Allometry and plasticity of meristem allocation throughout development in Arabidopsis thaliana

Authors

  • Stephen P. Bonser,

    Corresponding author
    1. Department of Biology, Queen’s University, Kingston, Ontario, Canada, K7L 3N6
      *Correspondence and current address: Department of Ecology and Evolutionary Biology, Corson Hall, Cornell University, Ithaca, New York 14853, USA, (tel. 607 254 4282; fax 607 255 8088; e-mail:spb23@cornell.edu).
    Search for more papers by this author
  • Lonnie W. Aarssen

    1. Department of Biology, Queen’s University, Kingston, Ontario, Canada, K7L 3N6
    Search for more papers by this author

*Correspondence and current address: Department of Ecology and Evolutionary Biology, Corson Hall, Cornell University, Ithaca, New York 14853, USA, (tel. 607 254 4282; fax 607 255 8088; e-mail:spb23@cornell.edu).

Summary

  • 1 The three principal fates of axillary meristems in flowering plants are branching or vegetative growth, reproduction, and inactivity. We examined the relationship between allocation to each of these fates and plant size in five genotypes of Arabidopsis thaliana. Nutrient availability was used to obtain a range of plant size.
  • 2Allocation to each meristem fate was recorded at weekly intervals throughout development and at final developmental stage.
  • 3 There were significant allometric relationships between meristem allocation and plant size for each meristem fate throughout development in each genotype. Allometry and plasticity tended to be strongest early in development. In addition, although there were differences in allometry between genotypes, meristem allocation through development was generally consistent across genotypes.
  • 4 We interpret the results of this study as an adaptive response to nutrient availability. Increased allocation to growth meristems in high nutrient treatments early in development should maximize both size and the total number of meristems that can be committed to reproduction later in development. This suggests that the pattern of meristem allocation is a component of an adaptive strategy through development in Arabidopsis thaliana.

Introduction

Fitness in plants should be closely related to the patterns of development that determine plant architecture. A major determinant of plant architecture is the pattern of allocation of meristems to growth and reproduction. In turn, meristem allocation controls such components of fitness as the number of reproductive units (Geber 1990), plant height and size (Duffy et al. 1999), and the ability to forage for resources (McPhee et al. 1997). Meristem allocation to growth and reproduction throughout development should reflect, in part, a program of development particular to a given species. Variation in meristem allocation within a species can be associated with the evolution of strategies that may maximize fitness in a given environment (Bonser & Aarssen 1996).

The relative importance in populations of variation between genotypes and within organisms has important implications for the evolutionary patterns of meristem allocation. If variability of a given meristem fate within a population is entirely between genotypes, then the evolution of meristem allocation will be due to selection among phenotypes that differ in genetically fixed patterns. If the expression of a meristem fate were due to organism-level plasticity, then evolution of meristem allocation would have occurred as a product of selection of the developmental program of meristem allocation within individuals (Diggle 1993). The developmental nature of phenotypic plasticity has been demonstrated in several studies (e.g. Pigliucci & Schlichting 1995; Pigliucci et al. 1997). Phenotypic plasticity is assumed to be an essential component of a plant’s ability to respond to changing environments (Bradshaw 1965; Sultan 1987; Watson et al. 1997). The modular nature of plants allows them to continue to alter the number and type of functional units throughout development (Watson et al. 1997). Recent studies have demonstrated the potential plasticity of meristem allocation in patterns of floral meristem allocation (Diggle 1993, 1994) and branching (a product of meristem differentiation) (e.g. Geber 1989; Bonser & Aarssen 1994; de Kroon & Hutchings 1995). Examining plasticity of meristem allocation throughout development may be important in interpreting variation in meristem allocation and understanding the evolution of adaptive strategies in plants.

