Influence of mechanical stress on biomass allocation in three species (Pachira aquatica Aubl., Sextonia rubra (Mez) van der Werff and Simarouba amara Aubl.) with contrasting posture control mechanisms

Abstract

Key message

The posture control in Pachira aquatica, Sextonia rubra and Simarouba amara is achieved through contrasted mechanisms involving tensile stress in wood only, bark only or both. This study evidenced that the restoration of verticality does not imply an overall cost for trees but modifies growth kinetics and biomass allocation to organs and tissues, improving the posture control of the trees in the three species.

Context

All trees need a motor system to correct their position through the generation of asymmetric tensile stress around the stem, leading to active bending. In angiosperms, depending on the species, tensile stress is generated either in wood (tension wood), in bark or in both.

Aims

Here, we investigated how gravitropic stimuli (tilted stems without any movement) may affect growth and biomass allocation and whether this process depends on the posture control mechanism of the species.

Methods

Tree growth kinetics, final biomass allocation and wood and bark proportion, localisation and density were measured on young tilted plants and straight plants as controls. Pachira aquatica, Sextonia rubra and Simarouba amara were selected according to the location of their their motor system within bark only, wood only or both wood and bark, respectively.

Results

In response to tilting, trees from the three species increased their diameter and decreased their slenderness, but the total biomass (including stem and roots) was not different from that in the control trees, suggesting that reaction to artificial tilting does not imply a specific cost for the plant. However, the species exhibited strong differences in growth kinetics, in the amount and organisation of the tissues or in biomass allocation to different organs (root vs shoot, wood vs bark), adapted to the specificity of the posture control mechanisms and improving their motor function.

Conclusion

Whatever the posture control mechanism, uprighting does not modify the total biomass invested. However, allocation of biomass to different organs is strongly modified to obtain an efficient control of the posture.

1 Introduction

The lifespan of trees is strongly dependent on their ability to remain stable in an erect position. Trees must struggle against dynamic external loads such as wind, animals, falling trees or debris as well as static load such as self-weight under gravity. When trees are growing on sloping ground or when they have been tilted by a landslide, they need to restore verticality. Even their own growth and thereby increased self-load tends to alter their vertical position by bending their stem downwards. To react to these disturbances, trees need to adjust their structure and require a motor system (Moulia et al. 2006) to maintain or adjust their position. These adaptative growth responses imply adjustment of carbon allocation to the different organs and induce belowground and aboveground morphological modifications in the size, symmetry and distribution of the tree organs toward an optimal design improving anchorage and stability (Wang et al. 2023). Numerous studies (e.g. Telewski and Pruyn 1998; Cordero 1999; Coutand et al. 2008, Bonnesoeur et al. 2016, Roignant et al. 2018 or detailed review by Wang et al. 2023) evidenced that mechanical perturbation decreased height growth and increased radial growth and allocation to the roots compared to that of undisturbed trees.

In this study, we aimed at describing the changes in tree morphology and biomass allocation to different organs while trees react to an artificial tilting. In these conditions, trees act to restore verticality (Yoshida et al. 2000). The principle of the uprighting is always based on an asymmetric repartition of stress around the stem, the upper side of the tilted stem being more tensile than the lower side, producing an upward bending moment of the stem (Alméras and Fournier 2009).

