Timing of Drought and Severity of Induced Leaf Desiccation Affect Recovery, Growth and Autumnal Leaf Senescence in Fagus sylvatica L. Saplings

Abstract

Increased water limitations due to climate change will pose severe challenges to forest ecosystems in Europe. We investigated the response of potted saplings of Fagus sylvatica L., one of the major European tree species, to a spring and a summer water-withholding period with control–control (C-C), control–drought (C-D), drought–control (D-C) and drought–drought (D-D) treatments. We focused on recovery capacity and phenological and growth traits and questioned the extent to which an earlier drought influenced the response to a second drought in the same growing season. To examine the impact of the level of drought stress, a distinction was made between saplings with less or more than half of their leaves desiccated due to the spring drought (D<50 and D>50). The timing of the drought influenced the immediate post-drought response: saplings severely affected by the spring drought (D>50) resprouted, whereas saplings severely affected by the summer drought (C-D and D<50-D) did not. The spring treatment influenced the onset of visual symptoms in the summer drought, with saplings less affected in the spring drought (D<50-D) developing symptoms three days later than the saplings not subjected to drought in the spring (C-D), whereas severely affected saplings (D>50-D) had not yet display symptoms seventeen days after the first visual symptoms in the spring control saplings (C-D). The timing of autumnal leaf senescence displayed the legacies of the spring treatment. The saplings heavily affected by the spring drought showed a slower decrease in relative chlorophyll content and delayed leaf senescence (D>50-C and D>50-D), which may enable the repair of damaged tissues. The saplings that were less affected by the spring drought (D<50-C) showed earlier autumnal leaf senescence, which is likely an acclimation response. Interestingly, a larger diameter increment in autumn for all of the saplings that experienced the summer drought (C-D, D<50-D and D>50-D) may indicate the recovery of hydraulic capacity by new xylem growth. Our results underline the plasticity of young F. sylvatica saplings in response to (repeated) drought.

1. Introduction

Forests play a key role in water and nutrient cycles, provide a diverse range of ecosystem services and are essential for the preservation and sustenance of biodiversity [1,2]. Increasing severe weather phenomena such as heat waves, droughts and storms, occurring due to climate change, can lead to increased tree mortality and result in the release of net atmospheric CO₂. Forests face vulnerability to various extreme climate events, with drought and its related disturbances globally exerting the most significant impact [3,4]. It is assumed that trees will probably not be able to adapt to abrupt increases in aridity through evolutionary mechanisms due to their extended generation cycles and the incapacity to migrate away from stressful conditions. The prospect of global warming causing extended and intensified droughts underscores the importance of investigating the impacts of drought on woody plants [4,5].

Trees shedding leaves is a common phenomenon during drought as a life-saving strategy [6]. Leaf shedding typically occurs after stomatal closure [7], and xylem embolisms in the leaves may be the main driver of leaf mortality during drought [8,9,10]. The shedding of leaves in response to drought diminishes the evaporative leaf surface area, thereby assisting woody perennials in postponing the initiation of cavitation in the conductive tissues of the stem [11,12]. Leaf shedding alleviates water stress on the remaining foliage of woody plants, effectively slowing down the rate of desiccation [13,14]. Yet, shedding leaves without complete nutrient resorption results in net nutrient losses, which may limit post-drought resprouting and photosynthesis and have long-term implications for tree performance [15]. Defoliation due to severe drought can lead to partial crown dieback and eventually result in tree mortality in the following years [16]. Rebuilding an affected crown after a severe drought demands additional carbon investment, sourced from non-structural carbohydrate reserves or from the assimilation of remaining or newly grown leaves once drought stress subsides [17].

While considerable focus has been dedicated to identifying the physiological factors leading to tree mortality during drought, it is equally crucial to comprehend the mechanisms involved in the recovery of trees that withstand dry periods [18]. Drought resilience can be quantified by evaluating both the disturbance impact and the rate of recovery following the disturbance [19]. The pace of post-drought hydraulic recovery is primarily determined by the magnitude of the negative water potential reached and the time spent at this extreme [4]. When little or no cavitation (i.e., formation of vapor-filled cavities or air bubbles in the xylem network) has occurred, recovery from drought after rainfall occurs rapidly and stomata open to assimilate new carbon from the atmosphere as plants rehydrate. When cavitation thresholds have been surpassed, the recovery of photosynthesis is considerably slower and proceeds in coordination with the gradual restoration of hydraulic conductance from the roots to the tree crown [20]. When extensive aboveground biomass loss occurs due to drought, the recovery process may be accelerated through resprouting from epicormic buds [20]. A reduction in growth or physiological processes can be compensated for by investment in other processes or tissues and, a process called compensatory recovery [17]. For instance, after drought stress, Fagus sylvatica saplings can delay autumnal leaf senescence, maintaining photosynthesis at a higher level for a longer time [21,22].