‘Different aspects of the phenotype of an organism are (and must be) correlated’ (Schlichting & Pigliucci 1998). Thus, it may be more important to examine the relative allocation to traits than the absolute allocation to a single trait, such as the relationship between plant size and form and function (e.g. through patterns of meristem allocation) or allometric relationships (i.e. size correlated variations in form and function). The relationship between plant size and the allocation to its component parts often reflect consequences of natural selection operating on the relationship between plant form and function (Niklas 1994). Allometric relationships are generally thought to be the product of biomechanical constraints (e.g. Givnish 1986; O’Brien et al. 1995; Henry & Aarssen 1999). However, there is a growing body of evidence that growth form in plants is environmentally dependent (e.g. Bell & Tomlinson 1980; Jones & Harper 1987; Slade & Hutchings 1987a,b; Geber 1989; Schmitt & Wulff 1993). An allometric relationship between two traits expressed in individuals across a range of environments can be interpreted as a type of plasticity. The potential adaptive nature of allometry has been demonstrated in recent studies (Weiner & Thomas 1992; Bonser & Aarssen 1994).

Defining a strategy for meristem allocation

Different patterns of plant form and function are fundamentally a product of patterns of meristem fates (Harper 1987). The above-ground meristems on a plant form a population of functional units that can be classified into three different kinds at any one point in time: an inactive (I) meristem is quiescent or undeveloped because of meristem suppression (by apical dominance), senescence or death; an R (reproductive) meristem forms a flower or inflorescence; and a G (growth) meristem forms a vegetative branch. Commitment of an axillary meristem to growth (G) does not terminate its activity. As the terminal meristem of a branch, a G meristem produces vegetative biomass (and more meristems) and then becomes either inactive (I) or reproductive (R) at the final developmental stage of the plant. G meristems therefore represent two functional units separated in time (G then I, or G then R). It is generally the case for flowering plants that each leaf axil subtends at least, and usually only, one meristem (or the product of its development) (Bell 1991). Hence, the number of R meristem functional units expressed over a plant’s life or growing season is equivalent to the number of flowers or inflorescences (both axillary and terminal), the number of G meristem functional units is equivalent to the number of branches, and the number of I meristem functional units is equivalent to the number of leaf axils without flowers or branches plus the number of vegetative shoot apices. However, at any earlier point in development, inactive meristems should be calculated only as the number of leaf axils without branches or flowers as vegetative shoot apices are normally involved in leaf production and new meristem production; hence, they are not technically inactive. The total number of meristems on a plant is the sum of the reproductive, growth and inactive meristems. Under the meristem based model, relative allocation to R meristems is a measure of reproductive effort. Similarly, relative allocation to I and G meristems measures apical dominance and branching intensity, respectively (Bonser & Aarssen 1996).

In this study, we grew genotypes of the short-lived annual plant Arabidopsis thaliana (L.) Heynh. on a range of nutrients in order to examine the potential for phenotypic plasticity in the relationship between meristem allocation and plant size throughout development. The relationship between meristem allocation and components of fitness in Arabidopsis has been established in a previous study (Duffy et al. 1999). Lifetime reproduction can be maximized by allocation to both reproductive and growth meristems. Branching in annual plants is primarily a mechanism to increase the total number of reproductive meristems at the end of the growing season and, compared with perennials, may not be as important to the accumulation of vegetative size (Duffy et al. 1999).

We addressed the following questions. (i) Are there significant allometric relationships between meristem allocation and plant size, where size variability is due only to environmental variability (i.e. is meristem allocation plastic). (ii) How does plasticity in meristem allocation change throughout plant development. Examining the strength of allometry through development can identify when plasticity of meristem allocation is possible or potentially important. (iii) Are allometry and plasticity variable across genotypes within a species? Genetic variability in allometry and plasticity will indicate the potential for plasticity in meristem allocation to evolve in populations.

Materials and methods

Five inbred lines (genotypes) of Arabidopsis (Bla, Ob, Kil, Fi and Rsch) were grown on a nutrient gradient. Genotypes were obtained as seed from the Arabidopsis Information Service (Frankfurt, Germany) and were originally collected from natural populations. Forty individuals of each genotype were transplanted individually into pots containing a sterilized potting mix (Fison’sTM Sunshine Basic Mix #2) at the emergence of both cotyledons. The pots were placed in a growth chamber under 16 h light (250 µmol m−2 s−1, 23.0 °C, 75% relative humidity) and 8 h dark (21 °C, 75% relative humidity) and were randomly re-positioned within the chamber once per week. Ten replicate plants of each genotype were assigned to each of four nutrient treatments, using 0.00, 0.25, 1.00 and 2.00 g L−1 solutions of a 20 : 20 : 20 N : P : K fertilizer. For each nutrient level, 25 mL of the appropriate solution was applied per pot at weekly intervals until the end of the experiment. Plants were grown to final developmental stage (i.e. when there was no further change in the number or allocation of meristems).

