Shading and mechanical stress (MS) modulate plant architecture by inducing different developmental pathways. Shading results in increased stem elongation, often reducing whole-plant mechanical stability, while MS inhibits elongation, with a concomitant increase in stability.
Here, we examined how these organ-level responses are related to patterns and processes at the cellular level by exposing Impatiens capensis to shading and MS.
Shading led to the production of narrower cells along the vertical axis. By contrast, MS led to the production of fewer, smaller and broader cells. These responses to treatments were largely in line with genetic differences found among plants from open and closed canopy sites. Shading- and MS-induced plastic responses in cellular characteristics were negatively correlated: genotypes that were more responsive to shading were less responsive to MS and vice versa. This negative correlation, however, did not scale to mechanical and architectural traits.
Our data show how environmental conditions elicit distinctly different associations between characteristics at the cellular level, plant morphology and biomechanics. The evolution of optimal response to different environmental cues may be limited by negative correlations of stress-induced responses at the cellular level.
In their natural habitat, plants are subjected to a plethora of different environmental stimuli which have their own specific effects on developmental pathways. In concert, these cues modulate the architectural blueprint of a given plant. Plastic responses to one environmental cue can limit or alter the response to other environmental cues (Cipollini, 1997, 2004; Sultan, 2000; Weinig & Delph, 2001; Bruce et al., 2007; Anten et al., 2009; Li & Gong, 2011; Huber et al., 2012; von Wettberg et al., 2012). However, there may be a negative correlation between abillities to show plastic responses to cues eliciting contrasting phenotypic responses (Henry & Thomas, 2002), thereby preventing plants from producing responses to multiple distinct cues. Shading and mechanical stress (MS) are two commonly occurring environmental factors which select for opposite phenotypic responses. Responses to MS (thigmomorphogenesis) generally involve the production of short and thick stems, while shading induces longer and more slender stems (Schmitt et al., 1999; Pruyn et al., 2000; Henry & Thomas, 2002; Liu et al., 2007; Anten et al., 2009; Chehab et al., 2009). However, in plants subjected to both shading and MS, the shade-induced stem elongation was reduced, indicating that the integrated phenotype is a compromise among the responses to the individual cues (Anten et al., 2009). Yet, it is still largely unknown how developmental mechanisms at the cellular level such as cell division, cell growth and spatial arrangement of cells affect and, at the same time, are affected by changes in plant architecture. We expect that the primary developmental processes operate at the cellular level, and that the changes at the organ level, such as changes in internode length, are a consequence of these cellular changes. Yet, the cues leading to modifications in processes at the cellular level may be detected in other parts of the plant. Furthermore, natural selection acts on integrated whole-organism phenotypes and selection on the specific responses at the cellular level will be determined by the relative costs and benefits of this response in terms of plant performance. Thus, bottom-up and top-down processes probably interact in determining the basic architectural blueprint of plants and environmentally induced changes therein. Here, we investigated to what extent changes at the cellular level contribute to plant architectural changes in response to shading and/or MS. Better knowledge of these patterns will allow us to understand both the evolutionary constraints and also the opportunities shaping architectural plasticity in response to multiple environmental cues.
Differences in plant architecture are basically achieved by variation in the number and size of units of which plants are composed. Organ size and shape are regulated by a combination of cell proliferation and cell growth. Although cell proliferation and enlargement are two fundamentally different processes, governed by different genetic mechanisms (Potters et al., 2007, 2009), they are likely to be coordinated and controlled at the level of the organ (Leevers & McNeill, 2005; Tsukaya, 2005; Weijschede et al., 2008; Malinowski et al., 2011). Previous research has revealed an apparent trade-off between cell number and cell size in leaf tissue and petioles, with plants being able to buffer variation in cell proliferation to some extent by alteration of cell size (Tsukaya, 2005; Weijschede et al., 2008). However, it has been shown for Trifolium repens that changes in cell number and size have different effects on biomechanical characteristics; genotypes that elongated their petioles mainly by cell division had less flexible tissue than genotypes that elongated preferentially by producing longer cells (Huber et al., 2008). This indicates that, even if variation in cell size is compensated by variation in cell proliferation, the emergent, integrated phenotype may have different characteristics (Huber et al., 2008). In addition, there may also be physiological constraints on cell size; for example, the efficiency of intracellular passive diffusion processes may be constrained by the cell volume to cell area relationship (Jorgensen & Tyers, 2004).
