Elevated Growth Temperature Reduced the Physiological Plasticity of Fagus sylvatica Seedlings in Response to Drought and Shade

1. Introduction

The significant increase in temperature with ongoing climate change along with more frequent and severe droughts (e.g. scenario SSP5-8.5) [1] poses challenges for forest species, with long reproductive cycles, slow migration capacity, and yet unknown limits for acclimation to this scenario. Indeed, plant acclimation to warmer growing temperatures is an underexplored process, which gets even more complex when concurrent with other environmental stress factors. Many plant species face the triple interaction of growing under continuously rising temperatures, suffering events of water deficit, and regenerating in the light-limiting conditions of the understory [2,3]. To gain insights into how future warming will affect the regeneration niches of key late-successional species, as well as to understand how tree populations respond to rapid environmental changes, we need to study the phenotypic plasticity of essential functional traits in relation to these environmental stressors [4,5,6]. Changes in key functional traits may alter plant ecological strategies, such as drought and shade tolerance, which in turn may challenge the species’ ability to compete at their regeneration niche and hence influence forest dynamics [7,8,9].

The temperature response of most biological processes, such as metabolic and growth rates, follows an asymmetric Arrhenius equation, which present a maximum rate at the optimal temperature [10]. For example, warming can enhance tree growth and increase leaf area at the expense of root investment, particularly in deciduous temperate species as long as they are well-watered and fertilized. However, growth decreases once the optimal temperature for growth is exceeded. [11,12,13]. This decline is attributed to a reduction in plant leaf area and net photosynthesis (A_n) when the optimal temperature for the photosynthetic activity is exceeded [11,14,15]. Additionally, the rate of plant respiration continues to rise after reaching the optimum for An, further contributing to decreased growth [16]. A reduction in carbon availability is also shown by decreased leaf mass per area (LMA) with temperature [17]. LMA is a relevant trait of the leaf economical spectrum usually associated with plant ecological strategies, such as drought or shade tolerance [18,19]. Additionally, leaves are generally smaller when developed at elevated temperatures to reduce the boundary layer and facilitate heat dissipation [20,21,22]. Both photosynthesis (A_n) and respiration (R_d) acclimate to changes in growth temperature by the downregulation of respiration [16,23,24] and the upregulation of the temperature optimum for photosynthesis [12,23,25,26] which allows plants to maximize carbon gain with warming [27]. However, a decoupling between A_n and stomatal conductance (g_sw) at high temperature has been also reported in many species, particularly in warm adapted species, in a way that A_n decreases whereas g_sw increases due to the so-called evaporative cooling [28,29,30,31,32], with the concomitant reduction in intrinsic water use efficiency (iWUE) and enhancing the risk of dehydration [33].

The extent to which stomata acclimate to warming depends on a balance among their key functions. Stomata not only play a role in photosynthesis and contribute to the carbon economy, but they also regulate leaf water loss through the transpiration rate (E), leaf temperature through evaporative cooling [34], ROS-mediated impairment of normal metabolism [35] and leaf water potentials [36,37]. Maintaining these water potentials within specific safety limits is crucial to prevent hydraulic system failure [38] and leaf dehydration [39]. Leaf transpiration is the product of vapor pressure deficit from leaf to the atmosphere (VPD_L) multiplied by the total conductance for water vapor. Therefore, for the same stomatal aperture, more water is loss to the atmosphere when air temperature rises as consequence of elevated VPD_L [32,40]. To sustain higher water flow at higher evaporative demand, the leaf water potential at the turgor loss point (Ψ_TLP) usually decreases in seedlings grown at warmer temperatures [41], as often occurs with species or populations inhabiting warmer and dryer habitats [42,43], but it is still unknown if these adjustments occur in response to warming or to the interaction with other stressors such as low soil or air moisture [44].

Acclimation of the hydraulic architecture to warming in forest species remains understudied, even though it is expected to change since warmer climate increases the evaporative demand [12,41]. For example, well-watered Populus tremuloides seedlings grown at warmer temperatures than ambient showed higher growth, allocation to roots, which also increased their hydraulic conductance and the sapwood area to leaf area (Huber value, HV) [41]. However, in a xeric site, both Juniperus monosperma and Pinus edulis showed lack of acclimation of the HV or xylem anatomy after four years of atmospheric warming and only the juniper reduced its leaf specific hydraulic conductivity (K_L) associated with a decrease in canopy conductance under warming [40]. Well-watered plants grown under warmer conditions can develop larger and/or a more frequent conduits in the xylem [41,45,46], which may increase conductivity, but not necessarily affect embolism resistance. On the contrary, the vessel lumen of Eucalyptus camaldulensis and E. grandis was smaller, the specific conductivity (K_s) lower and wood density (W_D) higher when seedlings grew above optimum growth temperatures [47,48]. Besides changes in xylem anatomy, as the temperature increases, the density of sap remains more or less constant, but the viscosity decreases (i.e. lower resistance of water to motion) increasing whole plant hydraulic conductance just by the effect of sap temperature [47,49].

Warming increases the water deficit experienced by plants [50,51], but water deficit also exacerbates the negative effects of growing at supraoptimal temperature [52,53]. Water deficit reduces temperature optima for both photosynthesis, as consequence of reduced diffusional CO₂ conductances [54,55], and tree growth due to the decreased photosynthesis rate, reduced production of leaf area and increased LMA [11]. However, an active osmotic adjustment shown in warmed trees can benefit drought tolerance [56]. Stomatal closure triggered by water deficit increases leaf temperature and, with it, the risk of tissue damage, even when thermal acclimation occurs [20,57]. When stomata are partially or fully closed as response to soil water deficit, water loss through the leaf cuticle becomes crucial for plant survival [58,59]. This leaf minimum conductance (g_min) usually decreases as consequence of drought, but high temperature during short time periods promotes the increase of g_min [60,61,62], although g_min may decrease over prolonged warmer temperatures due to the sustained higher VPD [33].