In most studied tropical species, and probably in most temperate species as well, the motor system relies on wood, whether associated with bark or not (Ghislain et al. 2019a). In these species, strong modifications of the fibre cell wall occur during the maturation process, allowing the generation of high tensile stress in the so-called G-layer in tension wood on the upper side of the tilted stem (Yoshida et al. 2000). This generates an asymmetric axial stress on both sides of the stem (Fig. 1) that is responsible for a bending moment able to curve the stem upwards (Alméras et al. 2005). When the motor system relies only (or mostly) on wood (Fig. 1C), like in Sextonia rubra (Mez) Van der Werff, the bark restricts the uprighting action of wood (Ghislain et al. 2019a). When the motor system relies mostly or only on bark, like in Pachira aquatica Aubl., the wood on the upper side of the leaning stem is characterised by a low fibre content and radially elongated parenchyma cells (Ghislain et al. 2019b). In these species, tensile stress is very low in wood but high in bark (Fig. 1A). Longitudinal tensile stress originates from its interaction between wood and bark. Wood radial growth increases the stem perimeter leading to stretch the bark in the tangential tension. This tangential strain is redirected into longitudinal strain due to the high Poisson’s ratio allowed by the trellis organisation of the stiff phloem fibres (Clair et al. 2019; Alméras et al. 2024). Asymmetric radial growth of wood allows the production of asymmetric longitudinal stress between the upper and lower bark and generates a strong bending moment able to curve the stem of saplings upwards. The presence of fibres with a G-layer in the phloem (Nanko et al. 1982; Nakagawa et al. 2012) may even increase tensile stress in the bark (Lehnebach et al. 2020b). In some species, such as Simarouba amara Aubl., the motor system relies on both wood and bark (Clair et al. 2019). Wood is under high tensile stress, and bark is under sufficient tensile stress to accompany the action of wood (Fig. 1B).

Fig. 1

Diagram of the posture control mechanisms. A largerr tensile stress is produced on the upper side than on the lower side of the tilted stem. Tension is generated either in the bark only (P. aquatica), in wood and bark (Si. amara) or in the wood only (Se. rubra). Drawing and anatomical sections showing eccentric growth and wood and bark thicknesses on both sides of the stems. Arrows: tensile stress. Schematic disc presents the diameters parallel (D_a) and perpendicular (D_b) to inclination as well as the bark thickness on the upper side (BTH_US) and the lower side (BTH_LS)

These types or combinations of motor systems (active wood only, active bark only or combinations of active wood and active bark) result in similar gravitropic responses (Ghislain et al. 2019a). However, as these mechanisms involve a different contribution of wood and bark, the investment in these tissues is expected to be contrasted in order to improve the efficiency of each mechanism. This tissue investment at the stem level may differ quantitatively (amount of bark compared to wood) or qualitatively (denser wood or bark allowing stiffer material) or in its distribution around and along the stem (eccentric secondary growth, slender or stocky stem). Species with posture control mechanism relying on wood only are expected to increase wood growth, especially on the upper side of the tilted stem, decrease bark growth and modify density to reach a stiffer tension wood and softer bark. Reversely, when the mechanism relies on bark only, an overgrowth of both wood and bark is expected (Ghislain et al. 2019a) with a denser bark (expected to be stiffer) and less dense wood to increase the efficiency of bark in bending the stem. When posture control relies on both wood and bark, a trade-off is expected between both. At the plant level, all species are expected to reduce their slenderness but different mechanisms may lead to different trade-offs in terms of growth kinetics and may affect the total plant biomass or the allocation to different the different organs or tissues.

The three selected species and their posture control mechanisms are as follows: P. aquatica displays an active bark; Si. amara displays both active tension wood and bark, both contributing to the uprighting of the stem; and Se. rubra displays active tension wood (Ghislain et al. 2019a). To avoid the effect of wind and to study only the effect of the type or combination of motor systems, seedlings were grown in a greenhouse with a constant tilting maintained all along the experiment, avoiding any movement of either the stem or the leaves.

This study aimed to address the following questions: (1) Does the reaction of a young stem to artificial tilting require an overall biomass investment from the plant and does this investment differ among these 3 species with different posture control mechanisms? (2) Does the reaction modify the biomass allocation within the different organs of the plant (root, stem, or leaves)? (3) How does the reaction modify the wood and bark quantity and quality (density) around the stem?

