200.00 180.00 160.00 140.00 120.00 100.00 80.00 60.00 40.00 20.00 0.00

(B)

50.00 40.00 30.00 20.00 10.00 0.00

(E)

27.00 26.00 25.00 24.00 23.00 22.00 21.00 20.00 19.00 18.00 17.00

100.00

(D)

(F)

80.00 No. fruits

Rosette diameter (cm)

10.00 9.00 8.00 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00

Inflorescence size (cm)

0.00

29

60.00 40.00 20.00

Low

Control

0.00

Low

Nutrients

Control Nutrients

An-1

Mt-0

Bla-12

No-0

Ct-1

Np-0

Edi-0

Nw-4

Eil-0

Oy-0

En-2

Rsch-4

Landsberg

Sf-2 Tsu-0

Lip-0 Wil-2

Figure 3. Reaction norm plots for the natural populations and the Landsberg isogenic line in response to nitrogen availability. A, duration of vegetative growth; B, number of rosette leaves; C, rosette diameter; D, duration of reproductive growth; E, inflorescence size; F, number of fruits.

30

M. PIGLIUCCI AND N. BYRD

T 3. Matrix comparisons using the Mantel test. Matrix I represents a matrix of genetic distances based on RFLP analysis; Matrix II is a series of matrices quantifying the differences among populations in the plasticity of a given trait. No correlation was significant Matrix I

Matrix II

genetics genetics genetics genetics genetics genetics

vegetative duration number of leaves rosette diameter reproductive duration inflorescence size number of fruits

Correlation

P-value

0.1849 0.1701 0.0403 0.0818 0.1159 0.0200

0.1494 0.1550 0.3736 0.3177 0.1580 0.4276

T 4. Matrix comparisons using the Mantel test. Matrix I and II represent a series of matrices quantifying the differences among populations in the plasticity of a given trait. Boldface indicates correlations significant according to a sequential Bonferroni test Matrix I

Matrix II

Correlation

P-value

vegetative duration vegetative duration number of leaves

number of leaves rosette diameter rosette diameter

0.8164 −0.0796 0.1766

0.0003 0.6869 0.1227

number of fruits number of fruits inflorescence size

inflorescence size reproductive duration reproductive duration

0.5429 0.0635 0.1437

0.0001 0.3094 0.1507

vegetative duration

reproductive duration

−0.0142

0.5232

rosette diameter

inflorescence size

0.6654

0.0001

number of leaves

number of fruits

−0.1775

0.9220

line effects). In stark contrast, no character showed a significant heterogeneity for plasticity among these lines, even though three traits came close to formal significance (Table 5, line by treatment effects). Figure 4 shows the actual pattern of the reaction norms of the mutants when compared to Landsberg. Notice that had the graphs in Figure 4 been plotted on the same scale used for the 16 natural populations (Fig. 3), the reaction norms would form a tight, parallel bundle; we have, however, plotted them on a finer scale to facilitate close inspection of several interesting features of this data set. We wish to emphasize that these features concern the overall mean behaviour of these lines, while they are independent of the environment, and therefore do not relate to the degree or type of phenotypic plasticity. First, rosette diameter, inflorescence size, and fruit production followed the same pattern among mutant lines that they did in the natural populations, with higher phenotypic values and genetic variance under the control treatment and a marked convergence towards identical phenotypes in the stress treatment (Fig. 4). Second, all mutants grew more slowly than the control under both conditions (longer duration of vegetative growth in the mutants vs. the wild type), but did not produce fewer fruits (in fact, chl2 managed to produce significantly more fruits than Landsberg) (Fig. 4). Third, the uptake-limited mutant, chl1–6, in most cases did not behave differently from the nitrate-reductase-limited mutants. Among the three nitrate-reductase mutants, chl6

34.00

7.00

(A)

32.00

No. rosette leaves

Duration of vegetative growth (days)

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

30.00 28.00 26.00 24.00 22.00 20.00

(C)

3.50 3.00 2.50 2.00 1.50 1.00

40.00

6.00 5.50 5.00 4.50

27.00 26.00 24.00 23.00 22.00 21.00 20.00 19.00

(F)

60.00

35.00 30.00 25.00 20.00

50.00 40.00 30.00

15.00

20.00

10.00

10.00

5.00

(D)

25.00

70.00

(E)

No. fruits

Inflorescence size (cm)

45.00

(B)

6.50

0.00

Duration of reproductive growth (days)

Rosette diameter (cm)

4.00

31

Low

Control

0.00

Low

Control Nutrients

Nutrients

Chl1–6 Chl2 Chl4 Chl6 Landsberg (wild type)

Figure 4. Reaction norm plots for the four mutant lines and the Landsberg isogenic line in response to nitrogen availability. A, duration of vegetative growth; B, number of rosette leaves; C, rosette diameter; D, duration of reproductive growth; E, inflorescence size; F, number of fruits.