A possible alternative but largely overlooked mechanism that could underlie organ size changes is alterations of the shape of cells and their spatial arrangement or organization. For example, the production of longer but narrower cells and the alignment of a greater proportion of the cells vertically along the internodes can contribute to increased internode length. From an economic perspective, changes in cell shape and alignment may have lower production costs, as they require fewer additional structures, but may result in costs in terms of reduced stem thickness. Whether selection acts by favoring increased cell number, cell size, cell shape or spatial alignment of cells will depend on the structural costs of a given response and the fitness consequences associated with the resulting biomechanical characteristics in a given environmental setting (Callahan et al., 2008). Detailed knowledge of the developmental pattern at the level of individual cells may provide further insight into the mechanisms and consequences of architectural changes in response to environmental cues.
In this paper, we build on data presented in a previous paper on Impatiens capensis (Anten et al., 2009), where we showed the effects of contrasting environmental stimuli (light and MS) on plant architecture and biomechanical characteristics. In the current paper, we describe the cellular mechanisms determining internode architecture and investigate to what extent patterns at the cellular level affect biomechanical characteristics. To the best of our knowledge, this is the first study that disentangles environmental effects on size and number of cells from the spatial alignment and shape of cells and which relates habitat-specific variation in internode architecture to patterns at the cellular level. Our experiment aimed to answer the following questions. How do spectral shading, MS and habitat of origin affect the basic traits determining plant architecture, namely cell size, cell number, cell shape and spatial cell alignment? How are these cellular characteristics related to plant architecture and performance in terms of mechanical stability? Is there a negative correlation between shading- and MS-induced plasticities of cellular, architectural or biomechanical characteristics?
Materials and Methods
Impatiens capensis Meerb. (jewelweed) is an annual herb that occurs over a wide range of habitats in North America (Tabak & von Wettberg, 2008). The species is characterized by hollow stem centers and can grow to be over 1.5 m in high-light field sites, but is typically only 30–50 cm high in low-light forest understories. It has a mixed mating system, producing both self-fertilized cleistogamous flowers early and outcrossed chasmogamous flowers later in the season (Waller, 1979). Seeds usually disperse < 1.5 m from parent plants (Schmitt et al., 1985). Consequently, substantial micro-geographic genetic differentiation exists in morphological and life history traits both within and among natural populations, and genetically and phenotypically differentiated open and closed canopy forms have been observed (Simpson et al., 1985; Argyres & Schmitt, 1991; Dudley & Schmitt, 1995; Donohue et al., 2000; von Wettberg & Schmitt, 2005; von Wettberg et al., 2008). Genetic differentiation in this species has been demonstrated between populations growing as close as a few meters apart (Lechowicz & Bell, 1991).
In the spring of 2003, seedlings were collected from grassland and forest populations at Weetomo Woods Tiverton Park in Rhode Island, USA (41.5°N, 71.2°W), as described in more detail in von Wettberg et al. (2008). The forest population grew in a mixed Acer rubrum– Fagus americana deciduous forest understory where canopy openness of the trees was c. 6% (von Wettberg et al., 2008). The grassland population grew in a large Carex meadow, at least 50 m from the forest edge, where canopy openness was c. 80%. The two populations have been shown to be morphologically and genetically differentiated despite the short geographic distance between them (von Wettberg et al., 2008). For the current experiment, we used the fifth generation of inbred seeds from 10 families (hereafter genotypes) from each of the two habitats.
The experiment was conducted in the glasshouse facility of Utrecht University, Utrecht, the Netherlands. On 28 August 2006, 24 seeds of each genotype were sown in 1.3-l pots, filled with a 1 : 1 mixture of sand and potting soil and 3 g of slow-release fertilizer (Osmocote, Scotts Co. LLC, Marysville, OH, USA; 10% N + 10% P + 10% K + 3% Mg + trace elements). On 19 September, 16 similar plants of each genotype, consisting of the still elongating hypocotyl and first internode, the cotyledons and the first leaf pair, were selected and randomly assigned to each of two shading and two MS treatments, for a total of four replicates per treatment combination. Individual pots were placed 0.35 m apart. The experiment was laid out in four blocks across the glasshouse, with each block containing one replicate 4 × 5 grid of each treatment combination. Grids were placed 1 m apart to minimize shading effects. Positions of plants within grids and grids within blocks were changed randomly every week to further minimize possible effects of position in the glasshouse.
