Plants in a crowded stand regulate their height growth so as to maintain similar heights to neighbours even when they have potential advantages in height growth

Abstract

Background and Aims

Although being tall is advantageous in light competition, plant height growth is often similar among dominant plants in crowded stands (height convergence). Previous theoretical studies have suggested that plants should not overtop neighbours because greater allocation to supporting tissues is necessary in taller plants, which in turn lowers leaf mass fraction and thus carbon gain. However, this model assumes that a competitor has the same potential of height growth as their neighbours, which does not necessarily account for the fact that height convergence occurs even among individuals with various biomass.

Methods

Stands of individually potted plants of Chenopodium album were established, where target plants were lifted to overtop neighbours or lowered to be overtopped. Lifted plants were expected to keep overtopping because they intercept more light without increased allocation to stems, or to regulate their height to similar levels of neighbours, saving biomass allocation to the supporting organ. Lowered plants were expected to be suppressed due to the low light availability or to increase height growth so as to have similar height to the neighbours.

Key Results

Lifted plants reduced height growth in spite of the fact that they received higher irradiance than others. Lowered plants, on the other hand, increased the rate of stem elongation despite the reduced irradiance. Consequently, lifted and lowered plants converged to the same height. In contrast to the expectation, lifted plants did not increase allocation to leaf mass despite the decreased stem length. Rather, they allocated more biomass to roots, which might contribute to improvement of mechanical stability or water status. It is suggested that decreased leaf mass fraction is not the sole cost of overtopping neighbours. Wind blowing, which may enhance transpiration and drag force, might constrain growth of overtopping plants.

Conclusions

The results show that plants in crowded stands regulate their height growth to maintain similar height to neighbours even when they have potential advantages in height growth. This might contribute to avoidance of stresses caused by wind blowing.

INTRODUCTION

Light competition is critical for survival, growth and reproduction of individuals in dense stands. Plants that could not receive sufficient light stop their growth and often die (Weiner et al., 1990; Nagashima et al., 1995; Matsumoto et al., 2008). As light is a unidirectional resource, taller plants intercept more irradiance and overshade shorter competitors (Schmitt et al., 1995; Dudley and Schmit, 1996; Huber and Wiggermann, 1997; Anten and Hirose, 1998; Huber et al., 1998). It is well known that many herbaceous plants enhance elongation of stems when the stand density or the leaf area index (LAI) is high (e.g. Ballaré et al., 1990; Weiner et al., 1990; Weiner and Thomas, 1992; Weiner and Fisherman, 1994; Nagashima, 1999; Nishimura et al., 2010). This response has been regarded as shade avoidance (Smith, 1982).

Height growth is, however, not always competitive but often apparently co-cperative. In many plant communities, height is similar among plants that reached the top of the canopy (e.g. Koyama and Kira, 1956; Kuroiwa, 1960; Ford, 1975; Hutchings and Barkham, 1976; Weiner and Thomas, 1992; Weiner and Fishman, 1994; Nagashima and Terashima, 1995; Nagashima et al., 1995; Vermeulen et al., 2008), which is called height convergence (Nagashima and Terashima, 1995). Since tall plants receive high light levels, one may consider that being tall is advantageous in light competition. However, plants do not necessarily keep this advantage. Height convergence is often observed among upper plants with different biomass, even though plants with larger biomass may have higher potential for growth (Weiner and Fishman, 1994; Nagashima and Terashima, 1995). Why do such plants not keep overtopping others?

Although being tall is advantageous in light competition, it entails costs for plants. As leaf height increases, plants need to invest biomass more than proportionately in the stem to support their own weight, which in turn reduces the fraction of leaf mass in the plant (Givnish, 1982; Niklas, 1992). Givnish (1982) proposed a game-theoretic model of plant height growth. He assumed that overtopping the other plants is advantageous in light capture and thus leads to a higher production rate per unit leaf mass, but the leaf mass fraction in shoot decreases with increasing plant height. His model predicted existence of the evolutionarily stable height (ESH) in a stand at a given density: no other height (neither taller nor shorter) is advantageous in biomass production in the stand in which plants have the ESH. His model also predicted that the ESH increases with increasing canopy cover. This was well supported by forest forbs (Givnish, 1982). Enhancement of height growth in response to crowding seems to be consistent with the prediction. Height convergence may be a result of evolutionary stable growth of competing plants.