32

M. PIGLIUCCI AND N. BYRD

T 5. Analysis of variance for the mutants and the Landsberg background included in the study. Type III MS are reported. Boldface indicates significant effects according to a sequential Bonferroni correction Character (df) Duration of vegetative growth Number of rosette leaves Rosette diameter Duration of reproductive growth Inflorescence size Number of fruits

Line (4)

Treatment (1)

Line by treatment (4)

356.23 (0.0001) 7.35 (0.0019) 1.71 (0.0038) 74.80 (0.0001) 403.92 (0.0004) 1498.42 (0.0001)

2.13 (0.6247) 0.22 (0.7111) 44.15 (0.0001) 88.26 (0.0010) 15842.69 (0.0001) 24743.77 (0.0001)

5.17 (0.6745) 1.97 (0.3050) 1.04 (0.0458) 14.45 (0.1181) 196.50 (0.0326) 543.60 (0.0580)

Error (106–107) 8.85 1.61 0.41 7.66 71.83 230.36

was markedly different for the duration of the vegetative growth and for leaf production, while chl2 was distinct for the duration of the reproductive growth and for fruit production (Fig. 4).

DISCUSSION

The study of the genetics of phenotypic plasticity has a long history (Scheiner, 1993a), and its importance for the tempo and mode of evolution of plastic responses is certainly a matter of heated debate (Schlichting & Pigliucci, 1993; Via, 1993a,b; Pigliucci, 1996b). In general, we would like to know under what circumstances phenotypic plasticity evolves by random drift or other non-deterministic forces, versus evolution by deterministic forces such as natural selection (Sultan, 1995). Also, much current research is focusing on the identification and characterization of genes controlling plastic responses in natural populations (Smith, 1990; MitchellOlds, 1995). In this paper, we addressed three specific questions related to the evolution of phenotypic plasticity in response to nutrient stress and to its genetic control in the model system Arabidopsis thaliana. Together, our results point towards the unlikeness of a major role of genetic drift or other non-deterministic factors, towards a possible involvement of genetic constraints and/or selection (deterministic forces), and towards no (or limited) role of nitrogen uptake and reductase genes in the control of plasticity to nitrogen availability.

Is there a correlation between differences in phenotypic plasticity among natural populations and their genetic similarity? We found evidence for divergence of the reaction norms of Landsberg and Columbia, the two closely related inbred lines, but the main aspect of such divergence

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

33

was the evolution of distinct genotypic means across treatments, not of distinct plasticities. The resistance of phenotypic plasticity to diverge in these lines may be due to either the fact that plasticities—unlike character means—are controlled by a reduced number of genes (so that it takes more time to accumulate new mutations at the relevant loci), or that plasticities are subjected to a much stronger stabilizing selection. In either case, this is evidence for the prevalence of deterministic forces of evolutionary change (or lack thereof) such as genetic constraints or some form of selection. Without knowledge of the details of the selective regimes to which these plants have been exposed, as well as of their genetic architecture, it is not possible to distinguish between these two alternatives. When we extended our study to consider a wider range of natural populations (but focusing only on nitrogen stress), we found no relationship between the divergence among populations in the patterns and intensities of phenotypic plasticity and the genetic distances among the same populations based on RFLP data. Again, we stress that we compared genetic and phenotypic degree of differentiation among populations; we did not attempt a one-to-one match between genetic and phenotypic markers. This is an important distinction, because if we had carried out the latter analysis, there would be no a priori expectation of a correlation, even if both sets of traits were evolving primarily by drift. This is because drift affects variation and means in a different fashion (Hartl & Clark, 1989). In particular, the evolution of character means in specific populations is unpredictable by the very nature of genetic drift. Therefore, one population could increase the mean of one trait and decrease the mean of a second trait by drift, while a second population could be characterized by exactly the opposite pattern, or increase or decrease both trait values. On the other hand, drift increases the variation across populations for any trait which is not subjected to deterministic forces such as selection. Therefore, we would expect a correlation between the divergence for the two types of traits in our populations if both sets were evolving primarily by drift. This notwithstanding, there are still two possible interpretations of our results: (a) phenotypic plasticity to nitrogen stress in A. thaliana indeed did not evolve via random genetic drift (the most likely evolutionary force determining genetic distances based on neutral molecular markers), as just discussed; or (b) there are very few genes controlling the plastic responses of interest, so that it is difficult to define a clear expectation for the phenotypic variation that they exert, whatever the evolutionary forces at play. Although we cannot exclude it, we do consider the second alternative less likely, for three orders of reasons. First, the plasticities that we study are quantitative characters, and our general understanding of these traits is that they tend to be affected by many genes (although not necessarily with small effects; Orr & Coyne, 1992). Second, at least in the case of the plasticity of the duration of the vegetative phase, already 22 loci have been identified as being directly involved in the control of this trait; of these, at least 19 perform some sort of regulatory action in response to specific environmental conditions, thereby influencing phenotypic plasticity (Coupland, 1995). Third, several of the reaction norms observed in this study showed a continuous distribution among populations, especially under the control conditions. This kind of non-discrete distribution is difficult to reconcile with the action of one or very few loci (Falconer, 1989). Research is currently in progress in our laboratory to test the hypothesis that the plasticities described here in natural populations are under the control of a reduced number of loci (M. Pigliucci & N. Byrd, in prep.).