In this study, we conducted a controlled recurrent drought experiment on F. sylvatica saplings. We withheld water from potted saplings in the spring and summer of the same growing season, followed by rewatering. In the experiment, we aimed to subject saplings to varying levels of drought stress, ranging from mild to severe, which was assessed based on the visual symptoms that developed during the drought periods. The factorial design allowed the study of the potential influence of former drought exposure. We hypothesized that the responses to the imposed droughts would depend on (i) the timing of the drought in the growing season, (ii) the severity of the drought effects and (iii) whether saplings had already experienced a drought. We were particularly interested in the recovery process and focused on the timing of autumnal leaf senescence and the influence on growth traits.

2. Materials and Methods

2.1. Source Material

We used 288 Fagus sylvatica saplings belonging to ten provenances (Table S1). The ten genetic backgrounds were evenly distributed between the different treatments (see Section 2.2), but no provenance-dependent analyses were performed because of the low number of saplings for each provenance in each treatment group. Vander Mijnsbrugge et al. [23] described the seed collection in 2016, the germination in 2017, the establishment of a common garden of potted seedlings in 2017, and a temperature manipulation experiment with the potted seedlings in 2018. After the temperature treatment in 2018, we moved the potted seedlings to an outdoor container field at the Research Institute for Agriculture and Fisheries in Melle (Belgium), where the saplings were watered using automatic sprinklers. In the winter of 2019–2020, we transferred the saplings to 4 L pots (height 17 cm, diameter 21 cm). All pots were filled with standard nursery potting soil (1.5 kg/m³ NPK 12 + 14 + 24, organic matter 20%, pH 5.0–6.5, E.C. 450 µS/cm and dry matter 25%) without additional fertilizer. The organic matter content and the pH range promote the water-holding capacity of the potting soil, while a dry matter content of 25% indicates a relatively aerated mixture. All saplings remained in the pots during the experiment and had access to the same type and amount of potting soil.

2.2. Drought Treatments

In the middle of May 2022, we moved all potted saplings to a greenhouse, in which an automatic grey shade net protected the saplings from heavy solar radiation. We performed spring and summer drought treatments following a full-factorial design in the growing season of 2022 (Figure 1). At the start and end of both treatments, we brought all saplings to full water saturation: we placed the pots overnight in a water basin with the water level reaching 5 cm above the base of the pots. At the start of the two treatments, we took the pots out of the basin and let any excess water drain naturally over the course of a few hours. During the first treatment (spring), half of the pots—the drought group—did not receive any watering (n = 144), while the other half—the control group—were watered on a regular basis by experienced technical staff, keeping the saplings well-moisturized (n = 144). The ten provenances were evenly (and randomly within each provenance) distributed among the control and drought groups (Table S2). For the second treatment (summer), we split both the control and drought groups of the spring treatment in half, with each half then being the control or drought group in the summer treatment (Figure 1). In addition, the ten provenances were equally (and randomly within each provenance) distributed over the control and drought groups (Table S3). The four treatment groups of the experiment are hereafter called control–control (C-C), control–drought (C-D), drought–control (D-C) and drought–drought (D-D), indicating the combination of the groups in the spring and summer treatments.

The spring treatment lasted from May 23 until June 15, and the summer treatment took place from August 8 until September 12. The spring treatment was ended when there was a distribution of saplings across the different visual drought stress groups (see results), from saplings displaying up to 100% desiccated leaves, as well as saplings without any desiccated leaves. A small number of saplings, for which all leaves were desiccating during the spring drought treatment (15 from a total of 144 saplings), were placed overnight in a water basin on June 13 (n = 7) and June 14 (n = 8) to keep mortality to a minimum and to have a reasonable amount of saplings in all visual drought stress groups (Table S2). To reduce sapling mortality to a minimum during the summer drought treatment, we again placed the saplings that were severely desiccated early on during the treatment in a water basin before the end of the summer drought treatment (Table S4, Figure S1). After the two subsequent treatments, we kept the saplings in the greenhouse until December 2022, ensuring they were consistently maintained under well-watered conditions, as assessed by knowledgeable technical staff. Seven saplings were not alive anymore in the spring of 2023.