Although the model of Givnish (1982) can provide important insights into plant growth under competition, it is still questionable whether his model fully accounts for the height convergence observed in plant communities. First, his model assumed that competing plants have equal biomass to each other, which is not the same as field situations where plants with different biomass have similar height (Weiner and Fishman, 1994; Nagashima and Terashima, 1995). Secondly, it is unknown whether the constraints assumed in the model are appropriate to explain the plastic height growth of plants found in open habitat. The model assumed that plants have constant mechanical stability and accordingly the biomass allocation to leaves depends only on plant height. There may be, however, another disadvantage in being tall. When plants overtop the neighbouring plants, they may be exposed to stronger wind than the neighbours, which might entail negative effects on plant growth due to excessive transpiration and mechanical stress (Drake et al., 1970; Grace, 1974; Putz et al., 1983). If this is the case, being tall may be more disadvantageous than that considered in Givnish (1982).

In the present study, an experiment was designed to examine whether plants avoid overtopping and whether the behaviour is according to the Givnish model. A stand of potted plants was made and the height of the pot (above or lower than neighbouring pots) was manipulated. The lifting treatment will benefit plants in light capture without any reduction in leaf mass fraction. This is in contrast to the situation of overtopping plants in the Givnish model, where overtopping plants increase light interception but reduce biomass allocation to leaves. The lowering treatment will decrease light availability of the plant despite the same leaf mass fraction as the neighbours. We postulated two hypotheses for response to the treatments. (1) Lifted plants will keep themselves overtopping their neighbours because there is no disadvantage in overtopping. Lowered plants may be suppressed due to the low light availability. (2) Treated plants may regulate their height growth so that their height becomes similar to the neighbours. Lifted plants will decrease the height growth and consequently be caught up by the neighbours. Lowered plants may increase height growth so as to have similar height to the neighbours. Biomass allocation changes caused by the treatments were also investigated. First it is expected, as has been assumed by Givnish (1982), that biomass allocation to leaves depends on stem length irrespective of treatments. If relationship between allocation to leaves and stem length is affected by the treatment, constraints other than that assumed in Givnish (1982) are suggested in the competing plants. The experimental results were also compared with the theoretical prediction using the model of Givnish (1982).

MATERIALS AND METHODS

The experiment was conducted with Chenopodium album, a broad-leaved summer annual, in an experimental garden of Tohoku University, Sendai, Japan (38°25′N, 140°83'E). Monthly mean air temperatures during the experiment were 18·0, 22·5 and 22·9 °C in June, July and August, respectively. Seeds of C. album were obtained from plants in a natural population in 1995. Three hundred and twelve plastic pots, each 12·5 cm in diameter and 20 cm high, filled with a mixture of 7 : 3 vermiculite and Akadama (granular loamy soil 0·5–1 cm in diameter), were tightly arranged on a bench 0·9 m wide, 5·4 m long and 0·5 m high and placed outdoors. On 5 June 1996, about five seeds of C. album were sown in each pot. Pots were watered at 0700–0715 h and 1600–1615 h every day by an automatic watering system. After emergence, each pot was fertilized with 100 mL of nutrient solution of 0·01 : 0·02 : 0·01 NPK every week (10 mg N, 20 mg P and 10 mg K week⁻¹ per pot) from 26 June until 24 July, and from 31 July with 100 mL of nutrient solution of 0·02 : 0·04 : 0·02 NPK (20 mg N, 40 mg P and 20 mg K week⁻¹ per pot). Seedlings were thinned to leave one per pot by 16 July: the plant density was 77 plants m⁻². On 2 August, the stand was surrounded by 80 % shade cloth to the height of the top of the plants to reduce edge effects. The height of the shade cloth was changed according to the growth of the plants. On 3 August, the length and basal diameter of the stems of all plants were measured. The stem length was measured from the base to the terminal shoot apex to the nearest 1 mm, and the basal diameter was measured to the nearest 0·1 mm in the middle of the first internode. Care was taken to allot plants of similar size for 21 target and 126 neighbouring plants. Pots were arranged to have a target plant surrounded by six neighbouring plants as shown in Fig. 1B.