Two levels of shading were applied by exposing plants to either 15% daylight with a red-to-far red ratio (R : FR) of 0.3 (denoted ‘shade’) or 50% daylight and R:FR = 1.2 (denoted ‘high light’) (Anten et al., 2009). The ‘shade’ treatment was meant to simulate the light conditions that plants experience under a forest canopy and was applied by creating cages covered with one layer of a plastic film (no. 122; Lee Colortran International, Andover, UK). As the plants grew taller, the heights of the cages were increased from 0.7 to 1.5 m. We left 0.2 m of open space below the film to allow free air movement; microclimatic measurements revealed no differences in air temperature and relative humidity between the two shading treatments. Mechanical treatments were applied by flexing plants either 0 (no MS) or 40 times once per day (MS). This was done by gently grasping the stem at c. 80% of its height and by bending it back and forth no further than 30° from the vertical direction. We chose this type of flexing because it simulates the mechanical effect of wind on plants (swaying of the stem) without affecting their microclimate (Smith & Ennos, 2003; Anten et al., 2010).
Between 9 and 12 October, a destructive harvest was conducted. On each day, one complete block was harvested. Total plant height and the length and diameter of the second internode were measured. The shoots were then destructively separated into stems, branches and petioles and the fresh mass of all aboveground plant parts was immediately determined. Stems were immediately packed in wet tissue paper to avoid loss of turgor and stored at 5°C for mechanical measurements and measurements of cell length and diameter. These measurements were performed on the second internode, which is situated between the first and the second leaf pairs.
We investigated the mechanisms of internode elongation based on the cells in the epidermal layer. Even though the contribution of the epidermis to the cross-section of stems is small, its contribution to bending resistance of herbaceous stems might be significant. In turgid herbaceous stems, the epidermis (and probably some cell layers under it) form a kind of ‘skin’ that is held in a state of tension by the hydrostatically inflated inner part of the stem (in much the same way as the interaction between the protoplast and the cell wall in turgid cells). A study by Niklas & Paolillo (1997) showed that the epidermis may thus contribute a significant fraction of stem flexural rigidity. In addition, the epidermis plays an important mechanical role in stem elongation (Schopfer, 2006 and citations therein). Therefore, while the epidermis only represents a small fraction of the stem cross-section, we believe that it can serve as a useful proxy for mechanical stem traits. The epidermal layer has previously been used to demonstrate that cell proliferation and extension can differ among environments and genotypes (Allard et al., 1991; Huber et al., 2008; Weijschede et al., 2008) and to determine the extent to which these cellular patterns at the epidermal layer are correlated with biomechanical characteristics of the whole organ (Ridge & Amarasinghe, 1984; Loodts et al., 2006). For the measurement of cell length and width, a thin peel of the outer epidermal layer was taken at harvest and immediately preserved in a solution of 50% formaldehyde. Photographs of these peels were taken under a light microscope with a digital camera attached to the microscope, and the length and width of 10 randomly chosen cells were measured to the nearest 0.01 mm in three areas. We measured 30 cells per internode, because cells can vary greatly in size and shape within tissue type. The values presented in this paper are the mean values of these measurements. We defined the growth direction of the plant (i.e. along the vertical stem axis) as the main axis of cell alignment, and care was taken that the epidermal peels were positioned in the correct direction when measuring. Cell length refers to the vertical cell dimension and cell width to the horizontal cell dimension (Fig. 1). These data were used to further calculate cell area (cell length × cell width), number of cells vertically along and horizontally around the perimeter of the internode, total number of epidermal cells on the internode, cell shape (cell length/cell width) and cell alignment (number of cells along the internode/number of cells around the internode) (Fig. 1). Cell area will be used as a measure for cell size. Cell shape values > 1 indicate that cells are relatively narrow and long, and cell shape values < 1 indicate the reverse. Differences in cell alignment may point at different patterns of cell division, namely cell division predominately taking place either in a vertical plane (high values of cell alignment) or in a horizontal direction (lower values of cell alignment). Alternatively, cell arrangement may have been determined after cell division in the early stages of internode growth. Our data do not allow us to differentiate between the effects of early cell division and later cell positioning on the relative alignment of cells in the epidermal layer.