2 Material and methods

2.1 Plant material

The study was carried out on saplings of three cooccurring lowland tropical tree species from the Guiana Shield in French Guiana with different posture control mechanisms (Fig. 1, Table 1). All three species are evergreen, P. aquatica and Si. amara are heliophilic, while Se. rubra is shade-tolerant. Se. rubra and Si. amara were collected as seedlings at the Paracou experimental station (5° 16′ 26″ N, 52° 55′ 26″ W) near Sinnamary and at the agronomic campus of the Research Unit EcoFoG in Kourou (5° 10′ 22″ N, 52° 39′ 18″ W). P. aquatica plants were collected as seeds along the Kourou River (5° 00′ 47″ N, 52° 41′ 54″ W). The plants were grown in a greenhouse at the agronomic campus in Kourou, equipped with a shade cloth transmitting 10.8% of the total irradiance. When the saplings were 1 year old, they were transplanted into 20 l pots with a mixture of 1 to 2 volumes of sand and forest soil and fertilised once at the beginning of the experiment. A wind shelter prevented the effect of wind during the experiment. Water was automatically supplied twice a day to avoid any water deficit. All saplings were staked and acclimatised in the greenhouse for at least 1 month. Two treatments were applied to each species: half of the plants were kept straight and fastened to the stakes as a control (C), and the other half was fastened to the stakes and tilted at 45° (T). A total of 25 to 29 saplings per species and treatment were selected among the larger samples based on height and diameter uniformity. The position of the pots was randomised to avoid position effects within the greenhouse. Table 1 summarises the growth monitoring periods and harvesting dates.

Table 1 Description of the plant material used for this study. N_control, number of individuals used as a control; N_tilted, number of plants used for tilting; N_tot, total number of plants; Growth monitoring period; Harvesting time; D₀, mean diameter at tilting day (± standard deviation); H₀, mean height at tilting day (± standard deviation); posture control mechanism

2.2 Growth monitoring

A subsample of 15 plants per treatment and per species was chosen. Growth monitoring started 1 to 2 weeks before the tilting date, which was considered the initial date (t₀) for each descriptor. Growth was monitored approximately every 2 weeks on this subsample until the stem diameter increased by at least 70% in tilted trees.

The following indicators of growth were recorded.

Height (H, mm). A mark was made with a pen on the bark at approximately 5 to 10 cm from the soil, and the height was precisely measured only once at the beginning of the experiments to avoid errors linked to the change in ground level during the period (e.g. soil compaction or eventual erosion in tilted pots). The following steps were performed only from the mark to the stem apex, and the basal length was subsequently recorded. The relative height increment (H_{Rel. Incr}, mm.mm⁻¹) at time t was calculated as the difference in height at time t (H_t) minus the height at tilting date t₀ (H₀) divided by the height at t₀. H_{Rel. Incr} = (H_t – H₀)/H₀.

Number of new leaves. Newly grown leaves were recorded at each measurement date.

Diameter (D, mm). The basal diameter was measured both parallel (D_a) and perpendicular to the pole (D_b; i.e. to the tilted direction for tilted saplings, Fig. 1). The relative diameter increment (D_{Rel. Incr}, -) at time t was calculated as the change in the mean diameter (mean of D_a and D_b) divided by the mean diameter at t₀: D_Rel.Incr. = (D_t − D₀)/D₀.

The ovalisation index (Ov, -) was calculated at each time as the ratio between the diameter parallel to tilting (D_a) and the diameter perpendicular to tilting (D_b): Ov = D_a/D_b. The ovalisation index is equal to 1 when the stem is circular and greater than 1 when the stem is oval, with the greatest diameter occurring in the tilting direction.

Slenderness (S, -) was calculated at each time t as the ratio of height (H_t) to mean diameter (D_t): S_t = H_t/(D_t*0.1). The relative slenderness (RS_t, -) at time t was calculated as RS_t = (S_t – S₀)/S₀.

2.3 Description of the plants at harvest

Saplings were harvested at the end of the monitoring period for P. aquatica. A longer period of growth was applied before harvesting for Si. amara and Se. rubra, as the lignification process in tension wood is known to occur very late after cell wall formation (Roussel and Clair 2015) and is expected to contribute to the stem biomass in tilted trees. Roots, stems, and leaves were collected separately, with distinctions made between leaves produced before the tilting date and leaves produced after the tilting date. For Si. amara and P. aquatica, leaflets were separated from the rachis and the petioles. Visual assessment of the root system confirmed that all the species still had sufficient space for root development in the pots at the harvesting date. The roots were carefully washed by hand.