Fig. 1.

Experimental design: treatment of height difference (A) and arrangement of pots (B). Target plants were lifted, lowered, or remained the same level as surrounding plants, as indicated.

The height differences between the target and neighbouring plants were created on 4 August, 60 d after sowing, when the mean plant height was 37 cm and LAI was 2·3. Seven target-plant pots were lifted by 7 cm (lifted plants), another six target-plant pots were lowered by 7 cm (lowered plants), and the other six target-plant pots were kept at the same level as the neighbouring pot (control; Fig. 1A). The height and diameter of target plants were measured in the same way as above. The height of six neighbouring plants was also measured for each target plant. Photosynthetically active radiation (PAR) was measured with quantum sensors (LI-190SA, LI-COR Inc., Nebraska, Lincoln, USA) at the top of the plant and outside the stand. For each plant, PAR was averaged from three measurements.

One and two weeks after the treatment, the height and diameter of target plants and the height of the neighbouring plants were measured. In the experimental period two neighbours were replaced with marginal plants to make the neighbourhood condition of target plants even. After measurements, target plants were harvested and separated into organs (leaf, stem and root), and dried at 80 °C for 3 d and weighed. LAI increased to 2·9 and 4·5 in 1 and 2 weeks after treatments, respectively.

Effects of the treatment on plant growth were analysed by the generalized linear model with a normal error distribution and identity link (JMP statistical software; SAS Institute Inc., Cary, NC, USA). Tukey–Kramer honestly significant difference test was used for post-hoc pairwise comparisons.

Simulation

Optimal leaf height of lifted or lowered plants was simulated according to the model of Givnish (1982). The model assumed single-leaf plants but the model can be extended to multi-leaf plants like C. album. Here photosynthetic production of an invader plant that competes with neighbours for light is considered. Plant photosynthetic production is obtained by photosynthetic production per unit leaf mass (g) multiplied by allocation of shoot mass to leaves [f(s)], where s is stem length, g is a function of the mean height of the leaves of target (h₁) and neighbouring (h₂) plants. When an invader is taller than the neighbours (h₁ > h₂)

(1)

where P_max and P_min are the photosynthetic rate per unit leaf mass of unshaded and shaded leaves, respectively; o is the probability of horizontal overlap of leaves; leaves are distributed on stems randomly within the range of h_i(1 ± σ). When an invader is shorter than the neighbours (h₁ < h₂),

(2)

For an explanation of how the equations are derived see Givnish (1982), but note that eqn 2 is in a different form from the equation given in that paper. Leaf mass fraction, f(s), is given as a function of stem length (s), where a and b are a constant.

(3)

In simulations, 0·374 and 0·043 µmol CO₂ g⁻¹ leaf dry mass s⁻¹ were adopted for P_max and P_min, respectively, which were calculated from the top (1·68 g N m⁻²) and bottom leaves (0·44 g N m⁻²) of dominant individuals in a monospecific stand of Chenopodium album established in the same experimental garden in 1999 (for details, see Hikosaka et al., 2003). Values of a and b were 0·73 and 0·0031, respectively, which were obtained from control plants in the present study (r² = 0·97), σ is 0·48 from the average of control plants, and o was regarded as one as the LAI exceeded one. Simulations were conducted in the case of an invader 7 cm higher or lower than neighbours (simulation A); and an invader whose pot is lifted or lowered by 7 cm (simulation B). In simulation A, height (h) and stem length (s) of an invader are identical values, while in simulation B, h and s are different values from each other. The evolutionarily stable height (h*), which is defined as the height of a stand where no invader with any other height has higher photosynthetic productivity (g × f), is given as follows (Givnish, 1982):

(4)

In simulation B, an optimal height maximizing their photosynthetic production was numerically found out for lifted or lowered invaders.