Two stem mechanical characteristics were measured on the second internode: the Young's elastic modulus (E; MN m−2), a measure of the tissue rigidity, and the breaking stress (σb; MN m−2), a measure of tissue strength (Niklas, 1992; Gere & Timoshenko, 1999). High values for E and σb indicate that the tissue is stiff and resistant to rupture, respectively. The stem-level traits were used to calculate two measures of mechanical stability at the whole-plant level. The first measure is the buckling safety factor (BSF), which indicates the ability of plants to carry their own weight. BSF < 1 indicates that the plant stem is too long or too thin to carry the aboveground plant structures, even in the absence of lateral forces, and will thus buckle globally (Gere & Timoshenko, 1999). The second measure is the maximum lateral force that plants resist before breaking (Fmax) and is a measure of the ability to resist external (wind) forces. Specifically, it indicates the minimum lateral force required to induce an amount of stress at the stem base that exceeds σb, causing rupture (Anten et al., 2005). Measurements were performed with a universal testing machine (Instron Model 5542; Instron, Canton, OH, USA) using a three point bending technique. Further details of measurements, including equations and the assumption underlying the calculation of E, σb,Fmax, and BSF, are given by Anten et al. (2005, 2009).
The responses of plant traits to treatments were analyzed with mixed model nested ANOVA (Proc GLM, sas 9.1; SAS Institute, Cary, NC, USA). Habitat, light and MS were treated as fixed effects and the random factor genotype was nested within habitats. Two- and three-way interactions were also included in the model. Block was included in the model to account for spatial variation in the glasshouse conditions and for the difference in harvest times among blocks. Data were transformed when necessary to improve normality and homogeneity of the residual variances.
Shade-induced plasticity was calculated as the percentage change of a trait in shaded, no-MS conditions as compared with high-light, no-MS conditions (Weijschede et al., 2006; Huber et al., 2008). MS-induced plasticity was calculated as the percentage change of traits in high-light, MS conditions as compared with high-light, no-MS conditions. This implies that the percentage response was scaled to the conditions where the specific stress did not occur. Consequently, a positive plastic response to MS indicates that the respective trait had a larger value in high-light, MS conditions compared with high-light, no-MS conditions, whereas a negative value for plasticity indicates the opposite. A value of 0 indicates that the respective trait value was not affected by MS. The calculations were performed using the within-treatment genotypic means. Genotypes of both habitats were pooled for the following analyses to increase the strength of the analyses. As there was a strong overlap in the relative responses of genotypes from forest and grassland habitats (see e.g. Fig. 6), we believe that the resulting correlations were not driven by habitat divergence.
The interrelationship among whole-organ traits and the effects of the environment on cellular traits, internode architecture and biomechanical characteristics were analyzed with a genotypic path analysis. Path analytical models can be used to explore and quantify patterns of variation in character correlations (Pigliucci & Kolodynska, 2006). Specifically, we were interested in how cellular characteristics (cell area, cell number, cell shape and cell alignment) contribute to internode architecture and to what extent biomechanical characteristics can be explained by traits at the cellular level. The four cellular traits were entered as exogenous traits. We calculated the correlation among these cellular traits and calculated how they were associated with the architectural traits internode length and internode thickness and the biomechanical traits internode flexibility and BSF (see Fig. 4). Genotypic means were used for all analyses. We performed three sets of analyses. In the first set, we investigated how cellular traits are associated with architectural and biomechanical traits under high-light, no-MS conditions. In the second set, we calculated how shading-induced changes of cellular characteristics are associated with shading-induced changes of architectural and biomechanical traits, and in the third set of analyses we performed the same analysis for MS-induced changes. The program amos (Arbuckle & Wothke, 1999) was used for these analyses.
We performed correlation analyses to test the hypothesis that there is a negative correlation between the expression of shade avoidance and the expression of thigmomorphogenesis (i.e. mechanical-induced changes of traits) by calculating Pearson's correlation coefficients of the percentage response to shading and to MS for cellular, architectural and biomechanical characteristics. Similar to the path analyses, these analyses were performed at the genotypic level, with plants from the two habitats being pooled.
Plant response to shading, MS and habitat of origin
Plants responded to shading by producing significantly longer, thinner internodes and stiffer tissue and having a lower BSF (Fig. 2; Table 1). These internodes were composed of longer and narrower cells (Fig. 3a,b,d; Table 2). Shading did not significantly affect the mean individual cell area and the total number of epidermal cells, but it did affect the alignment of cells; while on average shading reduced the total number of epidermal cells on the second internode by only 4.4%, it significantly increased the relative alignment of cells in the vertical direction (Fig. 3e–h; Table 2). MS-induced responses were largely opposite to those induced by shading. Plants subjected to MS produced shorter and slightly thicker internodes and had a higher BSF (Fig. 2a,b,f; Table 1). Their internodes consisted of shorter and slightly smaller cells (Fig. 3a,c; Table 2). MS reduced cell area by up to 30% (Fig. 3c; Table 2). MS tended to reduce the total number of epidermal cells (5.5%; P =0.052; Fig. 3g; Table 2). In contrast to the effect of shading, MS reduced the relative allocation of cells in the vertical direction (Fig. 3h; Table 2). Shade had a smaller stimulatory effect on cell elongation in plants subjected to MS than in unflexed individuals, as indicated by a significant MS × shade interactive effect on these traits (Fig. 3a; Table 2).