2.4 Plant biomass allocation

Roots and stems were dried at 103 °C for 48 h and subsequently weighed. Leaves or leaflets, rachises and petioles were dried at 68 °C for 48 h and weighed. A subsample of leaves per species was dried at 103 °C for 48 h. The mass ratio between 103 °C and 68 °C was 97.8% for Se. rubra, 97.2% for Si. amara and 97.9% for P. aquatica. This test allows us to conclude that the error made by any of the methods is less than a few percent. We used dry mass at 68 °C for all leaves and leaflets in this study.

Biomass allocation to each of the 3 compartments (root, stem, and leaf) was expressed as the dry mass (M, g) of the given compartment divided by the total biomass (M_tot, g) of the plant:

$$\begin{array}{lll} \text{Total biomass:} & \text{M}_{\text{tot}} = \text{M}_{\text{root}} + \text{M}_{\text{stem}} + \text{M}_{\text{leaves}} & (\text{g})\\ \text{Root mass ration:} & \text{RMR} = \text{M}_{\text{root}}/\text{M}_{\text{tot}} & (-)\\ \text{Stem mass ration:} & \text{SMR} = \text{M}_{\text{stem}}/\text{M}_{\text{tot}} & (-)\\ \text{Leaf mass ration:} & \text{LMR} = \text{M}_{\text{leaves}}/\text{M}_{\text{tot}} & (-)\\ \text{Specific stem length:} & \text{SSL} = \text{stem length before drying/M}_{\text{stem}} & (\text{mm/g}) \end{array}$$

A similar ratio of the whole shoot was computed to compare aboveground and belowground biomass:

$$\begin{array}{ccc}\mathrm{Shoot}\;\mathrm{mass}\;\mathrm{ratio}:&\mathrm{ShMR}\;=\;({\mathrm M}_{\mathrm{stem}}\;+\;{\mathrm M}_{\mathrm{leaves}})/{\mathrm M}_{\mathrm{tot}}&\left(-\right)\end{array}$$

Leaf area (cm²) is the total leaf area of a sapling. Leaves of Se. rubra and leaflets of Si. amara and P. aquatica were measured using a LI-COR 3100 area metre (LI-COR, Lincoln, Nebraska, USA) and were subsequently standardised according to the supporting biomass as follows:

$$\begin{array}{ccc} \text{Leaf area ratio}\;(\text{LAR},\text{m}^{2}.\text{kg}^{-1}):\;&\text{LAR}\;=\;\text{Leaf area}/\text{M}_{\text{tot}}&\;(\mathrm{m}^{2}.\mathrm{kg}^{-1})\end{array}$$

The specific leaf area (SLA, m² kg⁻¹) of the newly formed leaves (after t₀) was measured as follows:

$$\begin{array}{cc}\mathrm{SLA}\;=\mathrm{New}\;\mathrm{leaf}\;\mathrm{area}/\mathrm{New}\;\mathrm{leaf}\;\mathrm{mass}&(\mathrm m^2.\mathrm{kg}^{-1})\end{array}$$

The wood and bark specific gravity (WSG and BSG, g.cm⁻³) was measured as the ratio of the oven-dried mass (M_103°C) to the fresh volume (V_green). SG = M_103°C/V_green. Measurements were performed on both wood and bark isolated from a 2-cm portion of the stem sampled at approximately 10 cm from the soil. Wood samples included the pith. The sample volume was calculated using the inverse Archimedes principle method on a Sartorius CP224S balance (precision: 0.2 mg).