RESULTS

The PAR at the top of the plants was significantly lower in lowered plants, though the difference was small (P = 0·03, ANOVA). No significant difference was found between lifted and control plants (Table 1). In lifted plants, however, about nine expanded leaves were positioned higher than the canopy surface, while in control plants only three expanded leaves were exposed to the top of the canopy. Thus light interception was increased by the lifting treatment.

Table 1.

Photosynthetically active radiation (PAR) at the top of the plant relative to outside of stand just after the height difference treatment (for control, lifted and lowered plants, see Fig. 1)

	Control	Lifted	Lowered
Relative PAR at the top (%)	88·9 ± 1·0a	88·9 ± 1·4a	85·4 ± 4·3b

Different superscript letters indicate significant difference (P < 0·05) assessed by Tukey–Kramer test (n = 7 in each group).

Control plants exhibited the height growth rate of 23 cm per week during the experiment. Lifted plants slowed stem elongation (Table 2 and Fig. 2A). Two weeks after the onset of the treatment, the difference in apparent height from neighbouring plants became 1 cm (not significantly different from zero) in lifted plants (Fig. 2B). Lowered plants accelerated stem elongation (Table 2 and Fig. 2A), and the difference in apparent height from neighbours was reduced from 7 to 3·5 cm during the experiment (Fig. 2B). However, their height was still lower than that of their neighbours (Fig. 2B).

Table 2.

Stem dimensions, plant dry mass, dry mass partitioning, leaf area, growth parameters of target plants 2 weeks after treatments (see Fig. 1)

	Control	Lifted	Lowered	P-value
Stem length (m)	0·835 ± 0·007b	0·775 ± 0·006c	0·870 ± 0·008a	<0·0001
Height increment (m)	0·459 ± 0·008b	0·400 ± 0·006c	0·496 ± 0·007a	<0·0001
Diameter (mm)	4·31 ± 0·06	4·40 ± 0·08	4·21 ± 0·07	0·161
Diameter increment (mm)	0·49 ± 0·03b	0·66 ± 0·04a	0·47 ± 0·04b	0·0015
Length/diameter (m m−1)	194 ± 3b	176 ± 3c	207 ± 3a	<0·0001
Total mass (g)	6·68 ± 0·24	6·50 ± 0·14	5·99 ± 0·19	0·0348
Root mass (g)	1·71 ± 0·10ab	1·76 ± 0·06a	1·44 ± 0·10b	0·0210
Stem mass (g)	2·65 ± 0·10	2·50 ± 0·06	2·42 ± 0·06	0·0815
Leaf mass (g)	2·33 ± 0·07	2·24 ± 0·06	2·13 ± 0·07	0·107
Root/total mass (g g−1)	0·255 ± 0·007ab	0·271 ± 0·006a	0·240 ± 0·010b	0·0265
Stem/total mass (g g−1)	0·396 ± 0·006	0·385 ± 0·007	0·405 ± 0·006	0·0964
Leaf/total mass (g g−1)	0·349 ± 0·003	0·344 ± 0·003	0·355 ± 0·008	0·232
Lamina area (cm2)	602 ± 26	566 ± 14	613 ± 23	0·231
SLA (m2 kg−1)	28·9 ± 0·7b	28·4 ± 0·7b	32·4 ± 0·8a	<0·0001
LAR (m2 kg−1)	9·01 ± 0·22b	8·70 ± 0·15b	10·25 ± 0·20a	<0·0001

Values are means ± s.e. Different superscript letters indicate significant differences (P < 0·05) among treatments according to Tukey–Kramer test (n = 7 in each group).