Table 1. Result of mixed model ANOVA testing for the effects of habitat of origin (H), light availability (L) and mechanical stress (MS) on responses of morphological and biomechanical traits in Impatiens capensis
Buckling safety factor
F-values and their significance are given. Significance levels: ns, P >0.1; $, 0.1 ≥ P >0.05; *, 0.05 ≥ P >0.01; **, 0.01 ≥ P >0.001; ***, P ≤0.001. Significant values are presented in bold, and marginally significant values in italics.
Table 2. Result of mixed model ANOVA testing for the effects of habitat of origin (H), light availability (L) and mechanical stress (MS) on responses of cellular characteristics in Impatiens capensis
Cell number along internode
Cell number around internode
Total cell number
F-values and their significance are given. Significance levels: ns, P >0.1; $, 0.1 ≥ P >0.05; *, 0.05 ≥ P >0.01; **, 0.01 ≥ P >0.001; ***, P ≤0.001. Significant values are presented in bold, and marginally significant values in italics.
Plants originating from forest understory habitats produced significantly thinner internodes composed of fewer and narrower but on average longer cells than those from the grassland habitat (Figs 2b, 3a,c,d,g; Tables 1, 2). The total number of epidermal cells was on average 42% lower in forest compared with grassland plants (Fig. 3g; Table 2). This could be attributed to a significant reduction in both the number of cells aligned vertically along and the number of cells aligned horizontally around an internode (Fig. 3e,f; Table 2). Plants originating from the two habitats responded similarly to MS with respect to the cellular characteristics. However, habitat did affect the potential to respond to shading, as the cell shape of forest plants responded more strongly to shading than that of grassland plants, while under shaded conditions cell area was more similar between grassland and forest plants than under high-light conditions (Fig. 3c,d; Table 2).
Effects of cellular characteristics on plant architecture and biomechanical traits
The genotypic path analyses show the extent to which traits at the cellular level affect architectural traits and biomechanical characteristics, and whether cellular traits are interrelated with each other. Under high-light, no-MS conditions, there was a positive correlation between cell shape and cell area as well as between cell alignment and total cell number (Fig. 4a). That is, genotypes that aligned more cells in the vertical direction also had more cells overall, and genotypes with longer cells also had larger mean cell areas (Fig. 4a). Total cell number was negatively correlated with cell area, and genotypes that aligned a larger proportion of their epidermal cells vertically had on average smaller cells (Fig. 4a) under high-light, no-MS conditions.
Internodes were consistently longer in genotypes with a larger cell area or a larger number of cells (Fig. 4). Shading- or MS-induced changes of cell area or total cell number had a similar effect on internode length plasticity, as genotypes that produced more or larger cells in response to shading also produced longer internodes, while genotypes that produced shorter or fewer cells in response to MS also decreased the internode length (Fig. 4b,c). Interestingly, under high-light, no-MS conditions, genetic variation in internode diameter was not correlated with any of the cellular characteristics (Fig. 4a). However, shading- and MS-induced changes in internode diameter were positively associated with environmentally induced increases in total cell number and cell area, but negatively associated with cell width (Fig. 4b,c).
The association between environmentally induced changes of cellular characteristics and environmentally induced changes of biomechanical traits strongly differed for the different environmental conditions and was characterized by only a few significant correlations (Fig. 4). It should be noted, however, that MS-induced reductions of cell area and cell number were associated with a significant increase in tissue stiffness, as indicated by larger values of Young's modulus (Fig. 4c). Aligning more cells in the vertical direction in response to MS, however, was associated with the production of internodes with more flexible tissue (Fig. 4c). In contrast to these strong effects of MS-induced changes in cellular characteristics on internode tissue stiffness, there was no direct association with MS-induced changes in whole-plant stability (BSF, Fig. 4c).