Cross-sectional morphometric measurements. A photograph of the green section was taken for all saplings at approximately 8 cm from the soil. The total section area (A_tot, mm²), bark area (A_B, mm²), total wood area including the pith (A_w, mm²), bark thickness on the upper side (BTh_US, mm) and bark thickness on the lower side (BTh_LS, mm) along the pole direction (Fig. 1) were measured on the images using ImageJ (Schneider et al. 2012). The tissues were described as follows:

$$\begin{array}{lll}\mathrm{Bark}\;\mathrm{area}\;\mathrm{ratio}:&\mathrm{BAR}\;=\;{\mathrm A}_{\mathrm B}/{\mathrm A}_{\mathrm{tot}}\;&\left(-\right)\\\mathrm{Bark}\;\mathrm{thickness}\;\mathrm{ratio}:&\mathrm{BThR}={\mathrm{BTh}}_{\mathrm{US}}/{\mathrm{BTh}}_{\mathrm{LS}}&\left(-\right)\end{array}$$

2.5 Statistical analyses

Statistical analysis was performed using R (R Core Team 2020). The normality of the distribution of the parameters and the homogeneity of their variance were checked with the package stats and lawstat. Student’s t test or the Mann–Whitney test was used to compare treatments. Student’s t test was performed with package stats to test the difference with a theoretical mean of 1 for ovalisation index.

3 Results

3.1 Growth kinetics

After 14 to 15 weeks in the tilted position, sapling growth was strongly affected compared to that in the straight control trees (Fig. 2, Table 2). The relative diameter increment (D_{Rel. Incr}) was larger in the tilted than in the control saplings. This difference in secondary growth was already evidenced after 3 to 4 weeks for the 3 species (Fig. 2).

Fig. 2

Growth kinetics of P. aquatica (Pa), Si. amara (Sa) and Se. rubra (Sr). Changes during the treatment of the relative diameter-increment, ovalisation index, relative height increment, relative slenderness and number of newly grown leaves. The tilting day is t₀ = 0. The black lines represent tilted trees, and the grey lines represent straight trees. Error bars represent the standard error

Table 2 Mean value and standard deviation per species and per treatment of the parameters measured for growth at weeks 14–15, growth at harvest, plant description at harvest. Column P indicates the results of either Student’s t test or the Mann–Whitney test for comparison of two treatments. For the bark area ratio and bark thickness ratio, Student’s t test was used for comparison with the theoretical mean of 1. *P < 0.05; **P < 0.01; ***P < 0.001; NS: not significant

The stems of the control saplings of all 3 species were circular. In tilted saplings of P. aquatica, strong ovalisation of the stems occurred after 3 weeks of tilting then ovalisation remained stable throughout the experiment, whereas the stems of the control trees remained circular (ovalisation index of 1.11 ± 0.06 for tilted saplings at harvest; Table 2). In tilted Si. amara, ovalisation of the stems was gradual and reached similar values to P. aquatica after 8 to 10 weeks. In Se. rubra, the stems of both the tilted and control trees remained circular throughout the experiment (no difference with the theoretical value of 1) (Fig. 1).

The relative height increase (H_{rel_incr}) was lower in the tilted saplings than in the control for P. aquatica and Si. amara saplings beginning at weeks 3 and 2, respectively (Fig. 2 and Table 2). This difference was exacerbated in P. aquatica, in which H_{rel_incr} was barely 7% in tilted trees and 39% in control trees at week 14 (Table 2). In Si. amara, H_{rel_incr} was 43% in the control trees and 27% in the tilted trees at week 14. In Se. rubra, height growth was also lower for tilted trees than for control trees, but the difference in H_{rel_incr} between treatments was significant only at weeks 3 and 11. In this species, H_{rel_incr} was 27% in the control trees and 20% in the tilted trees at week 14.

After 2–3 weeks, the relative slenderness (RS_t) was lower in the tilted saplings than in the control saplings for all 3 species (Fig. 2).

Tilted saplings produced fewer new leaves than did the control saplings after week 3 in P. aquatica and only at weeks 4 and 6 in Si. amara. The number of new leaves on Se. rubra did not differ between the tilted and control saplings (Fig. 2 and Table 2). In tilted saplings of P. aquatica, the small number of new leaves is linked to the low relative height increase, as this species is characterised by rhythmic growth.