Another environmental stressor which hampers growth, and the regeneration capacity of many temperate tree species, is light availability [63,64]. In shaded conditions, plants allocate more resources to elongating stems and increasing leaf surface area to capture light [65,66,67], showing conflicting trade-offs with acclimation to water stress [68,69,70,71,72,73]. For example, lower capacity of osmotic adjustment of plants grown under shade [74,75,76], lower LMA [17] or lower biomass allocation to roots [77] can negatively affect the acclimation to drought and warmer temperatures. In contrast, large gaps in the forest expose seedlings to high light intensity and warmer temperatures, leading to increased metabolic activity and growth rates. However, these conditions also pose challenges such as heightened water stress and the risk of heat damage [20,78,79,80,81,82]. To understand the challenges that forest recruitment will face in a warmer world, we need to understand how elevated temperatures, water stress, and light conditions interact to affect plant performance. Additionally, it is important to recognize that the response of the stress combination is unique and cannot be considered as the additive contribution of each environmental factor [83,84,85].

Most studies of temperature acclimation did not control air humidity, which raised VPD and might flaw the effect of temperature due to the high sensitivity of transpiration to increasing VPD [41,86,87]. Moreover, the transpiration response to VPD was shown to acclimate with the prevailing VPD during growth [88,89,90]. Therefore, comparing results of temperature acclimation from studies conducted at different VPD may be controversial [91]. In the other hand, the direct response of gas exchange to temperature, by adjusting ambient humidity to maintain the same VPD, has been examined far less extensively [92]. Therefore, disentangling the direct effect of warming from the soil and the atmospheric drought (e.g. soil water deficit and high air VPD) requires further studies because shifts in resource allocation, morphological and physiological acclimation to warming can increase the vulnerability to overcome droughts [12,51].

In the present study we exposed one-year-old seedlings of Fagus sylvatica L. grown under optimal (25 °C) and supraoptimal (+7.5 °C) temperatures, to a cross-combination of water deficit and shade. VPD was kept constant at the two growing temperatures, with the aim to evaluate how warming affects the potential mechanisms that confer multi-tolerance to drought and shade [85]. We hypothesized that: 1) Warmer temperatures will reduce plant growth due to reduced leaf area rather than by a proportional decrease in photosynthesis rate, which will exacerbate the detrimental effects of combined water deficit and shade. 2) Stomatal conductance is expected to increase under warming when water is not limiting to foster evaporative cooling of the leaves, but it will be decreased when combined with water deficit. 3) The xylem anatomical traits associated with the plant hydraulic conductivity will adjust accordingly with the changes in leaf transpiration to coordinate water demand and supply. 4) Phenotypic plasticity to water availability and light intensity will be reduced due to warming.

2. Results

Changes in Morphology and Biomass Allocation

Warming had a significant effect in plant growth and biomass after 90 days from the beginning of the experiment (Table 2). Plants in T25 grew thicker and produced more biomass and bigger leaves, resulting in higher leaf area. Plants in T32 produced more leaves but were smaller (Figure 1; Table A1). Plant growth was closely related with leaf net photosynthesis (An) across treatments (Figure 1). Both water deficit and low light conditions reduced growth, biomass and the number of fully developed leaves, although the cumulative effect of light intensity exerted a stronger effect than water deficit (Figure 1). Water deficit increased the water use efficiency of leaves (Figure 1). Plants grew in low light produced the least biomass at both growth temperatures and water treatments. Growth temperature changed the biomass allocation pattern and plants in T32 allocated more resources to leaves and less to shoots than in T25, and water deficit increased the proportion of root biomass (Table 2; Figure A2).

Leaf Gas Exchange, Water Use Efficiency and Stomatal Traits

Leaf gas exchange was significantly affected by temperature, light intensity and watering during growth (Table 2). Unlike for net assimilation (A_n), for which light intensity exerted the stronger effect, water treatment was the main regulator of g_sw, R_l, g_min and iWUE, particularly in T32 (Table 2). Nevertheless, warming significantly decreased A_n and increased R_l, but g_sw and g_min increased only under well-watered conditions (Table 2), thereby decreasing iWUE, which aligns with more negative isotopic carbon composition of leaves grew at both light conditions (δ¹³C) (Figure 2). Warming when combined with water deficit had a more negative effect on leaf gas exchange for any light growth condition. Stomatal traits were not directly affected by warming although warming had significant interaction with the other two growing factors. Stomatal density (S_D) was mostly affected by the light intensity (higher at higher intensity) only at T25, but stomatal length (L_s) was mostly affected by available water (shorter stomata under water deficit) only at T32.

Leaf Water Relations

Water availability was the main factor affecting leaf water relations, although warming lowered leaf water potentials (Ψ_pd, Ψ_leaf and Ψ_stem) in T32 more than in T25. Growing at high light produced significant more negative leaf water potentials (Figure 3; Table 2). Most parameters derived from the P-V curves were significantly affected by water availability, although with strong interactions with light and temperature. In T25, well-watered plants showed higher osmotic potential at full turgor and loss of turgor (π¹⁰⁰, π⁰) than under water deficit, 0.8 MPa of osmotic adjustment at high light intensity, and higher leaf capacitance before the turgor loss point (C_π¹⁰⁰ or C_FT) but lower maximum modulus of elasticity (ε_leaf) than under water deficit at both high and low light intensity. However, warming prevented the capacity of osmotic adjustment of plants growing under high light intensity and reduced the plasticity to light and water in T32. By contrast, warming reduced the safety margin (SM_leaf, i.e. the difference between the minimum leaf water potential at midday and the potential at turgor loss point) in T32 in all treatments, particularly lower under water deficit treatments.

Hydraulic Traits and Stem Anatomy

All functional traits related with water flow were severely affected by water deficit, as expected, although the stem specific hydraulic conductivity (K_s) was more affected by light intensity than for water availability (Table 2), resembling what was observed for morphological traits (e.g. plant height, diameter and total leaf area). Warming did not have a significant effect on K_s and only a moderate increase on the plant hydraulic conductance (k_plant) under high light and well-watered conditions (Figure 4). In contrast, warming decreased the leaf specific hydraulic conductivity (K_L) and Huber value (H_v) in all treatments except for well-watered plants at high light intensity. Shade significantly decreased K_s at T25 and increased H_v and K_l in T25 well-watered plants, but not in T32, and decreased k_plant only at T32. The mean loss of xylem conductivity (PLC) was lower than 10% in well-watered plants and lower than 16% in water-stressed pointed to moderate drought during the whole experiment.