2.3. Measurements and Observations

We weighed all the pots at the start of each of the two drought treatments (i.e., after the overnight water basin) and on a weekly basis throughout the treatments. The weight reduction of the pots served as an indicator of water scarcity. For the summer treatment, we calculated the relative weight loss as the difference in weight of a pot between the initial water-saturated condition (on August 8) and its weight on August 22, divided by the initial weight at the saturated condition.

For all the saplings, we measured stem diameter with a measuring rod at 1 cm above the soil level and sapling height up to the place on the stem that was still alive. We performed the measurements at the beginning of the spring drought treatment (May 23), in-between the two drought treatments (August 1) and at the end of 2022 (December 26 for diameters) or the start of the growing season in 2023 (April 25 for heights). Height measurements in December 2022 were not reliable as it only became clear at the time of the spring budburst in 2023 which part of the stem had died off.

To identify the onset of visual drought stress symptoms on the saplings in the spring and summer drought treatments and to track the progress of desiccation, we scored the saplings of the drought groups phenotypically (protocol: see

Table 1. Four scoring protocols used to visually assess the effects of: drought stress on the saplings during the water withholding period [9,24], final leaf desiccation after drought, post-drought recovery and autumnal leaf senescence.

Score	Drought Stress * (% Leaves)	Leaf Desiccation (% Leaves)	Post-Drought Resprouting	Autumnal Leaf Senescence (% Green Leaves)
1	none	none	normal buds	100% (all leaves still green)
2	≤33%	1%–50%	swollen buds starting to open	75%–99% (first leaves turning yellow)
3	34%–66%	51%–75%	first leaves protruding from the buds	50%–75% (combination of green, yellow and first browning leaves)
4	67%–100%	76%–95%	first leaves unfolding	25%–50% (more than half of the leaves turning yellow to brown)
5		>95%	first leaves elongating	<25% (a majority or only yellow and brown leaves)
6				a majority to all leaves brown

* visual signs of drought stress: leaves curling up or becoming more rigid; leaves turning pale green to grey; later on, most of the affected leaves turned brown.

). During the spring treatment, visual drought symptoms became evident on June 10, and we scored all saplings on June 13–15. For the summer drought treatment, we scored the saplings 3 times a week from August 20 until September 12.

Following the first (spring) drought treatment, when it became visually evident which leaves had dried up and which ones had regained turgor, we assessed the proportion of desiccated leaves on the saplings of the drought group on June 21 using a scoring protocol (Table 1). The distribution of the saplings among the different categories of desiccated leaves and their further division among the control and drought groups in the summer treatment are shown in

Table 2. Number of saplings in the different categories of desiccated leaves after the spring drought period (n_D) and their further division among the D-C and D-D treatment groups (n_D-C/n_D-D).

Proportion of Desiccated Leaves	nD	nD-C/nD-D
0%	78	39/39
1%–50%	11	7/4
51%–75%	11	8/3
76%–95%	14	9/5
96%–100%	30	10/20

. The first results indicated that distinguishing saplings with less than 50% desiccated leaves (D<50, consisting of the groups with 0% and 1%–50% leaf desiccation) and saplings with more than 50% desiccated leaves (D>50, consisting of the groups with 51%–75%, 76%–95% and 96%–100% leaf desiccation) was justified (see Figure 2b). On July 14, we evaluated the post-drought recovery of the saplings in the drought group, focusing on buds that initiated new growth (i.e., resprouting). Budburst in woody saplings is normally observed in spring. We adapted a spring protocol for the post-drought new leaf phenology (Table 1). After the second (summer) drought treatment, only a few of the desiccated saplings displayed bud swelling, with a minor number of small leaves that hardly protruded from the buds and did not properly unfold. Therefore, no recovery was scored after the summer drought treatment.

We used a chlorophyll content meter (CCM-200, Edaphic Scientific, Melbourne, Australia) that determines the relative chlorophyll content as the ratio of optical transmission at 931 nm (near-infra-red) to optical transmission at 653 nm (green). We measured the relative chlorophyll content on one representative damage-free and mature leaf on a short shoot at the middle of each sapling during the growing season: May 31, July 4 and 18 (beginning of the spring drought, and two times after the end of the spring drought), as well as on August 1, September 26 and October 24 (before the summer drought, and two times after the end of the summer drought). Saplings that were totally desiccated due to the drought treatments were, quite evidently, not measured.