3.2 Biomass allocation between roots, stems and leaves

For all three species, neither the total biomass produced nor the specific leaf area of the new leaves (SLA) were different between tilted and control trees (Table 2). Compensation between compartments resulted in stable total biomass in P. aquatica and Se. rubra. Neither total biomass nor allocation to the different compartments differed in Si. amara (Table 2 and Fig. 3). In P. aquatica, leaf mass ratio (LMR) was lower and the root mass ratio (RMR) was larger for tilted trees. Despite differences in stem height, the stem mass ratio (SMR) and specific stem length (SSL) did not differ between treatments (Table 2 and Fig. 3). Leaf area ratio (LAR) was lower for tilted trees. In Si. amara, the biomass of roots, stems and leaves did not differ between tilted and control trees. However, SSL was lower in the tilted trees (91 ± 22 mm.g⁻¹ compared to 121 ± 24 mm.g⁻¹ in controls). No difference in LAR was detected between treatments (Table 2 and Fig. 3). Finally, in Se. rubra, LMR was also lower in tilted trees, but RMR was not different. Changes in the biomass allocated to leaves were compensated by an increase in stem mass. SSL was lower in the tilted trees (139 ± 96 mm.g⁻¹ compared to 202 ± 98 mm.g⁻¹ in the controls). LAR was lower for the tilted trees (Table 2 and Fig. 3).

Fig. 3

Diagram of the differences in the biomass allocation ratios between roots, stems and leaves of tilted trees and controls. Grey: no change. White: The biomass of the tilted trees was lower than that of the controls. Black: biomass was larger in the tilted than in the control trees

3.3 Wood and bark proportion, localisation and density

A schematic overview of the changes in wood and bark specific gravity and size and shape of the sections of tilted and control trees during growth (Fig. 4) summarised the following results.

Fig. 4

Schematic representation of the change in size and shape of the sections of tilted and control trees during growth. The colour level indicates the wood and bark specific gravity (g.cm⁻³). Pa: P. aquatica; Sa: Si. amara; Sr: Se. rubra

Wood specific gravity (WSG) of P. aquatica was lower in the tilted saplings than in the control saplings, whereas no difference was found in the specific gravity of bark (BSG) between the treatments (Table 2, Fig. 4). Wood and bark areas were not different between control and tilted trees, but the bark was thicker on the upper side of the tilted saplings, whereas bark thickness remained constant around the stem in control trees. On average, the upper bark was more than twice the thickness of the lower bark on the tilted trees. Similarly, bark area ratio (BAR) was lower for tilted trees, suggesting that for a constant stem area (see final diameter, Table 2) between treatments, wood growth is favoured over bark growth for tilted stems.

In Si. amara, WSG was larger for tilted trees, whereas BSG was lower. The bark thickness ratio was greater for the tilted trees than for the untilted trees, indicating a strong reaction on the upper side compared to the lower side. On tilted trees, the upper bark was on average more than 1.5 thicker than the lower bark. Both the wood and bark areas were greater for tilted trees, and the bark area ratio did not differ between the control and tilted saplings, suggesting that the increase in stem area observed in tilted trees (see final diameter, Table 2) was proportional in wood and bark.

In Se. rubra, the tilted trees did not have greater WSG (P = 0.063; Table 2), whereas the BSG was lower in the tilted trees. The bark thickness remained constant throughout the stem in both the tilted and control trees. The increase in stem area observed for tilted trees (see final diameter, Table 2) was driven by wood (increase in wood area for tilted trees), without differences in bark area leading to a lower bark area ratio for tilted trees.

4 Discussion

4.1 Whatever the posture control mechanism, reaction to artificial tilting does not require an over-investment in total biomass

Regardless of the posture control mechanism, we were unable to observe any change in the total biomass produced by the tilted trees compared to the controls. This would mean that motor function does not an overall higher cost for trees regardless of the posture control mechanism. However, it could also be that we were unable to observe any change because we cannot measure biomass before tilting and compare tree-by-tree changes during growth. Light differences would therefore be hidden by the initial variability. A much heavier protocol following the mass all along the growth (as proposed by Niez et al. 2019) would allow to confirm our observations.