Correlation between responses to shade and MS
We found a positive correlation of plastic responses to shading and to MS for cell shape, number of cells around the internode, total cell number, cell area and internode diameter (Figs 5, 6). For example, genotypes that increased their internode diameter more strongly in response to MS exhibited a smaller reduction in internode diameter in response to shading and vice versa (Fig. 5b). It has to be noted that this apparent positive correlation actually describes a negative correlation between the abilities to respond to shading and MS: as shade avoidance and MS have opposite effects on internode diameter, a positive correlation means that genotypes that elicited the expected response to one environmental cue inevitably expressed a reduced response to the other environmental cue.
In contrast to the cellular traits, shade- and MS-induced responses of biomechanical traits and internode length were not positively correlated. Shade- and MS-induced changes in resistance to rupture (breaking stress, σb) tended to be negatively correlated, indicating that genotypes that exhibited stronger increases in response to MS hardly responded to shading with respect to σb, while genotypes that increased σb in response to shading hardly responded to MS for this trait (Fig. 5d).
The correlation analyses also revealed that different groups of traits responded to either shading or MS (Figs 5, 6). Genotypes varied very little in the anatomy and morphology of internodes in the vertical direction (cell length, cell shape, number of cells vertically along internodes and internode length) in response to MS, but they showed a large variation in the response to shading for these cellular traits. The opposite was true for traits characterizing the plasticity of biomechanical stability and in terms of the maximum lateral force that plants can resist (Fmax), as genotypes varied much more in their response of Fmax to MS than to shading (Fig. 5e).
Variation in plant architecture is linked to processes at the cellular level
Individuals of the same species are often phenotypically variable. Variation in plant growth and development can be related to cell structural attributes – their size, number and geometry – as cells form the basic architectural units of plants. Yet, the degree to which plasticity in these cellular traits in response to different environmental cues scales to responses at the organ level and how this relates to the habitat in which plants evolved have been rarely investigated. We found that two common environmental cues, shading and MS, had significant effects on the balance between horizontal and vertical cell divisions in I. capensis, which contributed to organ-level variation in internode length and thickness. The specific patterns of these responses depended largely on the cellular traits in question and the type of environmental stimuli. Only cell length, shape and alignment responded to both shading and MS. Other traits, such as total cell number, number of cells aligned in the vertical direction and area of individual cells, remained relatively constant across light environments but responded to MS, while cell width and the number of cells around the internode were not affected by MS but responded to shading.
While cell number is widely regarded to be the major determinant of organ size (Johnson & Lenhard, 2011), cell size can theoretically also play an important role. This latter notion is supported by our results. Furthermore, we found a significant negative correlation between cell area and total cell number across genotypes, indicating that variation in internode length is not solely determined by cell number variation. This trade-off proved to be consistent over a wide range of conditions, as it was apparent not only in terms of genotypic variation in characteristics but also in terms of the phenotypic plasticity in these characteristics. That is, genotypes that were highly responsive to environmental stress by changing cell area had a more constant cell number and vice versa. Plants have only a limited amount of resources available to invest in growth, which can be invested in the production of either more or larger structures, leading to a trade-off between cell size and number (Huber et al., 2008; Weijschede et al., 2008; Fujikura et al., 2009; Kawade et al., 2010). Such a trade-off may have important evolutionary consequences, as high levels of phenotypic integration (Murren, 2012), especially negative correlations, have been argued to limit the evolution of plastic responses (Pigliucci, 2003; Gianoli & Palacio-Lopez, 2009).
Different environmental conditions select for different plant morphologies, requiring large flexibility at the level of the basic units. Thus, under shaded conditions, the production of longer stem internodes has been frequently shown to be advantageous (Dudley & Schmitt, 1996; Griffith & Sultan, 2006; Bell & Galloway, 2008; von Wettberg et al., 2008; Huber et al., 2011), while the negative effect of MS can be alleviated by the production of shorter and thicker internodes (Telewski, 1990; Telewski & Pruyn, 1998; Henry & Thomas, 2002). This indicates that plants have evolved a means to change their architecture in response to the environment despite the above-mentioned limitations on the number and size of the cells. In our experiment, internode length was one of the few characteristics that consistently responded plastically to both shading and MS.
Genetic variation in internode length under unstressed (high-light and no-MS) conditions was correlated with both cell number and cell area. We also found that both cellular traits can respond to environmental variation and in turn lead to environmentally induced plastic changes of internode length, a rather plastic trait in Impatiens. The limited response to the environmental manipulations of total cell number seemed to have mainly been compensated by the higher responsiveness of cell area. Theoretically, changes in cell alignment and cell shape may have provided an additional mechanism to further extend internodes either by producing narrower but longer cells or by aligning a greater proportion of cells in the vertical direction. However, though responsive to environmental conditions, cell shape and alignment were not directly associated with internode length or internode length plasticity.