34

M. PIGLIUCCI AND N. BYRD

Results similar to the ones reported here were found by Jasienski et al. (1997), and by Black-Samuelsson & Andersson (1997). The authors of both these papers interpreted their findings in a manner analogous to ours. The data of Jasienski et al. (1997) constitute a partial exception, because they found a significant correlation between phenotypic plasticity and variation in molecular markers in the case of moisture and temperature gradients (but not for other types of plasticity). They explained this result as indicative of that particular plastic response being polygenically controlled, implying that the high density of molecular markers they used was able to pick up some of the genes affecting plasticity. However, an equally reasonable conclusion would be that those reaction norms evolved by random drift while the other types of plasticity investigated in that study had been selected. The determination of when phenotypic plasticity is the result of selection or of passive reaction to environmental changes remains one of the most difficult tasks facing investigators in this field (Sultan, 1995). Are the plasticities of different characters correlated in natural populations? Schlichting (1986, 1989a,b) first proposed the term ‘plasticity integration’ to indicate the fact that reaction norms for different traits can be correlated. He provided experimental evidence for population differences in plasticity integration, as well as for the dependence of these correlations on the particular suite of traits or environments considered (Schlichting, 1989a,b). There are two main reasons that characters, and therefore plasticities, may be causally correlated. First, the correlation could be due to shared genetic control, or to very closely linked genes; that is, pleiotropy would represent the mechanistic link between two characters or plasticities. Second, the correlation could be the result of past natural selection. This would be the case if it is somehow adaptive for an organism to have those two traits, or two plasticities, covarying across a particular range of environments. The first category of explanation refers to genetic constraints, the second one to functional ecology. We found some interesting correlations between plasticities in A. thaliana, and also some interesting lack of correlation. In the first experiment, where we compared the plasticities to a variety of nutrient stresses in Columbia and Landsberg, we detected the same pattern of plasticity, almost regardless of the particular stress. The major differences were in the amount of the plastic response elicited by each stress, not in its direction. It seems, therefore, that the plant is reacting in a similar fashion to any nutrient-related limitation. This is not the case for the response to completely different kinds of stress, such as water or light (Pigliucci et al., 1995a,b). Lois (1994) found a similar relationship between the kind of stress and the response of A. thaliana when investigating the mechanistic basis of plasticity to UV radiation. It is tempting to speculate that such a consistency of response to related stresses reflects an adaptive mechanism of control at the whole-plant level, as hypothesized by Chapin (1991). However, the same considerations that make it difficult to convincingly demonstrate adaptive plasticity for single traits (see above), apply even more strictly to the case of plasticity integration. In the second experiment, we focused on the correlation between plasticities of different traits in response to the same type of stress. We did find a relationship between the plasticity of time to flowering and the plasticity of leaf production. On the other hand, neither of these two reaction norms correlated with the reaction