Stem xylem anatomy was remarkably affected by water stress, which induced an increase in vessel density by unit of xylem area (V_D) but lower vessel area (V_A) and hydraulic diameter (D_h) than under well-watered conditions. Growing at high light intensity produced higher V_A and lower V_D. Indeed, we found a strong trade-off between V_D and V_A and an exponential relationship between TLA and V_A across all treatments (Figure 4). Warming decreased V_A and increased V_D under well-watered conditions and slightly under shade and water deficit (LS). However, for the rest of the treatments, growth temperature did not affect xylem anatomy. No significant differences between treatments were observed in the xylem area occupied by radial parenchyma.

Plasticity in Response to Temperature, Light, and Water Availability

Light and water availability significantly affected F. sylvatica phenotypes at different organization levels. These changes were also significantly modulated by growth temperature and for many traits, growing at warmer temperatures limited their phenotypic plasticity (Figure 5). For example, the phenotypic plasticity index (PPi) in response to growing light intensity was consistently higher at T25 than at T32 under well-watered conditions, although only higher regarding leaf gas exchange when evaluated under water stress (Figure 5). PPi was also higher for water availability for most traits except for leaf gas exchange at high light intensity, because the combination of warming and water stress reduced leaf gas exchange more at T32 than at T25. Under shade, plasticity for water deficit was higher at T32 in all but for leaf water relations, and higher adjustments were observed at T25 than at T32 (Figure 5). Overall PPi for all treatments and variables is presented in Figure A6.

The degree of plasticity differed among traits. Traits related with cumulative growth were among the most plastic, including biomass partitioning (Figure 6). By contrast wood density (ρ) showed little plasticity although significant for many treatments including an increase due to water deficit or warming (Figure 6). Stomatal morphology showed little variation and density was more plastic that stomata size. The minimum conductance to water vapour, or epidermal conductance (g_min), vessel density (V_D) and the number of leaves showed the highest increases in response to warming at high light whereas biomass, particularly roots, and capacitance at the turgor loss point (C_TLP) presented the highest reductions (Figure 6). Finally, when plants grew with low light and low water availability, a higher growth temperature resulted in decreased H_v and leaf specific conductivity (K_L) and plant hydraulic conductance (k_plant) (Figure 6).

Correlations Among Traits and Principal Components Analysis (PCA)

Plant growth traits (height and diameter) were strongly correlated among them and in general exhibited strong positive correlations with leaf gas exchange rates, K_s and V_A, irrespectively of growing temperature (Figure 7 & Figure 8). However, the temperature closest to the optimum (T25) produced significant positive correlations between plant height and ρ and trade-offs between height and H_v, ρ and H_V and ρ and iWUE, which were absent at T32. In contrast, warming (T32) led to significant positive correlations of plant diameter and LMA and k_plant. At T25, iWUE correlated negatively with g_sw, ρ, leaf C_FT, k_plant, leaf water potentials and V_D, and positively with δ¹³C, ε_leaf, RWC_TLP and H_v. But at T32 iWUE only negatively correlated with g_sw, k_plant leaf water potentials, and SM_leaf and positively with K_L. In T32, L_s correlated positively with LA, number of leaves, g_sw, g_min, leaf water potentials, k_plant, V_A and SM_leaf, and negatively with K_L and V_D, but at T25 there was not direct correlation with stomatal size and only with PCI and LMA, A_n and stem diameter. Leaf water relations showed higher levels of correlation with other traits (particularly with the ones related with leaf water potentials and hydraulics, but also anatomy stem anatomy and density) at T25 than at T32.

PCA was useful to characterize the two main drivers of variation: water (PC1) and light available (PC2) dividing the space into four well-characterized quadrants in clockwise order: HW, LW, LS and HS (Figure 9). The first component of PCA (PC1) explained 44 % of the variation and was positively associated with well-watered plants characterized by high values of k_plant, V_A, E, Ψ_pd, Ψ_leaf, and TLA; and negative scores were associated with water stressed plants for traits related with WUE, V_D, K_L and ε_max. PC2 explained 22 % of variation and was positively related with light intensity through LMA, δ¹³C, A_n, and S_D and negatively with SWC, π₀ and C_π100. Growing temperature pushed T25 plants towards more positive score of PC1 when well-watered, but less negative PC1 and more positive PC2 under water stress and high light and less negative PC2 under shade and water stress (Figure 9).

3. Discussion

Temperature is one of the most important determinants of plant function and development, although we know little about how temperature directly affects tree performance and physiology compare with other stresses such as drought or shade. The present study is among the first to disentangle the effect of growth temperature after controlling vapor pressure deficit (VPD) on key morphological, physiological, and anatomical traits in forest tree seedlings. Furthermore, we full fill a gap of knowledge by addressing the effect of warming in combination with superimposed stressors such as water deficit and shade in tree species, what was recently termed “multifactorial stress combination” and studied mostly in model plant species [93,94]. We revealed that warming exerted a direct influence on many of the functional traits related to plant growth and physiology (mostly negative; Figure 6), phenotypic plasticity to shade (positive) or water deficit (highly negative) and that some specific adjustments commonly observed at temperatures close to the optimum (e.g. T25) were absent when seedling were growing at supraoptimal temperatures (i.e. T32). Nevertheless, F. sylvatica seedlings showed strategies towards a better acclimation to warming that arguably may have improved plant performance. We discussed the extent of plant acclimation to warming and how warming may have affected the shade and drought tolerance of this late successional species.