While shading and MS had opposite effects on internode length and diameter and affected the individual cellular traits differently, we found very similar correlation patterns between plasticity of cellular patterns and changes of internode architecture in response to these two stimuli. Independent of the type of stimulus, environmentally induced changes in internode length were correlated with changes in cell area and total cell number, while changes in internode diameter could be related to changes in cell shape and area. This shows that, despite the different natures of these two environmental stimuli (shading but not MS entails an overall reduction of assimilates available to the plant), changes at the cellular level had similar effects on internode architecture, indicating that response patterns were canalized to some extent.
Patterns at the cellular level influence biomechanical characteristics to some extent
The biomechanical properties of an organ can depend on the characteristics of the cells comprising the tissues (Huber et al., 2008; Onoda et al., 2008; Kitajima & Poorter, 2010). In I. capensis, cellular characteristics and shade-induced changes at the cellular level had surprisingly few direct effects on the tissue flexibility. In contrast to the more canalized pattern found for internode architecture, the effects of plasticity of cellular traits on biomechanical characteristics differed between shaded and MS conditions. Plasticity of cellular traits had the strongest effects on tissue flexibility in plants subjected to MS, but these differed between cellular traits. In genotypes with a greater MS-induced reduction in cell area and cell number, there was also a larger increase in tissue flexibility (i.e. lower stem Young's modulus, E). Conversely, MS-induced alignment of more cells in the vertical direction along the internode resulted in reduced tissue flexibility. MS-induced increases in tissue flexibility have been observed in other studies (Anten et al., 2005; Liu et al., 2007). Stem flexibility may enable plants to more easily reconfigure under wind-loading, thus reducing the magnitude of the force to which they are exposed (Puijalon et al., 2008, 2011). Selection can directly act on the underlying cellular traits that facilitate this form of stress avoidance. In contrast to the results for MS-induced plasticity of cellular traits, we did not find a concomitant change of shading-induced plasticity of cellular traits and mechanical stability of plants. This may be explained by the fact that under shaded conditions stem stiffness is of less importance. Under natural conditions, high levels of shading are often caused by conspecific competitors which also act as wind shields, resulting in lower MS (Nagashima & Hikosaka, 2011, 2012). The relatively low variation in the plasticity of whole-plant biomechanical characteristics in response to shading may also indicate that selection has led to canalized responses by minimizing investment in biomechanical stability in shaded plants.
Under high-light, no-MS conditions, total cell number was closely associated with mechanical stability at the whole-plant level, while cell area affected tissue flexibility. Genotypes that produced more cells had a lower BSF (i.e. a reduced ability to carry their own weight) under high-light, no-MS conditions. As increased cell proliferation was mainly invested in the production of longer, but not thicker internodes, internodes of the same diameter had to support longer stems, and thus were associated with a higher risk of mechanical failure. This increased mechanical risk may have selected against genotypes investing preferentially in height growth. However, plants may have also evolved other mechanisms, such as a changed internode structure and efficient use of stem cavity, to gain the required internode strength (Anten et al., 2009).
Cellular characteristics strongly differ in plants originating from different habitats
Differential selection on the response pattern of the integrated phenotype may occur between habitats and populations; these differences can be interpreted as a signature of past evolutionary events leading to different developmental trajectories (Pigliucci, 2002). Plants in contrasting light environments, such as open habitats or forest understories, are typically also exposed to different levels of MS that impose different selective forces on their tissue and structural characteristics. In the forest understory, under both light-limited and wind-protected conditions, selection for stiffer stems provides sufficient self-support at a relatively small resource investment. Conversely, in more open habitats where plants are exposed to both high light and stronger wind forces, the production of more flexible shoots, which reconfigure more easily in the wind and thus reduce drag forces, might be more beneficial (Anten et al., 2009; Puijalon et al., 2011). Our results show that these contrasting selective forces have also led to genetic differences in cellular characteristics.