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

35

norm of rosette diameter. This established a link between the duration of the vegetative phase in this plant and meristem allocation during that phase, but not with overall growth and the resulting final size of the rosette. On the contrary, during the reproductive phase we found a significant correlation between the plasticities of size and meristem allocation, but not between either of these and the length of that phase. These combined results point to an interesting differential coupling of length of the growth period, size and meristem allocation during the two major phases of the life cycle of A. thaliana. Accordingly, when we looked for across-phase correlations, we did not find any relationship between the plasticities for the lengths of the two periods or between the plasticities of the two measures of meristem allocation, but we did find a significant relationship between the reaction norms for size during the two phases. Once again, are these patterns the result of past selection, or are they the manifestation of genetic constraints? The answer to this question would require a combined analysis of genetic correlations within populations and of comparative studies within A. thaliana, a study which is currently ongoing in our laboratory (Pigliucci et al., in prep.). On the one hand, a concordance between across-population and within-population correlations would suggest that indeed the same genes (or closely linked ones) are affecting the correlated plasticities. Linkage is a possibility in the case of A. thaliana, given its high level of selfing (Abbott & Gomes, 1989). However, we found surprisingly few correlations among plasticities considering the relatively high level of linkage disequilibrium one would expect in a high selfer with only five chromosomes. As for the possibility of selection, an intra-specific phylogeny of A. thaliana populations coupled with information about the ecology of these natural populations would provide insights into how these correlation patterns evolved. K. Cammell, M. Pigliucci and J. Schmitt (subm.) have used an intra- and inter-specific phylogeny of A. thaliana to investigate reaction norm evolution in response to light availability. Their results point to a negligible phylogenetic effect on the correlations among plasticities, suggesting the concerted action of natural selection. If we were to interpret the genetic distances used in the current study as an indication of phylogeny, we would reach a similar conclusion. Such interpretation, however, should be taken with caution. Are known loci involved in nitrogen uptake and metabolism capable of affecting the intensity or pattern of phenotypic plasticity? The genetic basis of specific plastic responses is usually unknown (Scheiner, 1993a), even though their quantitative genetics has been studied in several instances (Hoffmann & Parsons, 1993; Windig, 1994; Shaw et al., 1995; Bennington & McGraw, 1996). An argument has been made for a distinction between the potentially high number of genes generally involved in plastic responses and a much smaller number of specific ‘plasticity genes’ (Schlichting & Pigliucci, 1995; Pigliucci, 1996a). The idea is that reaction norms are quantitative characters, and as such they are likely influenced by many genes with varying degrees of effect. These genes probably account for most of the genetic variation for plasticity observed in natural populations. However, it is logical to expect, and it has been empirically demonstrated, that a much smaller number of high level regulatory gene products detect environmental signals and initiate a cascade of specific morphogenic reactions (Aphalo & Ballare`,

36

M. PIGLIUCCI AND N. BYRD

1995; Bagnall et al., 1995; Lopez-Juez et al., 1995; Quail et al., 1995; Visser et al., 1995; Yamaguchi-Shinozaki et al., 1995; Crews, 1996; Rozema et al., 1997). We refer to the first type of genes as ‘plasticity modifiers’, and to the second as ‘plasticity genes’ (but see the Introduction for a finer distinction within the category of plasticity genes between signal receptors and signal transducers). In general, our suggestion is that plasticity genes may account for the existence and direction of an adaptive plastic response, while the modifiers may fine-tune the intensity of that response. In the case of phenotypic plasticity to nitrogen stress, information is available on the genetics, biochemistry, and physiology of nitrogen metabolism (Braaksma & Feenstra, 1982; Field, 1983; Cheng et al., 1991; LaBrie et al., 1991, 1992; Tsay et al., 1993; Lillo, 1994; Zonia et al., 1995). Furthermore, we have studies documenting the ecological role of plasticity to nitrogen availability in plants (Snyder & Bunce, 1983; Sultan & Bazzaz, 1993; Longnecker & Robson, 1994; Thompson, 1994). We therefore hypothesized that some of the genetic elements already known to be involved in nitrogen uptake and metabolism, and in particular in the processes of nitrogen uptake and nitrate reduction, might be potential candidates for genes controlling plasticity to nitrogen stress. It turned out that all the mutants we analysed were indeed phenotypically different from the wild type Landsberg control, but chiefly in their across-environment response, not in their plasticity. Admittedly, we did find some marginally nonsignificant differences in the reaction norms of some mutants when compared to the wild type. However, we prefer to be statistically conservative and steer clear from any speculation regarding these effects. Of course, it is possible that these effects are biologically meaningful, and that a larger sample size might allow to make a more convincing case than we were able to. Predictably, all mutants grew at a smaller pace than the wild type during the vegetative (but not the reproductive) phase. Interestingly, all mutants allocated a higher number of meristems to leaf production, thereby effectively behaving as ‘late flowering’ ecotypes. Even more surprisingly, there were no major differences in plant size (either during the vegetative or during the reproductive phase), and very little in reproductive fitness (and one mutant actually produced more fruits than the control). These results can be explained to some extent by the fact that A. thaliana has a second nitrate reductase gene which is normally characterized by very low activity. Whenever the primary pathway is damaged, this redundant system is turned on, counter-balancing the decreased functionality of the primary enzyme (Cheng et al., 1991). This is an interesting example of the fact that living organisms may not normally work ‘at full capacity’, but instead rely on a series of backup systems to cope with environmental or genetic alterations (genetic redundancy; Goldstein & Holsinger, 1992; Pickett & Meeks-Wagner, 1995). However, the quest for genes directly controlling the plasticity to nitrogen stress continues. Perhaps better candidates could be found in genes affecting the differential growth of the root system, especially in response to nutrient deprivation. Some of these genes have recently been isolated in A. thaliana (Benfey & Schiefelbein, 1994; Cary et al., 1995; Smith & Fedoroff, 1995; Bates & Lynch, 1996) and will be the focus of future studies. ACKNOWLEDGEMENTS