Impact of Warming on Plant Morphology, Physiology and Anatomy at High and Low Irradiance

Warming reduced plant biomass and stem diameter (Figure 1), whereas height was less affected or even promoted under shaded conditions, which is consistent with previous studies of tree species growing near their warmer range limit [13,95,96]. Wood density (ρ) increased at warmer temperature which could compensate for the higher height/diameter ratio of warmed plants and favor stem biomechanics [97]. Indeed, the increase in ρ was associated with higher V_D and lower V_A in warmed plants growing at high irradiance (Figure 4), hence increasing the wood volume occupied by cell walls. Denser wood, associated with lower hydraulic diameter, can limit hydraulic conductivity and efficiency of water transport, although it can be an advantage in case of xylem embolism [98,99]. In fact, a strong positive correlation was found between V_A and K_s, but negative with V_D, which is indicative of the trade-off between mechanical and hydraulic efficiency. Nevertheless, despite the reduction in V_A in response to growth temperature, K_s could be partially compensated for the decrease in viscosity of water as temperature increased [100,101]. Well-watered seedlings of trembling aspen reduced leaf hydraulic conductance by 30 % as consequence of +5 ºC warming, although kept stem K_s similar across treatments (Figure 4) [41]. Our results highlight the acclimation capacity to warming of xylem conductive elements of beech seedlings confirming our third hypothesis, in contrast to the limited xylem responses of gymnosperms exposed to prolonged experimental warming [40].

Warmed plants at high irradiance (T32 HW) also produced lower leaf area (TLA) than plants at T25. However, under shaded conditions (LW) T32 plants produced higher TLA and more leaves with lower LMA (Figure A3). Although leaf area was positively associated with growing temperature in natural beech forest of Central Europe [57,102], an excess of warming may induce a significant reduction in leaf thickness, cell size, and total leaf area, alongside decreased starch content under high irradiance [103]. Here, the reduction of TLA at high irradiance was due to reduced leaf size, despite the number of leaves increased. Smaller leaves can cooldown more easily by evaporative cooling due to lower resistance of the boundary layer [22,57,104]. The same pattern was observed in warm-adapted Populus trees, which down-regulated leaf temperature by producing smaller and thinner leaves with higher stomatal density [105]. Hence, warming promoted a conservative production of transpiring tissue under high irradiance despite VPD was kept equal among temperature treatments, which correlated with a concomitant decrease in V_A indicating a close correlation between transpiration rate at the plant level and water supply [106,107,108].

Biomass of T32 HW plants was reduced by half compared to T25 HW plants, which was associated with lower carbon gain capacity (TLA decreased from 1800 cm² at T25 to 1000 cm² at T32, and 15 % reduction in A_n), together with increased carbon losses (e.g. R_d was 25 % higher at T32), confirming partially our first hypothesis (Figure 2 and Figure A2). The temperature optimum for A_n in beech was estimated at 24.5 °C [109], hence the little reduction of A_n at 32 °C could be attributed to thermal acclimation, maybe by rising A_n optimum temperature at T32 [12,26]. In any case, the depletion of A_n might be associated with metabolic constraints, particularly the downregulation of leaf photochemistry, which frequently occurs under elevated temperatures [110,111], than to diffusional limitations to CO₂ transfer by stomata. T32 HW plants increased g_sw (+48 %) in response to warming at similar VPD, confirming previous studies of warming at constant VPD [92], and our second hypothesis. Other studies combining warming with elevated VPD showed no changes or decrease in g_sw, although the transpiration rate (E) increased significantly (Figure 2) [33,111,112]. Even though we ensure a similar growing air VPD, the leaf to air VPD should be slightly higher for plants growing at T32, and thus E was slightly higher at T32, because leaf temperature and the saturated water vapor partial pressure within the leaf was higher at T32. The rise in g_sw and E experienced by T32 HW plants correlated also with higher k_plant suggesting strong coordination between water supply and demand at the plant level. Higher temperatures also promoted faster water use which eventually leads to faster depletion of soil water for warmed plants [113]. Reduction in TLA at T32 HW was important to avoid lowering Ψ_leaf, beyond the leaf safety margin (SM_leaf) since Ψ_tlp did not change with temperature. Reduced LMA was evident in T32 HW to regulate leaf temperature and reduce leaf construction costs [114], although the lower LMA precluded lower leaf capacitance either before or after turgor loss than in T25 HW leaves (Figure 6), partially supporting previous studies [41].

Plasticity of g_min to warming or shade remains largely unexplored [61,115]. g_min doubled at T32HW and was significantly higher at T32LW than at T25LW (Figure 2), confirming that g_min is strongly dependent on temperature (Table 2) [33,61,116] found for beech that both high VPD and high temperature reduced g_min in the absence of soil drought, whereas our results pointed to an opposite direction, in line with [29,60]. g_min accounts for how easily the water vapor exits the leaf once stomata are closed. However, incomplete closure of stomata may contribute to g_min, along with higher permeability of cuticle + epidermis [117]. At moderate high temperature (<40 °C) the stomatal component of g_min is of greater importance than at higher air temperature and VPD, where the cuticle component is governing g_min [118]. Accordingly, we found strong correlations between g_min and stomatal traits, especially PCI, which suggested a lack of complete closure of the stomata as the main cause explaining the rise in g_min at T32.

Growing under shaded conditions, when water was not limited (LW), mitigated the negative effects of warming described above and stimulated plant growth and biomass. Indeed, T32 LW seedlings produced higher: biomass (particularly leaves and roots), number of leaves, and slightly higher wood density, stem height and diameter, in contrast to what was observed at high light intensity. Warming induced a divergent ecological strategy in well-watered beech seedlings under low light levels [69] aimed to compensate for the reduced carbon gain by unit of leaf area (lower A_n, but higher R_l) with increased TLA, greater allocation of biomass to leaves, higher number of cheaper leaves (lower LMA), lower iWUE and δ¹³C, and reduced H_V and K_L, but higher transpiration rate to increase C_i and reduce leaf temperature (Figure 2 and Figure 4), in a way to minimize photorespiration losses [34]. As warming accelerated cell division, the total number of leaves increased [119]. At least for this late successional and highly shade-tolerant species, shade was beneficial to offset the negative effects of increased temperatures [120]. Beech seedlings growing under shade and warming followed an acquisitive strategy, prioritizing the consumption of water to cool down the leaves and mitigating leaf warming, but stimulating leaf growth with lower H_V and K_L to compensate for the reduced A_n [121,122]. T25 LW followed a more conservative strategy, with lower TLA and gas exchange, but higher Ψ_leaf and tougher leaves (high LDMC and LMA). This fast growing strategy of T32 LW could be risky if water is limited, since higher leaf area, but reduced leaf capacitance and Ψ_leaf lowered the leaf safety margin (SM_leaf) (Figure 3), which make beech shaded phenotypes more vulnerable to water stress under warming [123].