P-values are the result of generalized linear model analyses (d.f. = 2). Significnat values (P < 0·05) are highlighted in bold.

SLA, specific leaf area (lamina area/lamina mass); LAR, leaf area ratio (lamina area/total mass).

Fig. 2.

Growth of stem length of target plants (A) and difference in height of the top of plants between target and neighbouring plants that was obtained as [height of a target plant (+7 cm, for lifted plants; –7 cm, for lowered plants)] – (mean height of six adjacent plants) (B), after pot-elevation treatment (see Fig. 1). Different letters indicate significant differences (P < 0·05) among treatments according to Tukey–Kramer test (n = 7 in each group). Bars represent ± s.e.

Total biomass was significantly different among treatments (Table 2). However, the biomass of lifted plants was not greater than that of control plants (Table 2), even though lifted plants might have received greater irradiance. Lowered plants had smaller biomass than the other two (Table 2).

There was no significant effect of the treatment on leaf mass and leaf mass fraction (leaf/total mass) despite the differing stem lengths among the treatments (Table 2). The treatment significantly affected the root mass and root mass fraction (Table 2). Stem mass and the stem mass fraction showed a marginally significant difference (Table 2). These results suggest that there were changes in allocation between roots and stems across the treatments: lifted and lowered plants tended to invest more biomass in roots and stems, respectively. Specific leaf area (SLA) increased more in lowered plants than in the others, resulting in a greater leaf area ratio (LAR) in lowered plants (Table 2). Lifted plants slowed stem elongation but increased stem thickness enlargement, while lowered plants accelerated elongation but tended to reduce diameter growth (Table 2).

Table 3 summarizes the simulation results. Here it is assumed that neighbours have an evolutionarily stable height (h*) (eqn 4). When a stand is composed of pots at the same height level (simulation A; no lifting or lowering), a plant taller or shorter than h* has lower photosynthetic production than the neighbours with h*. This is because, for example, in a higher plant the positive effect by the increase in g was smaller than the negative effect by the decrease in f. In simulation B, on the other hand, a lifted or lowered plant has the same stem length as neighbours, but is 7 cm taller or shorter, respectively, than the neighbours. A lifted plant has greater photosynthetic production than the neighbours because they increase g without a decrease in f. If the lifted plants decrease their stem length and increase f, photosynthetic production further increases. Photosynthetic production of the lifted plants is maximized when they are 2 cm higher than the neighbours, suggesting that keeping 2 cm taller than neighbours is advantageous for a lifted plant. A lowered plant in simulation B produces less photosynthetically than the neighbours irrespective of stem length. Photosynthetic production of the lower plant is maximized when the height is 4 cm lower than the neighbours.

Table 3.

Simulation results

Height of invader plant (cm)	Photosynthetic production (mmol CO2 g−1 d−1)
Simulation A
Unchanged pot height
h*	2·333
h* + 7	2·307
h* – 7	2·317
Simulation B
7 cm lifted
h*	2·532
h* + 7	2·521
h* + 2 (optimal)	2·533
7 cm lowered
h*	2·135
h* – 7	2·136
h* – 4 (optimal)	2·140

Photosynthetic production per above-ground biomass is calculated as the product of photosynthetic production per leaf mass (g) and leaf mass fraction (f). In every simulation, height of neighbouring plants is an evolutionarily stable one (h* = 153 cm). In simulation A, the height of an invader plant is altered from the neighbours with an identical pot height to the neighbours. ‘h* ± 7′ means that the height of an invader is 7 cm higher/lower than h*. In simulation B, the pot height of an invader plant is lifted up or lowered by 7 cm compared with neighbours. ‘h* + 7′ in ‘7 cm lifted’ in simulation B means that the height of an invader is 7 cm higher than h* due to pot elevation but the stem length is the same as the neighbours. ‘h* + 2′ means that height and stem length are 2 cm higher and 5 cm lower than the neighbours and the invader that has this height has a higher photosynthetic production than those at other heights. Similarly, ‘7 cm lowered’ means that the pot height of an invader is 7 cm lower than the neighbours.