We thank Hilary Callahan, Mark Camara, Mitchell Cruzan, M.J. Lawrence, Anna Maleszyk, and Carolyn Wells for critical readings of previous versions of this

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

37

manuscript. We are indebted to S. Eaker for help with Experiment I and to K. McKinney for Experiment II. Thanks also to G. King for providing the RFLP data used in our analyses. This work was supported by NSF grant DEB-9527551 to MP.

REFERENCES

Aarsen LW, Clauss MJ. 1992. Genotypic variation in fecundity allocation in Arabidopsis thaliana. Journal of Ecology 80: 109–114. Abbott RJ, Gomes MF. 1989. Population genetic structure and outcrossing rate of Arabidopsis thaliana (L.) Heynh. Heredity 62: 411–418. Aphalo PJ, Ballare` CL. 1995. On the importance of information-acquiring systems in plant–plant interactions. Functional Ecology 9: 5–14. Aufsatz W, Grimm C. 1994. A new, pathogen-inducible gene of Arabidopsis is expressed in an ecotype-specific manner. Plant Molecular Biology 25: 229–239. Bagnall DJ, King RW, Whitelam GC, Boylan MT, Wagner D, Quail PH. 1995. Flowering responses to altered expression of phytochrome in mutants and transgenic lines of Arabidopsis thaliana (L.) Heynh. Plant Physiology 108: 1495–1503. Bates TR, Lynch JP. 1996. Stimulation of root hair elongation in Arabidopsis thaliana by low phosphorous availability. Plant, Cell and Environment 19: 529–538. Benfey PN, Schiefelbein JW. 1994. Getting to the root of plant development: the genetics of Arabidopsis root. Trends in Genetics 10: 84–88. Bennington CC, McGraw JB. 1996. Environment-dependence of quantitative genetic parameters in Impatiens pallida. Evolution 50: 1083–1097. Black-Samuelsson S, Andersson S. 1997. Relationship between reaction norm variation and RAPD diversity in Vicia dumetorum (Fabaceae). International Journal of Plant Science, 158: 593–601. Blows MW, Sokolowski MB. 1995. The expression of additive and nonadditive genetic variation under stress. Genetics 140: 1149–1159. Braaksma FJ, Feenstra WJ. 1982. Isolation and characterization of nitrate reductase-deficient mutants of Arabidopsis thaliana. Theoretical and Applied Genetics 64: 83–90. Brakefield PM, Gates J, Keys D, Kesbeke F, Wijngaarden PJ, Monteiro A, French V, Carroll SB. 1996. Development, plasticity and evolution of butterfly eyespot patterns. Nature 384: 236–242. Breto MP, Asins MJ, Carbonell EA. 1994. Salt tolerance in Lycopersicon species. III. Detection of quantitative trait loci by means of molecular markers. Theoretical and Applied Genetics 88: 395–401. Cary AJ, Liu W, Howell SH. 1995. Cytokinin action is coupled to ethylene in its effects on the inhibition of root and hypocotyl elongation in Arabidopsis thaliana seedlings. Plant Physiology 107: 1075–1082. Chandler PM, Robertson M. 1994. Gene expression regulated by abscissic acid and its relation to stress tolerance. Annual Review of Plant Physiology and Plant Molecular Biology 45: 113–141. Chapin III FS. 1991. Integrated responses of plants to stress. BioScience 41: 29–36. Cheng CL, Acedo GN, Dewdney J, Goodman HM, Conkling MA. 1991. Differential expression of the two Arabidopsis nitrate reductase genes. Plant Physiology 96: 275–279. Cheverud JM, Wagner GP, Dow MM. 1989. Methods for the comparative analysis of variation patterns. Systematic Zoology 38: 201–213. Clauss MJ, Aarssen LW. 1994a. Patterns of reproductive effort in Arabidopsis thaliana: confounding effects of size and developmental stage. Ecoscience 1: 153–159. Clauss MJ, Aarssen LW. 1994b. Phenotypic plasticity of size–fecundity relationhips in Arabidopsis thaliana. Journal of Ecology 82: 447–455. Coupland G. 1995. Genetic and environmental control of flowering time in Arabidopsis. Trends in Genetics 11: 393–397. Crews D. 1996. Temperature-dependent sex determination: the interplay of steroid hormones and temperature. Zoological Science 13: 1–13. de Jong G. 1995. Phenotypic plasticity as a product of selection in a variable environment. The American Naturalist 145: 493–512. Dubcovsky J, Galvez AF, Dvorak J. 1994. Comparison of the genetic organization of the early salt-stress-response gene system in salt-tolerant Lophopyrum elongatum and salt-sensitive wheat. Theoretical and Applied Genetics 87: 957–964.