In order to assess plasticity of meristem allocation throughout development, weekly meristem counts were conducted on all individuals from each of the five genotypes. Seven weekly records of allocation to R, I and G meristems were made, starting 1 week after transplanting. In Arabidopsis, inactive meristems occur only in leaf axils that do not subtend R or G meristems. Because flower primordia form and can be seen on the terminal ends of each branch at the same time as the axillary meristem is committed to growth (S. Bonser, personal observation), no terminal meristems are inactive. Due to the large size and advanced developmental stage of most plants, meristem counts were not conducted on weeks eight and nine. Plants were harvested at the end of the tenth week, when all plants had completed their developmental program. Total numbers of R, I and G meristems were recorded for all plants at final developmental stage.

Data analysis

Meristems from each individual across the four nutrient treatments for each genotype at each age were used to test for significant plasticity in meristem allocation. For a given genotype at a given age, the total number of reproductive (R) and vegetative (I + G) meristems in each plant was used to test for an allometric relationship between reproductive effort and plant size across individuals that differ in size due only to differences in environmental conditions. Total number of vegetative meristems (I + G) was used as a measure of plant size rather than total number of meristems (R + I + G) because the latter can result in spurious correlations that may obscure the relationship between reproductive and vegetative components (e.g. Samson & Werk 1986; Klinkhamer et al. 1990). Similarly, the relationship between I and R + G meristems can be used to test for allometry and plasticity of apical dominance, and the relationship between G and R + I meristems can be used to test for allometry and plasticity of branching intensity.

Reduced major axis (model II) regression was used to test for an allometric relationship between each meristem fate and plant size. Model II regression is appropriate where there is error in both the x and y variables of the regression model (Sokal & Rolf 1995). An allometric relationship is indicated where the slope (scaling exponent) of the relationship between the logarithm of one meristem fate and the logarithm of the sum of the two other meristem fates differs from one (isometry) (Niklas 1994). Use of the reduced major axis regression minimizes the chance of erroneous conclusions due to an artificially reduced slope of the model I regression (Henry & Aarssen 1999).

Plasticity in meristem allocation is indicated by allometry, i.e. where the slope of the above relationship is not unity. A significant difference of a reduced major axis slope (b) from a known slope (β) can be performed on the test statistic (Clarke 1980; McArdle 1988):

image(eqn 1)

where r2 is the square of the correlation coefficient between a given meristem fate and the sum of the other two. The test statistic T is approximately t-distributed with 2 + ((n − 2)/(1 + 0.5r2)) degrees of freedom. Where β = 1 (the scaling exponent for an isometric relationship), a significant difference (P < 0.05) for a given genotype at a given age between b and β indicates significant allometry and plasticity of meristem allocation.

The strength of the allometric relationship may vary throughout development (where there is a difference between the slope of the allometric relationship at two different ages within a genotype) or between genotypes (where there is a difference between the slope of the allometric relationship between two genotypes) at a given age. Variability in the allometric relationship between two ages within a genotype, or between two genotypes at a given age can be assessed by testing for a significant difference between the slopes of two reduced major axis regressions (b1 and b2) where each regression is calculated from the logarithm of one meristem fate vs. the logarithm of the sum of the other two, using the following equation (Clarke 1980; McArdle 1988):

inline image

where r21 is the square of the correlation coefficient of the relationship between allocation to one meristem vs. the sum of the other two for the first age (or genotype) of the pair, and r22 is the square of the correlation coefficient of the relationship for the second age (or genotype). The test statistic T is approximately t-distributed with degrees of freedom calculated following McArdle (1988).