DISCUSSION

The present study showed that plants in crowded stands regulate their height growth so as to maintain similar height to neighbours: plants whose top had been lowered relative to their neighbours accelerated the elongation rate while plants whose top had been lifted higher than the neighbours reduced the elongation rate (Fig. 2).

Lifted plants did not keep themselves overtopping neighbours, even though they had two advantages at the onset of experiment: increased light interception and less cost of tissues to support leaves at higher positions. At the end of the experiment, the height difference between lifted and the neighbour plants became 1 cm (Table 2). The value was similar to that expected from simulation B (Table 3). However, growth was quite different from that expected from the Givnish model. First, fraction of leaves in the plant mass was not different among the treatments despite the different stem lengths (Table 2). This result suggests that the cost of competition may not be fully accounted for by a reduction in leaf mass fraction with an increase of plant height (stem mass fraction). Secondly, lifted plants did not increase growth, even though they had received more light than their neighbours (Table 2). This suggests that overtopping is not advantageous for photosynthetic rate per leaf mass. Some other costs or constraints therefore need to be incorporated for understanding height growth regulation in crowded stands.

What was the cost or constraint of overtopping, then? Constraint of being tall has been considered mainly based on safety against buckling: a stem bends due to its own and leaf weight (e.g. McMahon, 1973; King, 1986; Niklas, 1993a, b, 1994a, b; Casal et al., 1994; King et al., 2009). However, buckling safety does not necessarily guarantee safety against breaking damage caused by wind blowing. Because wind speeds increase dramatically above the boundary layer of the canopy (Goudriaan, 1977; Speck, 2003), overtopped plants may be exposed to stronger winds than the neighbours. As wind speed increases, the risk of mechanical failure increases (Niklas, 2000). During the present experiment, the maximum instantaneous wind speed in the day was 7·3–25·4 m s⁻¹ in Sendai (Japan Meteorological Agency). In a potted plant of an erect annual, Xanthium canadense, wind speed of 20 m s⁻¹ bent the stem by 50–60 ° (H. Nagashima, Tohoku University, Sendai, Japan, unpubl. res.). Thus lifted plants in the present study might have suffered from wind blowing and changed biomass allocation and morphology to avoid mechanical failure. In fact, lifted plants reduced the height : diameter ratio of the stem and increased the root mass fraction (Table 2), both of which might have contributed to mechanical stability (Niklas, 1992; Henry and Thomas, 2002; Anten et al., 2005, 2006, 2009).

Another constraint may be transpiration. It is notable that lifted plants did not have greater biomass than control plants (Table 2), though the intercepted light might have been greater. This suggests that light use efficiency in photosynthesis (carbon gain per unit absorbed light; Hikosaka et al., 1999) was lower in lifted plants. Grace (1974) showed that an increase in wind speed from 1 to 3·5 m s⁻¹ enhanced transpiration and reduced water content of leaves in Festuca arudinacea. It is probable that wind induced stomatal closure to prevent water loss (Drake et al., 1970), leading to a reduction in photosynthesis rates. Lifted plants increased the root mass fraction (Table 2), which is also in accord with general responses to water stress (Watts et al., 1981).

Lowered plants, on the other hand, accelerated height growth, nearly reaching the canopy top (Fig. 2). The height difference was –3·5 cm, which was close to the prediction by the Givnish model (Table 3). Enhanced height growth was realized at the expense of stem diameter growth, leading to a higher height : diameter ratio of the stems (Table 2). This change in stem morphology is in accord with the previous studies in which subordinate plants were more slender than dominant and solitary plants (Weiner et al., 1990; Weiner and Thomas, 1992; Weiner and Fishman, 1994; Nagashima and Terashima, 1995). Biomass allocation was also changed (Table 2); lowered plants tended to increase stem mass at the expense of root as compared with the other plants (Table 2). This also agrees with previous observations that plants allocate fewer resources to roots when they are shaded or in a dense stand compared with plants that are exposed or solitary (e.g. Corré, 1983a, b; Maliakal et al., 1999; Nishimura et al., 2010). These results imply that accelerating height growth is costly: reduced diameter growth may increase a risk of mechanical failure of plants and reduced root mass fraction may be disadvantageous for acquisition of below-ground resources as well as lower mechanical stability. However, the elongation may benefit the plants by carbon gain, since wind speed is lower than it is on the outside of the canopy (Goudriaan, 1977; Speck, 2003).