38

M. PIGLIUCCI AND N. BYRD

Falconer DS. 1989. Introduction to quantitative genetics. New York: Longman. Field C. 1983. Allocating leaf nitrogen for the maximization of carbon gain: leaf age as a control on the allocation program. Oecologia 56: 341–347. Gloeckner G, Beck CF. 1995. Genes involved in light control of sexual differentiation in Chlamydomonas reinhardtii. Genetics 141: 937–943. Goldstein DB, Holsinger KE. 1992. Maintenance of polygenic variation in spatially structured populations: roles for local mating and genetic redundancy. Evolution 46: 412–429. Gromko MH. 1995. Unpredictability of correlated response to selection: pleiotropy and sampling interact. Evolution 49: 685–693. Hartl DL, Clark AG. 1989. Principle of population genetics. Sunderland, MA: Sinauer. Hoffmann AA, Parsons PA. 1993. Direct and correlated responses to selection for desiccation resistance: a comparison of Drosophila melanogaster and D. simulans. Journal of Evolutionary Biology 6: 643–657. Houle D. 1991. Genetic covariance of fitness correlates: what genetic correlations are made of and why it matters. Evolution 45: 630–648. Hubel A, Schoffl F. 1994. Arabidopsis heat shock factor: isolation and characterization of the gene and the recombinant protein. Plant Molecular Biology 26: 353–362. Jasienski M, Ayala FJ, Bazzaz FA. 1997. Phenotypic plasticity and similarity of DNA among genotypes of an annual plant. Heredity 78: 176–181. Jones ME. 1971. The population genetics of Arabidopsis thaliana. II. Population structure. Heredity 27: 51–58. King G, Nienhuis J, Hussey C. 1993. Genetic similarity among ecotypes of Arabidopsis thaliana estimated by analysis of restriction fragment length polymorphism. Theoretical and Applied Genetics 86: 1028–1032. Kiyosue T, Yamaguchi-Shinozaki K, Shinozaki K. 1994. Cloning of cDNAs for genes that are early-responsive to dehydration stress (ERDs) in Arabidopsis thaliana L.: identification of three ERDs as HSP cognate genes. Plant Molecular Biology 25: 791–798. Kunkel BN. 1996. A useful weed put to work: genetic analysis of disease resistance in Arabidopsis thaliana. Trends in Genetics 12: 63–69. LaBrie ST, Wilkinson JQ, Crawford NM. 1991. Effect of chlorate treatment on nitrate reductase and nitrite reductase gene expression in Arabidopsis thaliana. Plant Physiology 97: 873–879. LaBrie ST, Wilkinson JQ, Tsay YF, Feldmann KA, Crawford NM. 1992. Identification of two tungstate-sensitive molybdenum cofactor mutants. Molecular and General Genetics 233: 169–176. Lillo C. 1994. Light regulation of nitrate reductase in green leaves of higher plants. Physiologia Plantarum 90: 616–620. Lois R. 1994. Accumulation of UV-absorbing flavonoids induced by UV-B radiation in Arabidopsis thaliana L. I. Mechanisms of UV-resistance in Arabidopsis. Planta 194: 498–503. Longnecker N, Robson A. 1994. Leaf emergence of spring wheat receiving varying nitrogen supply at different stages of development. Annals of Botany 74: 1–7. Lopez-Juez E, Kobayashi M, Sakurai A, Kamiya Y, Kendrick RE. 1995. Phytochrome, gibberellins, and hypocotyl growth. Plant Physiology 107: 131–140. Manly FJ. 1986. Randomization and regression methods for testing for associations with geographical, environmental and biological distances between populations. Research in Population Ecology 28: 201–218. Mantel NA. 1967. The detection of disease clustering and a generalized regression approach. Cancer Research 27: 209–220. Meyerowitz EM. 1989. Arabidopsis, a useful weed. Cell 56: 263–269. Mitchell-Olds T. 1995. The molecular basis of quantitative genetic variation in natural populations. Trends in Ecology and Evolution 10: 324–327. NT-SYS. 1996. NT-SYS pc. Setauket, NY: Exeter Software. Orr HA, Coyne JA. 1992. The genetics of adaptation: a reassessment. The American Naturalist 140: 725–742. Pickett FB, Meeks-Wagner DR. 1995. Seeing double: appreciating genetic redundancy. The Plant Cell 7: 1347–1356. Pigliucci M. 1996a. How organisms respond to environmental changes: from phenotypes to molecules (and vice versa). Trends in Ecology and Evolution 11: 168–173. Pigliucci M. 1996b. Modelling phenotypic plasticity. II. Do genetic correlations matter? Heredity 77: 453–460.