As discussed previously, both shading and mechanical stimulation of I. capensis had significant effects on cellular traits that contribute to internode length and thickness. However, the strength and direction of these treatment effects varied depending on the habitat from which plants originated. While the overall number of epidermal cells on the internode remained unaffected by treatments, plants originating from the forest understory produced fewer cells than plants originating from grassland sites. Plants originating from different habitats hardly differed with respect to the shape and alignment of cells under high-light conditions but showed distinctly different response patterns to shading. Plants originating from forest understory responded to shading primarily by producing longer cells, while plants originating from grassland responded mainly by aligning a greater proportion of the cells in the vertical direction along the internode.
Interestingly, the response to shading was in the same direction as the habitat differentiation: the forest habitat plants had on average longer cells than plants from grassland habitats. Similarly, plants exposed to MS and plants from the more open and wind-exposed grassland habitats both produced shorter and wider cells. This consistent response to long-term past evolutionary events and present environmental conditions indicates that under sheltered, light-limiting conditions the production of long and narrow cells is favored by selection, while under open, disturbed conditions the production of shorter and wider cells is selected for. The habitat-specific difference in cell number, however, was larger than the differences in cell number induced by the experimental treatments. This supports the notion that cell proliferation is a more stable trait that can be altered by selection processes but not by relatively short-term environmental fluctuations. As total cell number is the product of vertical and horizontal divisions, the relatively low plasticity of total cell number points at a tight control among these different processes. The variation in cell number across habitats may indicate that cell proliferation is associated with fitness consequences, but additional experiments under a more natural setting would be needed to test this hypothesis.
Trade-off in shading- and MS-induced plasticities
We tested the hypothesis that genotypes cannot optimally respond to different cues inducing opposite responses (Henry & Thomas, 2002), leading to a negative correlation in the responses to MS and shading. Our results show a positive correlation for cellular traits in the responses to shading and to MS. Genotypes that reduced their cell area or total cell number in response to MS also produced fewer cells in response to shading (Fig. 6c,g), while genotypes that were more responsive to MS, producing wider and shorter cells, were less able to produce long and narrow cells in response to shading. Interestingly, this apparent positive correlation actually points at a negative functional relationship between the abilities to respond adequately to shading and MS. Previous research has shown that shade avoidance and MS select for opposite morphological responses, that is, the production of thinner internodes under shaded conditions and sturdier internodes under MS. A positive correlation between shade- and MS-induced plasticities in internode thickness thus means that plants that display an adequate response to one environmental cue (i.e. improving the ability to deal with the environmental effect, for example by increasing internode thickness in response to MS) actually displayed a reduced or a ‘wrong’ response to the other environmental cue (e.g. retaining relatively thicker internodes under shaded conditions). The same reasoning can be applied to the various cellular traits. This apparent trade-off may be explained by functional arguments. While strong responses to shade entail mechanical risks and would be detrimental under MS, similarly strong responses to MS come at the cost of competitive ability (Henry & Thomas, 2002). The presence of neighboring plants may also explain this apparent trade-off in the ability to respond to shade or MS. In a dense stand, neighboring plants negatively affect light availability but they also provide wind shielding (Speck, 2003), which may have resulted in a negative correlation in the strength of these contrasting selective forces, thereby impairing the ability of plants to elicit optimal responses to both, MS and shading. Our results not only reveal that different environmental cues select for opposite morphological and cellular responses, they also indicate that at the genotypic level the ability to respond adequately to different cues can be limited as a result of trade-offs in the responsiveness of cellular characteristics to MS or shading. This in turn may constrain genotypes in the responses to a wide range of environmental cues, such as shading and MS, if they select for opposite response patterns.
Plants have only a limited repertoire of possible responses at the cellular level – for example, faster or slower division, elongation in one or more directions, or thicker or thinner walls. Using these responses they have to deal with a variety of environmental changes and with simultaneously acting cues that can elicit opposite responses. With this limited repertoire, using differences in meristem activation patterns, plants can produce a remarkably large range of architectures. Differences in responses between forest and grassland genotypes suggest that there is strong selection pressure on this program of cellular adjustments to varying stress factors, reflecting the large risks involved – for example, being overtopped by neighbors, or succumbing to a sudden MS event.
We thank S. Huggers, S. van Hal, H. Noordman, M. Pawlowski, Liesbeth Pierson, F. Siesling and B. Verduyn for providing technical assistance, and K. Bishop, J. Richards and H. de Kroon for helpful discussion. The original collection and maintenance of genotypes was supported by US NSF DDIG 0408015, and EPA STAR and NIH NRSA fellowships to E.v.W., while support from Florida International University and Fairchild Tropical Botanic Garden facilitated the analysis of cellular data.