Impact of Warming and Water Deficit at High and Low Light Intensity

The hydraulic architecture (e.g. lower K_L, H_v and k_plant), significantly lower A_n, but higher carbon losses related to respiration and changes in morphological traits such as lower LA, TLA and LMA were related to lower biomass production, limited leaf osmotic adjustment and lower safety margin for plants growing under water-stress and warming (Figure 6), indicating important interactions between carbon investment and hydraulics [102,124,125]. Shade aggravated the negative effects of combined water stress and warming [69]. The impact of the concurrently multiple stress condition (i.e. water deficit, shade and warming) on beech seedlings conferred stronger pressure on plant’s performance than any of the stress factors individually or in combination in line with previous studies with model plants [93,94]. This negative outcome can be viewed as the result of the trade-offs emerging to mitigate the combined effect of the two limited resources (i.e. soil water content and light intensity) amplified by the effect of warming (i.e. a state variable which constitutes a non-resource factor of stress) which exacerbated the already resource-limited stresses [126,127,128].

As water availability diminishes, elevated growth temperatures lead to a steeper reduction in A_n, primarily influenced by increased stomatal closure to avoid hydraulic dysfunction, but at the cost of decreased carbon uptake and growth [55]. Thus, to avoid excessive tension in the xylem and the risk of embolism, stomatal conductance decreased [129] and hydraulic resistance increased (lower k_plant) limiting water transfer (Figure 2 and Figure 4) [130]. Water-stressed plants showed lower K_s but a more redundant xylem, which can be an advantage in case of xylem embolism [99]. The start of embolism formation (P12) in Fagus sylvatica was estimated at Ψ_stem below -1.8 MPa [131,132], although beech seedlings from the same population and acclimated to water deficit reached P12 at much lower levels (-2.4 MPa), suggesting very narrow xylem safety margin to preserve xylem functioning under drought [132]. Lower Ψ_stem and Ψ_leaf in droughted T32 plants may be the consequence not only of higher biomass allocation to leaves and decreased water viscosity [133] but also of lack of leaf osmotic and elastic adjustments. T32 HS leaves did not adjust osmotically (i.e. not change of water potential at full turgor) that together with the lower Ψ_leaf decreased significantly SM_leaf to almost 0.2 MPa. Lack of thermal acclimation was also observed for leaf capacitance and ε_max, decreasing further the plant's ability to tolerate fluctuations in Ψ_leaf and maintain turgor pressure (Figure 3) [134]. Maintenance of volume in chloroplasts by osmotic adjustment may preserve leaf photosynthetic capacity under drought [135] and prevent damage to chloroplasts from toxic concentrations of ions [136]. Reduced A_n in droughted T32 plants may also arise from lowered synthesis of protective enzymes, antioxidants and heat dissipation proteins, which are needed for mitigating the higher thermal and oxidative stress produced by the higher stomatal closure and limited leaf cooling capacity [78,137,138]. The capacity for osmotic or elastic adjustment was significantly hindered by light-limiting conditions and warming, which ultimately constrained the ability to maintain leaf turgor under conditions of water stress [139], reducing leaf gas exchange and growth [140].

In contrast, T25 HS plants were able to double A_n of T32 HS, achieve more than 0.8 MPa of active osmotic adjustment, lowering Ψ_TLP below 3 MPa (Figure 2 and Figure 3; Table A3), and increased ε_max. Higher ε_max implied more inelastic (higher stiffness) cell walls, an adjustment that allowed leaves to rapidly change the water potential within the cells as the leaf loses water and extract water from drier soil [141,142]. T25 HS also showed higher leaf shedding, which increased H_V and the supply of water to the remaining leaves (higher K_L), lowering vessel area and producing a more redundant xylem, which may protect the hydraulic system from larger hydraulic failure (Figure 3; [59,143,144,145]), overall improving the whole plant water balance [146] and sustaining higher leaf gas exchange rates (Figure 2). Higher A_n in T25 HS could be the consequence of osmotically active metabolites [147,148] which not only stabilize enzymes and membranes, but also regulates the stomatal sensitivity to leaf water potential, allowing T25 HS archive 65 % higher g_sw than T32 HS (Figure 6) [149,150]. Both temperature treatments decreased g_min in response to water deficit which may offer significant advantages for the conservation of xylem function, particularly during warmer and drier conditions [33,151].

Phenotypic Plasticity and Regeneration Niche of Beech Seedlings Under Multi-Stress Factors

Phenotypic plasticity allows plants to modify their phenotype to match their current environment and it is considered beneficial when individuals can express superior phenotype in each environment compared to a fixed phenotype. However, there are limits and costs associated with plasticity [152]. Not every plant species is equally plastic for every trait, suggesting trade-offs derived from their potentially high cost of plastic responses and the complex interaction among traits, which are considered critical to their adaptive value for plant fitness [153,154]. Beech seedling showed high phenotypic plasticity (Figure 4 and Figure 5), and as result of the long-term applied environmental factors, traits associated with whole-plant biomass demonstrated greater plasticity compared to those measured at the organ or tissue levels (Figure 5), although we detected strong synergistic effects with leaf gas exchange traits, and antagonistic relationships with the number of leaves, H_V and K_L (Figure 6). Some traits exhibited limited plasticity within the species and studied environments, such as those related with stomatal morphology, RWC_TLP, SWC, RP and ρ (Figure 6, Figure A1), indicating higher costs [155]. Warming reduced overall plasticity, in agreement with our fourth hypothesis related to the cost associated with maintaining plasticity in resource-limited environments [156]. Interestingly, warming in the absence of water deficit increased the phenotypic plasticity of F. sylvatica seedlings growing at low light intensity (Figure 6), revealing the strong adaptation of this shade-tolerant species to grow in the humid forest understory [120,157]. Extensive phenotypic plasticity in key functional traits is often considered favorable for the persistence of populations under rapid climate change [4]. Nevertheless, the often-associated increase in VPD with temperature may further limit beech gas exchange [143,158] and growing in mixture with competing species, like oaks, which are also more drought tolerant but higher water consumers, may also compromise current regeneration potential for beech seedlings under climate change [54,159,160].