4.2 Reaction to artificial tilting modifies the biomass allocation within the different organs of the plant

The mechanical response to artificial tilting was associated with large changes in tree growth. When tilted, trees exhibited decreased primary growth (height) and increased secondary growth (diameter), consequently decreasing their relative slenderness (Fig. 2). The overall mass was maintained but with different allocation ratios among the roots, stems and leaves (Table 2). Due to tilting, there was an increase in root mass ratio in P. aquatica which was compensated for by a reduced leaf mass ratio, whereas the stem mass ratio did not differ. With the opposite posture control mechanism, the root mass ratio in Se. rubra was not different, but a greater stem mass ratio and lower leaf mass ratio were observed in tilted trees. The higher biomass in the stems of tilted trees of this species suggested that the posture control based on tension wood requires more investment into this organ. However, the variations observed in biomass allocation were small compared to the variations in morphology (e.g. stem ovalisation, stem slenderness). This finding is in agreement with the observations of Poorter et al. (2012) that plant plasticity mainly affects organ morphology rather than allocation of biomass.

Previous works studying biomass allocation changes following mechanical disturbance were all interested in thigmomorphogenetic stimuli where plants were naturally (wind) or artificially swayed during their growth (Osler et al. 1996; Telewski and Pruyn 1998; Cordero 1999; Coutand et al. 2008). In our study, the stem was tilted but not swayed and was attached to a pole, avoiding any bending of the stem during growth. However, we observed a common trend toward a reduction in the slenderness of the stem, caused by reduced primary growth, increased secondary growth or both. This ‘stubby’ shape is adapted to minimise the buckling risk, regardless of the mechanical disturbance.

In thigmomorphogenesis experiments involving high mechanical perturbation, the mean leaf area (Telewski and Pruyn 1998) or leaf area, and therefore LAR (Cordero 1999), was reduced. As a result, the total biomass was reduced (Cordero 1999). In our study, tilting without any effect of wind did not affect leaf area, and the total biomass was similar to that of the control plants. As the morphology (size and SLA) of new leaves after tilting was similar among the treatments for the three studied species, we may hypothesise that trees do not need to adjust their assimilation function in response to tilting.

4.3 Reaction to artificial tilting modifies the amount and quality of wood and bark quantity and improves the posture control of the trees

The most contrasted responses to artificial tilting were observed on the wood and bark amount and organisation around the stem and their density.

In P. aquatica, the motor function is exerted by the bark through asymmetric growth; this growth is strong on the upper side and reduced on the lower side (Ghislain et al. 2019a), resulting in greater bark tensile stress on the upper than on the lower side. Indeed, growth monitoring revealed strong ovalisation of the stem very soon after tilting in P. aquatica. This ovalisation was associated with a change in the bark/wood thickness ratio with a decreased bark proportion (Table 2), a high wood eccentricity (Ghislain et al. 2019a) and a strong change in the bark thickness ratio, with the bark being more than 2 times thicker on the upper side than on the lower side (Fig. 4, Table 2). These strong modifications between both sides of the stem appeared to be due not only to exacerbated growth on the upper side but also to an almost complete lack of cambial activity on the lower side (see anatomical section Fig. 1A).