GENETICS AND EVOLUTION OF PLASTICITY IN ARABIDOPSIS

39

Pigliucci M, Schlichting CD. 1995. Reaction norms of Arabidopsis (Brassicaceae). III. Response to nutrients in 26 populations from a worldwide collection. American Journal of Botany 82: 1117–1125. Pigliucci M, Schlichting CD. 1996. Reaction norms of Arabidopsis. IV. Relationships between plasticity and fitness. Heredity 76: 427–436. Pigliucci M, Schlichting CD, Whitton J. 1995a. Reaction norms of Arabidopsis. II. Response to stress and unordered environmental variation. Functional Ecology 9: 537–547. Pigliucci M, Whitton J, Schlichting CD. 1995b. Reaction norms of Arabidopsis. I. Plasticity of characters and correlations across water, nutrient and light gradients. Journal of Evolutionary Biology 8: 421–438. Pyke K. 1994. Arabidopsis—its use in the genetic and molecular analysis of plant morphogenesis. New Phytologist 128: 19–37. Quail PH, Boylan MT, Parks BM, Short TW, Xu Y, Wagner D. 1995. Phytochromes: photosensory perception and signal transduction. Science 268: 675–680. Redei GP. 1992. A heuristic glance at the past of Arabidopsis genetics. In Koncz C, Schell J, eds. Methods in Arabidopsis research. Singapore: World Scientific, 1–15. Rice WR. 1989. Analyzing tables of statistical tests. Evolution 43: 223–225. Rozema J, Staaij J van der, Bjorn LO, Caldwell M. 1997. UV-B as an environmental factor in plant life: stress and regulation. Trends in Ecology and Evolution 12: 22–28. Scheiner SM. 1993a. Genetics and evolution of phenotypic plasticity. Annual Review of Ecology and Systematics 24: 35–68. Scheiner SM. 1993b. Plasticity as a selectable trait: reply to Via. The American Naturalist 142: 371–373. Schlichting CD. 1986. The evolution of phenotypic plasticity in plants. Annual Review of Ecology and Systematics 17: 667–693. Schlichting CD. 1989a. Phenotypic integration and environmental change. BioScience 39: 460–464. Schlichting CD. 1989b. Phenotypic plasticity in Phlox. II. Plasticity of character correlations. Oecologia 78: 496–501. Schlichting CD, Pigliucci M. 1993. Evolution of phenotypic plasticity via regulatory genes. The American Naturalist 142: 366–370. Schlichting CD, Pigliucci M. 1995. Gene regulation, quantitative genetic and the evolution of reaction norms. Evolutionary Ecology 9: 154–168. Shaw RG, Platenkamp GAJ, Shaw FH, Podolsky RH. 1995. Quantitative genetics of response to competitors in Nemophila menziesii: a field experiment. Genetics 139: 397–406. Short TW, Briggs WR. 1994. The transduction of blue light signals in higher plants. Annual Review of Plant Physiology and Plant Molecular Biology 45: 143–171. Smith DL, Fedoroff NV. 1995. LRP1, a gene expressed in lateral and adventitious root primordia of Arabidopsis. The Plant Cell 7: 735–745. Smith H. 1990. Signal perception, differential expression within multigene families and the molecular basis of phenotypic plasticity. Plant, Cell and Environment 13: 585–594. Smith H. 1995. Physiological and ecological function within the phytochrome family. Annual Review of Plant Physiology and Plant Molecular Biology 46: 289–315. Snyder FW, Bunce JA. 1983. Use of the plastochron index to evaluate effects of light, temperature and nitrogen on growth of soya bean (Glycine max L. Merr.). Annals of Botany 52: 895–903. Sokal RR, Rohlf FJ. 1981. Biometry. New York, NY: Freeman & Co. Stegner RW. 1993. Plant nutrition studies. Chestertown, MD: La Motte. Sultan SE. 1995. Phenotypic plasticity and plant adaptation. Acta Botanica Neerlandica 44: 363–383. Sultan SE, Bazzaz FA. 1993. Phenotypic plasticity in Polygonum persicaria. III. The evolution of ecological breadth for nutrient environment. Evolution 47: 1050–1071. SYSTAT. 1996. SYSTAT 6.0 for Windows. Chicago: SPSS Inc. Thompson L. 1994. The spatiotemporal effects of nitrogen and litter on the populatin dynamics of Arabidopsis thaliana. Journal of Ecology 82: 63–68. Tsay YF, Frank MJ, Page T, Dean C, Crawford NM. 1993. Identification of a mobile transposon in Arabidopsis thaliana. Science 260: 342–344. Turelli M. 1988. Phenotypic evolution, constant covariances, and the maintenance of additive variance. Evolution 42: 1342–1347. van Tienderen PH, Hammad I, Zwaal FC. 1996. Pleiotropic effects of flowering time genes in the annual crucifer Arabidopsis thaliana (Brassicaceae). American Journal of Botany 83: 169–174. van Tienderen PH, Koelewijn HP. 1994. Selection on reaction norms, genetic correlations and constraints. Genetical Research 64: 115–125.