4. Materials and Methods

Plant Material, Growth Conditions and Treatments

Beech seedlings were established from nuts collected in the Natural Reserve of ‘Hayedo de Montejo’ in 2021 (458,249; 4,551,615 UTM ETRS89). In February 2022, after three months at 4 °C in darkness and moist conditions, one seed per pot germinated within few days of difference in 9 L pots, filled with a mixture of peat and sand (75-25% v:v). Each pot was fertilised with a release fertiliser (Nutricote®, 16N:4.4P:8.3K, 2.5 gr L^-1 of substrate). Pots were randomly arranged in two different walk-in growth chambers set at two different temperatures but the same air VPD. One chamber was set with day/night temperatures and relative humidity of 25/15 ± 0.5º C and relative humidity of 40/68 ± 1% (VPD of 1.91/0.86 ± 0.06 kPa) (named T25), close to the temperature optimum for photosynthesis for beech [109], and the other to 32.5/22.5 ± 0.5º C and 61/80 ± 1% (VPD of 1.91/0.86 ± 0.06 kPa) (named T32), both with a 12 h of photoperiod. +7.5 ºC is the expected increase in mean temperature during June to August for Western and Central Europe and Mediterranean region according to the scenario SSP5-8.5 for 34 models of AR6 CMIP6 (interactive-atlas.ipcc.ch) (IPCC, 2023). In each chamber, we established two treatments of light intensity: high PPFD (ca. 600 μmol photons m⁻² s⁻¹ at the canopy height) and low PPFD (ca. 200 μmol photons m⁻² s⁻¹ at the canopy height). Low PPFD (e.g. shaded conditions) was provided by grey polyethylene shade nets mounted on frames and was measured as the average of six different measurements using a quantum meter (MQ 200; Apogee Instruments Inc., Logan, UT, USA) in a horizontal position on leaves. Plants were developed under well-watered conditions for two months when seedlings with similar height and number of leaves were assigned to two contrasted watering treatments. Well-watered plants were watered daily in the morning to full capacity. Water-stressed plants were without watering until stomatal conductance reached 1/3 of well-watered plants (ranging from ca. 10 days in high PPFD to ca. 24 days in low PPFD) and then plants were watered every three days with an initial amount of 150 ml of water for plants under high PPFD and 65 ml of water for plants under low PPFD (about 1/3 of daily evapotranspiration in both cases). Once a week, plants were weighed and the volumetric soil water content (SWC) was measured by time domain reflectometry (TDR, Trase System I, Soil Moisture Equipment, USA) and watering amounts were adjusted to keep the same proportionality. At this time the final multifactorial design (two factors: water and light) was set for each growing temperature: (1) high PPFD (ca. 600 μmol photons m⁻² s⁻¹ at the canopy height) and well-watered conditions (HW); (2) high PPFD and moderate water stress (HS); (3) low PPFD (ca. 200 μmol photons m⁻² s⁻¹ at the canopy height) and well-watered conditions (LW); (4) low PPFD and moderate water stress (LS). Every week plants were randomly changed of position within each treatment. Every three weeks, plants were interchanged between chambers maintaining the specific warming treatment (i.e. same conditions, but different growing chambers).

Growth and Biomass Allocation

Growth and biomass were measured in six randomly selected plants of each treatment at the beginning of the experiment that were not sampled for any other measurement until harvest. Height, stem diameter and number of fully developed leaves were measured six times throughout the experiment (every 15 - 20 days) (see Figure A1). After 90 days growing at their specific treatments the same plants were harvested and biomass divided into roots, stems and leaves. Roots were separated from the soil by gentle washing under clean tap water. Samples were oven dried at 60 ºC for 7 days and weighed. We calculated root (RM), stem (SM) and leaf mass (LM) fractions as the ratio of organ dry mass to total plant dry mass. One representative fully mature leaf from the top canopy of each plant was scanned to get leaf area using ImageJ [161] and then oven dried at 70 ºC for 48 h for constant dry mass (Mettler Toledo AB 204) to determine the leaf mass per area (LMA) as the ratio of leaf dry mass to projected area. The total leaf area (TLA) of each plant was estimated from LM and LMA. The mean leaf area for each plant (LA) was calculated as TLA divided by the number of leaves.

Gas Exchange and Water Use Efficiency

One fully mature leaf from the upper plant canopy was selected to measure the net photosynthetic rate (A_n, μmol CO₂ m^-2 s^-1), stomatal conductance (g_sw, mol H₂O m^-2 s^-1) and transpiration rate (E, mol H₂O m^-2 s^-1) were measured twice during the experiment in five plants per treatment using a portable IRGA system (LI-6400XT) with the transparent chamber (6400-08 clear chamber) to record the leaf gas exchange at growing conditions. Measurements were conducted at growth conditions of air temperature (block temperature was set at 25 ºC for T25, and 32ºC for T32) and relative humidity to get a VPD close to 1.9 kPa. The reference CO₂ concentration was set at 400 μmol m^-2 s^-1 and flow rate of 300 μmol m^-2 s^-1. After record of gas exchange at growing conditions, the light source was switched off inside the chambers and after 20 min of acclimation in the dark, dark respiration (R_d) was measured. Mitochondrial respiration in the light (R_l) was determined as 0.5 R_d. All measures of gas exchange variables were corrected with the minimum conductance (g_min) (see below) according to [147]. The leaf water use efficiency (WUE) was calculated as (A_n/E) and the intrinsic water use efficiency (iWUE) by (A_n/g_sw).