Pairwise comparisons of the slope at each age (or genotype) against all other ages (or genotypes) were conducted to test for differences in the strength of allometry (plasticity) throughout development within a genotype, or across genotypes at a given age. Pairwise comparisons throughout development or across genotypes were subjected to table-wide sequential Bonferoni correction to control for multiple comparisons (Rice 1989).

Results

Allometry of meristem allocation

Meristem allocation showed size-dependence across environments in each of the five genotypes (Figs 1 & 2). At final developmental stage (at age 10 weeks), allocation to reproductive meristems was significantly greater in larger plants than in smaller plants for two of the five genotypes (Fi and Rsch) (slopes greater than 1 in Fig. 1a). Allocation to inactive meristems, however, was significantly greater in smaller plants than in larger plants for these genotypes (slopes less than 1 in Fig. 1b). Allocation to growth meristems increased with increasing plant size for four of the five genotypes (Fi, Bla, Kil and Rsch) (Fig. 1c). Plasticity of meristem allocation tended to be greater in the earlier weeks of development than at final developmental stage (Fig. 2). Thus, slopes for both reproductive effort and branching intensity tended to be greatest early in life, and decreased towards isometry throughout development. Similarly, slopes for apical dominance were lowest early in life and increased towards isometry at final developmental stage in apical dominance in all five genotypes (Fig. 2). This indicates that the strength of allometry and plasticity is variable throughout development in Arabidopsis. There were very few instances of plastic allometry detected for the Ob genotype (Fig. 2b). This is probably due to the high mortality throughout development (low sample sizes) for plants of this genotype. Results for the significance test for each model II regression are reported in an appendix (Appendix 1, in the Journal of Ecology archive on the World Wide Web; see the cover of a recent issue of the journal for the WWW address).

Figure 1.

Model II regressions of the allometric relationship between (a) log R vs. log (I + G) meristems, (b) log I vs. log (R + G) meristems, and (c) log G vs. log (R + I) meristems for plants growing under different environments (nutrient levels) (i.e. plasticity). Results are shown for plants at final developmental stage (10 weeks). Lines represent the model II regressions for Fi (●), Ob (▪), Bla (◊), Kil (▿) and Rsch (▴) genotypes. The leading diagonal line represents the slope of an isometric relationship (slope = 1). Solid lines represent slopes that are significantly different (P < 0.05) from isometry; dashed lines represent slopes that are not different from isometry.

Figure 2.

Scatterplots of the coefficient of allometry (the slope of the relationship between meristem allocation and plant size) for each of the five genotypes at each age where meristem allocation was recorded. Each graph shows the coefficient of allometry for reproductive effort (▪), apical dominance (◆), and branching intensity (●). Closed symbols indicate where the coefficient of allometry was significantly different (P < 0.05) from isometry (slope = 1) (i.e. plastic allometry); open symbols indicate where the coefficient of allometry was not significantly different from isometry. The horizontal line represents isometry.

Allometry and development within and between genotypes

Very few significant differences were detected between any of the pairwise comparisons of allometric relationships between meristem allocation and size between weeks (results not shown). However, in general, allometry (non-isometry) was strongest early in development in each of the genotypes (Fig. 2). These trends of plasticity through development can be seen most clearly in the Kil genotype (Fig. 3). Similarly, pairwise comparisons of allometric coefficients show that while there are a few cases where genotypes at a given age show differences in the allometric coefficients, only one (Table 1) remained significant after the Bonferoni correction. The simultaneous effects of variability in age (developmental stage) and genetic identity on the strength of plasticity of the relationship between meristem allocation and plant size is displayed in Fig. 4.

Figure 3.

A three dimensional representation of allometry due to plasticity throughout development for the Kil genotype. Each line represents the model II regression of the relationship between (a) log R vs. log (I + G) meristems, (b) log I vs. log (R + G) meristems, and (c) log G vs. log (R + I) meristems at each of six weeks. Regression lines for ages with different letters have slopes that are significantly different (P < 0.05). Data points have been omitted to increase clarity.