40

M. PIGLIUCCI AND N. BYRD

Via S. 1993a. Adaptive phenotypic plasticity: target or by-product of selection in a variable environment? The American Naturalist 142: 352–365. Via S. 1993b. Regulatory genes and reaction norms. The American Naturalist 142: 374–378. Via S, Gomulkiewicz R, Jong GD, Scheiner SM, Schlichting CD, Tienderen PHv. 1995. Adaptive phenotypic plasticity: consensus and controversy. Trends in Ecology and Evolution 10: 212–216. Visser JW, Heijnk CJ, Hout KJM van, Voesenek LACJ, Barendse GWM, Blom CWPM. 1995. Regulatory role of auxin in adventitious root formation in two species of Rumex, differing in their sensitivity to waterlogging. Physiologia Plantarum 93: 116–122. Westerman JM, Lawrence MJ. 1970. Genotype–environment interaction and developmental regulation in Arabidopsis thaliana. I. Inbred lines; description. Heredity 25: 609–627. Windig JJ. 1994. Reaction norms and the genetic basis of phenotypic plasticity in the wing pattern of the butterfly Bicyclus anynana. Journal of Evolutionary Biology 7: 665–695. Yamaguchi-Shinozaki K, Urao T, Shinozaki K. 1995. Regulation of genes that are induced by drought stress in Arabidopsis thaliana. Journal of Plant Research 108: 127–136. Zhang J, Lechowicz MJ. 1994. Correlation between time of flowering and phenotypic plasticity in Arabidopsis thaliana (Brassicaceae). American Journal of Botany 81: 1336–1342. Zhang J, Lechowicz MJ. 1995. Responses to CO2 enrichment by two genotypes of Arabidopsis thaliana differing in their sensitivity to nutrient availability. Annals of Botany 75: 491–499. Zonia LE, Stebbins NE, Polacco JC. 1995. Essential role of urease in germination of nitrogenlimited Arabidopsis thaliana seeds. Plant Physiology 107: 1097–1103.