In the model of Givnish (1982), total biomass of plants was assumed to be identical among individuals for simplification. The model showed that, if a stand consists of individuals with evolutionarily stable height (ESH), plants at another height cannot have higher biomass production rates. However, this does not necessarily hold when the stand consists of individuals with different total biomass. If an individual has a greater biomass than neighbours, it can have a higher production rate even though its height is greater or shorter than the ESH of the neighbours (Table 3). In other words, the Givnish model may not explain why height convergence is often observed even among upper plants with different biomass (Weiner and Fishman, 1994; Nagashima and Terashima, 1995). The present results clearly suggest that overtopping others is not necessarily advantageous even when the individual has a greater potential for height growth. Plants keep their height similar to their neighbours to avoid constraints that arise by overtopping others.

Convergence of height growth leads to J-shaped or bimodal frequency distribution of plant height (J-shaped: the population consists of many similarly tall plants with a smaller number of shorter plants) (Koyama and Kira, 1956; Kuroiwa, 1960; Ford, 1975; Hutchings and Barkham, 1976; Weiner and Fishman, 1994; Nagashima and Terashima, 1995; Nagashima et al., 1995). Height convergence also accounts for a non-linear relationship between log-transformed height and diameter, which has commonly been observed in various plant communities; height is positively correlated with diameter among subordinate short plants, while height is relatively similar in dominant tall plants irrespective of diameter (Ogawa et al., 1965; Assmann, 1970; Kira, 1978; Rai, 1979; Kohyama et al., 1990; Niklas, 1995; Aiba and Kohyama, 1996; Thomas, 1996; Sterck and Bongers, 1998; Weiner and Thomas, 1992; Weiner and Fishman, 1994; Nagashima and Terashima, 1995). Subordinate individuals tend to increase their height, while dominant plants suppress their height growth to keep their height similar to their neighbours even though they have a greater potential of height growth. Such an apparently co-cperative growth of height may make plant–plant interactions less competitive, which allows coexistence of dominant plants. Nagashima et al. (1995) earlier showed in C. album stands that the number of upper plants that reach the canopy top was similar irrespective of initial density, which may partly result from co-cperative growth of the upper plants.

Several environmental cues have been suggested for the regulation of height growth. In particular, the red : far-red (R/FR) ratio in incident light affects stem elongation (for review, see Ballaré, 1999, 2009; Smith, 2000). Height convergence was impeded by treatments to reduce the R/FR effect (Ballaré et al. 1994; Aphalo and Rikala 2006). On the other hand, mechanical stimuli such as swaying due to wind may also affect stem growth (Biro et al., 1980; Telewski, 1990; Jaffe and Forbes, 1993; Henry and Thomas, 2002; Anten et al., 2005, 2006, 2009). Both factors can account for our results: lifted plants might have received a higher R/FR ratio in incident light and stronger winds than others, leading to a suppression of stem growth. This will be analysed in a forthcoming paper.

In summary, it was found that plants in crowded stands regulate their height growth to maintain a similar height to their neighbours even when there are potential advantages to height growth. Lifted plants increased neither biomass allocation to leaves nor biomass production, indicating that overtopping neighbours is not necessarily beneficial in competing plants. There may be other constraints for overtopping. A likely constraint is wind blowing, which may reduce mechanical stability and/or cause a deterioration in the water status of the plants.