Table 1.  Pairwise comparisons (between each of the five genotypes) of the coefficients of allometry for reproductive effort, apical dominance and branching intensity at 7 weeks of age. T statistic and P-values are shown for each comparison. P-values in parentheses were not significant after sequential Bonferoni correction. Significant P-values are shown in bold. Slopes for each genotype at each age are graphically displayed in Fig. 2
  ObBlaKilRsch
RE     
FiT0.301.772.942.30
 P0.770.09(0.006)(0.03)
ObT 0.750.061.32
 P 0.460.950.20
BlaT  0.920.75
 P  0.360.46
KilT   0.28
 P   0.78
AD     
FiT1.061.753.122.67
 P0.300.09 0.004(0.01)
ObT 0.080.380.48
 P 0.940.710.64
BlaT  1.021.0
 P  0.310.33
KilT   0.28
 P   0.78
BI     
FiT0.481.832.370.92
 P0.640.080.020.36
ObT 0.260.460.48
 P 0.800.650.63
BlaT  1.310.53
 P  0.200.60
KilT   0.11
 P   0.91
Figure 4.

A three-dimensional representation of the effects of variation in age and genetic identity on the strength of allometry and plasticity for each of the three meristem allocation traits: (a) reproductive effort, (b) apical dominance, and (c) branching intensity. Each point represents the coefficient of allometry, i.e. the slope of the model II regression between allocation to each meristem fate and plant size for plants growing under different environments (nutrient levels). The mesh connects each point to highlight the overall changes in the strength of plasticity of meristem allocation through development and between genotypes. Note that in order to improve clarity, the scale of the age axis reads from old to young plants for reproductive effort and branching intensity, while it reads from young to old for apical dominance.

Discussion

Allometry

There was significant allometry for the relationship between each meristem fate and plant size (the sum of the other two fates) throughout development for each of the five genotypes. In general, proportional allocation to reproductive effort and branching intensity increased with increasing plant size (caused by increasing nutrient availability) and allocation to apical dominance decreased with increasing plant size (Figs 1 & 2). Thus, life history traits defined by proportional allocation to meristem fates are phenotypic ally plastic across a range of nutrients in Arabidopsis. Significant within-organism variation suggests that plasticity in developmental properties within individuals is potentially important in defining adaptive strategies. Each meristem experiences a temporal series of potential allocation fates that contribute to the relative proportions of reproductive, inactive and growth meristems. Variability in meristem allocation across different levels of nutrient availability demonstrates that there is lability in the fates of axillary meristems throughout development (see Diggle 1993).

We predicted that an adaptive response to greater nutrient availability would be an increase in proportional allocation to growth meristems, thus maximizing both size (and competitive ability) and the total number of meristems that may be eventually committed to reproduction (Bonser & Aarssen 1996; Duffy et al. 1999). The capacity for the environment to affect the probability of branching has been repeatedly documented in plant populations (see Sutherland & Stillman 1988; references therein). Following meristem allocation throughout development may be particularly important in detecting plastic responses to nutrient availability, as early life allocation to branching is of primary importance in maximizing size at the end of development (Geber 1989, 1990). Within a developmental stage, increased allocation to one meristem fate results in decreased allocation to one or both other fates (Bonser & Aarssen 1996). Thus, allocation to G meristems at final development may come at a cost in terms of reduced potential reproductive output. While there was significant plasticity in meristem allocation at final development for some of the genotypes, the departure from isometry tended to be greater at younger ages for each genotype (Fig. 2). This indicates that plasticity in meristem allocation as a component of an adaptive strategy of growth and reproduction may be most important early in life, and may not even be detected later in life.

We detected significant plasticity in the degree of apical dominance (allocation to inactive meristems). High apical dominance in soils with low nutrient availability may be due to there being insufficient resources available to allocate to branching or reproduction. In such nutrient poor soils, growth will be slow and plants allocating meristems to extensive growth may leave insufficient time for successful reproduction before the end of the growing season. The allometric coefficient of inactive meristems and plant size tended to increase towards isometry throughout development (Fig. 2), possibly because any acquired resources are allocated to the initiation of reproductive meristems before final development.