The efficiency of the bark posture control also relies on the balance in rigidity between bark and wood. P. aquatica was the only species for which the bark was denser than the wood (Fig. 4) for both the control and the tilted trees, suggesting that the bark was stiffer than the wood. This higher density was emphasised in tilted trees where the bark was two times denser than the wood. In tilted trees, as previously shown (Ghislain et al. 2019b), the wood produced for this high eccentric growth is mainly composed of elongated parenchyma cells and a reduced amount of fibre. It therefore affects wood density, as observed in this study, and presumably the modulus of elasticity (lower density and reduced amount of fibres). This loss of wood stiffness is profitable for ensuring the efficiency of remaining upright, as it reduces the stiffness of the wood core when it is bent by the bark. However, this decrease in wood rigidity (not compensated for by any change in bark rigidity) also affects the mechanical properties of the stem, reducing the self-supporting ability of the stem and therefore its critical height (Niklas 1992). P. aquatica trees seemed to compensate for this weakness by virtually stopping height growth (7% height relative increment compared to 39% in control trees), allowing them to reduce their slenderness by 40% compared to before tilting (Fig. 2 and Table 2) and subsequently increase their mechanical safety. This decrease in primary growth directly affects the production of new leaves, which were reduced on tilted trees (Fig. 2).

Se. rubra uses a totally different posture control mechanism. The motor function is fully taken by wood with gelatinous layers in tension wood on the upper side of the stem (Ghislain et al. 2019b). Interestingly, no ovalisation of the stem was observed during growth of the tilted trees, suggesting that ovalisation could be useful only when posture control mechanism rely on bark. This circular shape of the stem may, however, hide eccentric wood growth, which is greater on the upper side than on the lower side (Ghislain et al. 2019a). The tilted stems remained circular, but their diameter increased with a limited decrease in height (Fig. 2, Table 2). A greater diameter with similar wood density compared to that of the controls increases the bending stiffness, and the slenderness was less affected than that in other species. In this species, the mechanism does not rely on bark; moreover, bark acts as an impediment to the uprighting driven by wood (Ghislain et al. 2019a). Interestingly, bark area was not different between the tilted trees and the controls, whereas wood growth was greater. This led to a strong decrease in the bark/wood weight ratio in tilted trees, minimising the counteraction of bark against wood. This effect was reinforced by a lower bark density, suggesting a lower stiffness in tilted trees, minimising the impeding action of bark against wood action. This lower density can be explained by the dilution effect of the phloem fibre in the growing stem. Fibres are deposited by cambial cells but cannot be divided later. To accompany the increase in the bark perimeter linked to wood growth, phloem tissues need to elongate or divide tangentially. This can only be done by parenchyma cells (Angyalossy et al. 2016). Consequently, we propose that the proportion of fibre, and thus the density, decreases in the inner bark of tilted trees.

Owing to a limited slowdown in height growth, leaf production is maintained. We speculate that this maintained leaf production was stimulated by a higher demand for the plant, which produced more wood and denser wood, leading to a greater stem biomass on tilted trees (Table 2).

In Si. amara, a species that uses both efficient tension wood and a trellis structure on the bark, ovalisation of the stem increases continuously with tree growth on tilted trees. Compared with those of the control trees, the stem diameter of the tilted trees increased strongly due to the increase in both wood and bark area. This concomitant growth allowed trees to maintain a similar bark/wood ratio between the tilted trees and the control. The contribution of bark was emphasised by the change in the bark thickness ratio, with the bark being more than 1.5 times thicker on the upper side than on the lower side of tilted trees (Fig. 4, Table 2). However, the bark was less dense and the wood was denser in tilted trees than in the controls. This may imply that the wood contribution was greater than that of the bark. This shift in motor system from the bark to the wood is expected to occur during ontogeny. During the young stage, the mechanism relying on bark is expected to be more efficient, whereas mechanism relying on wood appears more efficient when the stem diameter is greater (Lehnebach et al. 2020a). These observations confirm the hybrid motor system of Si. amara with co-occurrence of improvement of parameters of both mechanisms such as a denser tension wood and an asymmetry of bark growth. These modifications in both wood and bark occurred with a change in relative slenderness and height relative increase, which are intermediate between those of P. aquatica and Se. rubra (Fig. 2).

All the changes in wood and bark composition and/or organisation appear to improve the mechanical function with respect to the posture control mechanism of the species. In future studies, it will be interesting to observe the consequences of such modifications on other wood functions, such as the hydraulic properties of the stem and the possible trade-offs that may have driven species to such different mechanisms during the evolutionary process.