Stomatal Traits and Carbon Isotope Composition (δ¹³C)

The same leaves used for gas exchanges measurements were collected for carbon isotope composition (δ¹³C) and then for stomatal traits. Two leaf discs taken on each side of leaf blade were dried for 3 days at 65°C and analysed at the Stable Isotope Facility of the UC Davis (CA, USA). Isotope composition was measured with an elemental analyser (PDZ Europa ANCA-GSL, Sercon Ltd, Cheshire, UK) interfaced to an isotope ratio mass spectrometer (PDZ Europa 20-20; Sercon Ltd) with internal standards reaching δ¹³C_VPDB standard deviation < 0.07 ‰. Leaf trichomes were removed using duct tape and three negative impressions were taken of the middle portion of the leaf avoiding main veins and using nail varnish. The impressions were attached to a microscope slide using transparent tape and imaged under a light microscope (Leica DM2500 LED) using the software Leica Application Suite X 3.7.2.22383 (Leica Microsystems GmbH, Wetzlar, Germany). Photomicrographs were analyzed using ImageJ [161]. Stomatal density (SD) was calculated as the number of stomata per unit leaf area (nº mm^-2) on three leaf areas of 1,15 mm² each using x10 magnification. Within each leaf area 10 stomata were randomly selected to measure in μm the guard cell length (SL), stomatal pore length (Pls), stomatal complex width (Ws) and guard cell width (GWs) using x40 magnification. The potential conductance index (PCI) was computed as PCI = (GCL)² SD 10^-4 [162], as a surrogate of potential maximum stomatal conductance.

Minimum Leaf Conductance or Residual Conductance (g_min)

At the same time of gas exchange measurements, one healthy, fully expanded leaf from the same plants was immediately stored in plastic zip bag with a moist tissue paper and kept in a portable cooler at 4 °C until they were taken to the laboratory. In the lab, leaves were stored at 4 ºC in darkness with the petiole submerged in water overnight until the next day for measurements. We estimated g_min through bench dehydration of the rehydrated leaves with the petioles sealed with a high-strength adhesive (Loctite, Prism 401)[163]. Briefly, leaf area was measured at the beginning of the dehydration and leaves were placed on top of a perforated rack at constant air temperature and relative humidity (ca. 25 °C and 46 %, respectively) without direct light and were repeatedly weighed with an analytical balance to determine the steady state rate of water lost by unit of leaf area once stomata were closed, i.e. epidermal transpiration (E_min, $\mathrm{m}\mathrm{o}\mathrm{l} {\mathrm{H}}_2\mathrm{O} {\mathrm{m}}^{-2} {\mathrm{s}}^{-1}$). To get the epidermal minimum conductance by one side of the leaf (g_min, $\mathrm{m}\mathrm{o}\mathrm{l} {\mathrm{H}}_2\mathrm{O} {\mathrm{m}}^{-2} {\mathrm{s}}^{-1}$) we followed the next approximation (g_min = E_min P/(2 VPD)), where P is the atmospheric pressure (94 kPa) and VPD is the air vapor pressure deficit (1.65 kPa), and the boundary layer conductance was assumed negligible.

Pressure Volume Curves and Relate Traits

Leaf water relations were measured using pressure-volume (P-V) curves of leaves sampled at the end of night period on five seedlings per treatment. Leaves were rehydrated for six hours in the dark and at 4 °C with only the petiole immerse in distilled water. After that leaves were scanned to get the initial leaf area and were allowed to slowly dehydrate at constant room temperature using the free transpiration method [164,165]. Leaf water potential (Ψ_leaf) was measured with a pressure chamber (model 1000; PMS Instrument Co., Albany, USA) and leaf mass with a precision balance at intervals, starting with Ψ_leaf > -0.05 MPa and until ca. -3.5 MPa for HW and LW treatments and -6 MPa for HS and LS treatments. We checked all curves for oversaturation during the first steps of dehydration and when present was corrected as in [166,167]. Leaf dry mass was determined as described above. The osmotic potential at full turgor (π¹⁰⁰ sometimes represented by Ψ_FT), the turgor loss point (π⁰ or Ψ_TLP), leaf capacitance before the turgor loss point (C_{_}π¹⁰⁰ or C_FT) and after (C_{_}π⁰ or C_TLP), the maximum modulus of elasticity (ε_leaf), and the relative water content at the turgor loss point (RWC_TLP) were derived from the P-V curves according to [164] and [168].

Predawn and Midday Water Potentials

The same leaves used for gas exchanges measurements were used to determine midday leaf water potentials (Ψ_leaf) after cutting the petiole and sealed inside a plastic bag previously exhalated and stored in ice boxes, then transported to the lab and measured with a pressure chamber (PMS instrument; model 1505D) within 1–2 h of excision [169]. One adjacent leaf was previously covered and allowed to equilibrate for at least 3 h before midday, with cling wrap and aluminium foil to prevent transpiration, to determine the midday branch water potential (Ψ_stem) (i.e. non transpiring leaf). Leaf predawn water potential (i.e. end of the night period) (Ψ_pd) was measured the night before on the same plants. Safety margin (SM_leaf) was calculated as the absolute difference between the turgor loss point (π⁰) and the transpiring leaf water potential at midday (Ψ_leaf). The degree of anisohydry was calculated as the ratio of Ψ_leaf to Ψ_pd.