A number of other factors could contribute to the allometric relationships. (1) The meristem based approach used here allows a three way continuum of trade-offs and correlations, where each meristem fate may or may not be correlated with the other two (Duffy et al. 1999). Thus, functional responses of one meristem fate (e.g. G meristems) to nutrient availability may also result in significant plastic responses in the other two meristem fates (i.e. R and I meristems). (2) If a minimum vegetative size must be achieved before allocation to reproduction can be initiated, an allometric (plastic) relationship between reproductive effort and size can be observed in populations (Weiner 1988; Schmid & Weiner 1993). Size thresholds may also result in phenotypic plasticity in apical dominance and branching intensity. (3) Differences in rate of development across environments may result in significant plasticity in meristem allocation due to differences in developmental stage at a common age, confounding interpretation of phenotypic variation (Clauss & Aarssen 1994; Coleman et al. 1994; Gedroc et al. 1996; Huber & Stuefer 1997).

Variability in meristem allocation through development and between genotypes

Allometry and plasticity in Arabidopsis can be seen from the early life stages and is maintained throughout development. This experiment, in which individual plants were subjected to a given level of nutrients from the emergence of the first leaves until the final developmental stage showed that plants can respond to their environment from the initial developmental stage as well as throughout development. However, we did not test for the ability to respond to an environment that changes within the lifetime of the plant. It is possible that early allocation to branching in response to environmental conditions early in life may fix the plant’s architecture and reduce its ability to respond to environmental changes later in the growing season. Phenotypic plasticity can be highly constrained by patterns of development (Coleman et al. 1994). Further studies are required to determine if plants can respond to continually changing environments by altering their patterns of meristem allocation throughout the course of development.

In general, the allometric coefficients between any two genotypes at a given age were not significantly different, demonstrating that each of the five genotypes has similar patterns of development and similar responses to nutrient availability. The role of meristem allocation in defining plant body plan may constrain the range of meristem allocation between genotypes as all genotypes share the same basic architecture.

Plasticity and the evolution of meristem allocation strategies

The contribution of nutrient availability, genetic differences and development to variation of meristem allocation in an experimental population of Arabidopsis is encapsulated in Fig. 4. This integrated view of variation within populations allows us to simultaneously examine the potential for: (i) plants to display allometric patterns of meristem allocation, (ii) plasticity to change through development, and (iii) variability due to genetic identity. Total variability in the relationship between meristem allocation and plant size is due to complex interactions between environmental variability, age (or developmental stage) and genotype. Low power in the tests between allometric coefficients (due to the high number of comparisons, and Bonferoni correction) could account for the lack of significance of pairwise comparisons across ages and genotypes.

Plasticity in all meristem fates throughout development is obviously not possible since the fundamental architecture of a species requires that some meristems have a fixed developmental fate. However, it is not difficult to imagine a fitness advantage if the fates of a proportion of meristems on an individual could be developmentally and functionally labile. The observed pattern of stronger plasticity early in life suggests that the importance of developmental plasticity may be diminished as plants near final development. In semelparous annual plants such as Arabidopsis, maximum allocation to reproduction should be favoured at final development at the cost of reduced plasticity of meristem fates.

Genetic variation in developmental plasticity has been documented in a number of recent studies (Diggle 1993, 1994; Pigliucci & Schlichting 1995; Pigliucci et al. 1997). This variation suggests that phenotypic plasticity can be the target of selection. Genetic variability in developmental plasticity could be the result of between-genotype differences in meristem senescence or longevity (Richards 1982; Watson et al. 1997) or in the ability to hold and release meristems from dormancy (Newton & Hay 1996; Watson et al. 1997). Opposing selection pressures favouring either early or late reproduction may select for plasticity in the fates of meristems early in development. The evolution of strategies of meristem allocation should be a product of continual modifications on the developmental fates of individual meristems. Future studies should reveal the relative importance of allometry and genetic variation in allometry in the evolution of strategies of meristem allocation throughout development in plants.

Acknowledgements

M. Watson, C. Eckert, K. Preston and J. Weiner made valuable comments on earlier versions of this manuscript. This research was supported by an NSERC research grant to L.W.A. and NSERC and OGS postgraduate scholarships to S.P.B.

Received 29 February 2000 revision accepted 19 June 2000

Ancillary