Hydraulic Conductivity and Huber Value

Loss of hydraulic conductivity and maximum hydraulic conductivity were measured in nine seedlings per treatment at the end of the experiment following [170]. Plant transpiration was stopped 3 h before harvesting by switching off the lights and enclosing plants in black bags. Then pots were immersed in water and the stem was cut at the base. The entire above-ground portion was kept under water, tearing leaves away from shoots, and the plant distal end was cut to ensure xylem relaxation for 30 min [132]. The stem was then sequentially cut back underwater and 5 cm long segment above the cotyledons was connected to a XYL’EM apparatus (Bronkhorst, France) to measure hydraulic conductivity at low pressure (≤ 2 kPa) before (K_init) and after flushing the sample with degassed, filtered 2 mmol KCl solution at high pressure (0.2–0.3 MPa) for 20 min (K_max). Loss of hydraulic conductivity (PLC) was determined as:

$$PLC=100(1-K_{\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{t}}/K_{\mathrm{m}\mathrm{a}\mathrm{x}}),$$

(1)

Maximum specific hydraulic conductivity (K_s) was calculated by dividing K_max by sapwood area and leaf specific conductivity (K_L) by dividing K_max by total supported leaf area. The Huber value (H_V) was determined as the ratio of K_L and K_s (i.e. sapwood area relative to leaf area). Finally, the hydraulic conductance of the whole plant (k_plant) was estimated as in [171]:

$$k_{plant}={\displaystyle \frac{E}{\left({\mathrm{\Psi }}_{pd}-{\mathrm{\Psi }}_{leaf}\right)}}$$

(2)

Where E was the leaf transpiration rate and (Ψ_pd – Ψ_leaf) was assumed to be the water potential gradient from the soil to leaf (MPa), assuming full equilibration of plant and the soil at the end of the night period (i.e. Ψ_soil ≈ Ψ_pd).

Stem Xylem Anatomy

Five stem segments per treatment of those used for hydraulic conductivity measurements were selected for anatomical measurements. One centimetre in length was used to determine wood density (W_D) following [132]. 15 µm-thick crossed and tangential sections were cut using a sliding microtome (Leica SM 2400), stained with 0.1 % safranin and mounted for subsequent image analysis. Digital images of the thin sections were taken using a Moticam A1 camera (Motic, Hong Kong, China) attached to an Olympus BX50 light microscope. Images were analysed using ImageJ [161]. We measured three sectors per sample from the cross-section to quantify average vessel lumen area (V_A), vessel density (V_D), the equivalent circle diameter (D) and the hydraulic diameter (D_H) following [172]:

$$D=\sqrt{{\displaystyle \frac{4V_A}{\pi }}}$$

(3)

$$D_H={\left({\displaystyle \frac{\sum D^4}{N}}\right)}^{{\displaystyle \frac{1}{4}}}$$

(4)

Where V_A is the lumen conduit area, N it is the number of vessel elements. To determine the parenchyma surface area, a further three sectors per sample were taken from the tangential section.

Trait Variation and Plasticity

We calculated the effect of each treatment one by one and through the representation of the variation of each trait measured (see Table 1 with definitions) as a fraction of unity (ranging from -1 to 1), such as negative values meant a decrease in the trait as consequence of the treatment but positive meant an increase. If the trait decreases, or increases, due to the effect of the treatment and compared to the value of the treatment used as reference, we can use one of the next formulations to calculate plasticity:

If treatment mean of a variable (μ_treatment) < reference mean (μ_reference), the plasticity is negative:

$$NP={\displaystyle \frac{\left({\mu }_{treatment}-{\mu }_{reference}\right)}{{\mu }_{reference}}}$$

(5)

If the mean of treatment (μ_treatment) > the mean of reference (μ_reference), the plasticity is positive:

$$NP={\displaystyle \frac{\left({\mu }_{treatment}-{\mu }_{reference}\right)}{{\mu }_{treatment}}}$$

(6)

Note that we can get the individual trait response ratio as:

$${\displaystyle \frac{{\mathrm{\mu }}_{\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}}}{{\mathrm{\mu }}_{\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}}}}=NP+1$$

(7)

Or:

$${\displaystyle \frac{{\mathrm{\mu }}_{\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}}}{{\mathrm{\mu }}_{\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}}}}={\displaystyle \frac{1}{1-PP}}$$

(8)

Where NP and PP are the value of the negative or positive plasticity for the trait respectively.

The Phenotypic Plasticity Index (PPi) was calculated to assess the overall plasticity of Fagus sylvatica seedlings to different environmental conditions, evaluating the effects of growth temperature on light and water plasticity changes, as described by [173]. This form of expressing plasticity is relevant when several traits are analysed together, so we can visualize which traits are more plastic than the others. We calculated the PPi for each trait as the difference between the maximum and minimum mean values ($max\mathrm{\mu }-min\mathrm{\mu }$) divided by the maximum mean value:

$$PPi={\displaystyle \frac{max\mathrm{\mu }-min\mathrm{\mu }}{max\mathrm{\mu }}}$$

(9)

Statistical Analyses

After checking the normality of the data, three-way analyses of variance (ANOVA) were performed in R v. 4.3.2 (R Development Core Team 2024) using 'Tgrowth', 'light' and 'water' as fixed factors. In all cases, the data met the assumptions of normality and homoscedasticity, a significance level of P ≤ 0.05 was used and model assumptions were checked. Statistically significant differences between groups were assessed using the HSD Tukey post-hoc test in R v. 4.3.2 using the "agricolae" package. Principal Component Analysis (PCA) was performed in R version 4.3.2, utilizing the "stats" package for matrix calculations and the "factoextra" package for visualization support. The PCA was based on the correlation matrix, and the eigenvectors generated were analysed to identify variables exhibiting a strong association with specific principal components (PCs). Pearson correlation coefficients were estimated using the “rcorr” function in R. Results are presented as means ± standard error (SE).

5. Conclusions

Our findings underscore the significant impact of elevated temperature on plant growth, biomass partitioning and physiology of European beech seedlings. Specifically, we observed that warming decreased overall plant plasticity, reduced leaf size and allocated higher biomass to leaves while reduced leaf shedding and increased metabolic and water demands, suggesting that global warming could constrain the adaptive capacity of beech regeneration, especially near the drier margins of its distribution. Warming reduced biomass due to lowering total leaf area and net photosynthesis but increased respiratory costs and diminished the potential for beech seedlings to perform osmotic adjustment and decreased the leaf safety margin, which may compromise future survival of the species under dry conditions. Nevertheless, beech growth under shaded conditions may be increased due to an increase in the number of less costly leaves (lower LMA) which enhanced transpiration rate to cool down the leaf if warming occurs under deep soil and wet atmospheric conditions. This highlights the importance of understanding species-specific responses to multifactorial stress factors, which is essential for projecting the long-term persistence and ecological success of forest species under future warming scenarios.