In the autumn of 2022, we scored leaf senescence in all groups of saplings twice a week from October 26 to November 28, observing the entire sapling and giving a mean score (Table 1). Leaves desiccated due to the previous drought treatments were omitted from the scoring.

2.4. Statistical Analysis

We analyzed all data using the open-source statistical software R, version 4.4.2. [25]. We applied linear and general linear mixed models using the lme4 package [26] and addressed ordinal data with cumulative logistic regression using the ordinal package [27]. Figures were made with ggplot2 [28]. Mixed-effects modeling proved advantageous for the analysis of ecological data due to its ability to accommodate nested data, handle unbalanced datasets, and incorporate random effects [29]. In the context of experimental studies on tree seedlings and saplings, growth traits are often analyzed using (general) linear mixed models, whereas phenological data can be effectively processed with cumulative logistic regression (e.g., [30,31]). The independent variables applied in the fixed parts of the mixed models are explained in

Table 3. Variables used as fixed effects in the mixed models. Abbreviations of the variables are explained in the description using bold text.

Variable	Description	Type
DAY	day of measurement	continuous
DAYcat	categorical time variable for spring or summer treatments	categorical, 3 levels spring: first week (May 31) and after treatment (July 4 and 18) summer: before (August 1) and after treatment (September 26 and October 24)
DES	proportion of desiccated leaves after the spring drought	categorical, 5 levels: none (score 1) ≤50% (score 2) 51%–75% (score 3) 76%–95% (score 4) >95% (score 5)
Dspring	sapling diameter at the start of the spring treatment (May 23)	continuous (in mm)
Dsummer	sapling diameter in-between the two treatments, before the summer treatment (August 1)	continuous (in mm)
Hspring	sapling height at the start of the spring treatment (May 23)	continuous (in cm)
Hsummer	sapling height in-between the two treatments, before the summer treatment (August 1)	continuous (in cm)
TRspring	spring treatment	categorical, 3 levels: C = control D<50 = drought resulting in less than half of the leaves being desiccated D>50 = drought resulting in more than half of the leaves being desiccated
TRfull	full, i.e., combined spring and summer treatments (leaf desiccation categorized as in TRspring)	categorical, 6 levels: C-C = control–control C-D = control–drought D<50-C and D>50-C = drought–control D<50-D and D>50-D = drought–drought

. In the presented model formulas, each β is an indication of the effect of each independent variable on the response variable. A unique plant identifier was added as a random effect to account for repeated measures on the same saplings when applicable.

2.4.1. Recovery After the Spring Drought Period

We examined the relationship between the occurrence of post-drought resprouting after the spring treatment and the severity of the experienced drought stress. For this, we looked at the saplings in the drought group and studied the effect of the level of leaf desiccation and initial sapling height on the recovery by resprouting (RES). We used cumulative logistic regression to model the probability of having resprouted up to a given resprouting level (p_R), i.e., having reached a certain resprouting level or a lower level:

log(p_RES/(1 − p_RES)) = β₀ − β₁DES − β₂H_spring

2.4.2. Onset of Visual Drought Symptoms in the Summer Treatment

We examined whether the onset of visual drought symptoms in the summer treatment (ODR_summer) was dependent on the effects of the spring treatment. During the summer treatment, none of the saplings in the drought group that had more than half of their leaves desiccated due to the spring drought (D>50-D) showed visual drought symptoms before the end of the treatment. Hence, we used only part of the drought group, i.e., the saplings with less than half of their leaves desiccated after the spring treatment (D<50-D), and all controls from the spring treatment (C-D) for the modeling. We first transformed the scores of visual symptoms into binary data. All saplings without visual drought symptoms were given a score of 0, and all others scored 1. We fit a logistic regression with p_ODRsummer, showing the probability of displaying drought symptoms on a given day during the summer drought treatment:

log(p_ODRsummer/1 − p_ODRsummer) = β₀ + β₁H_summer + β₂TR_full + β₃DAY

The difference in time between the onset of the visual drought symptoms in the C-D and D<50-D groups was calculated based on the coefficients of the model above and using the mean summer sapling height.

2.4.3. Relative Chlorophyll Content

We used the measurements of the relative chlorophyll content (RCC) in the leaves for two purposes. Firstly, we quantified the experienced drought stress caused by the spring drought, by comparing the relative chlorophyll content of the spring drought group with the controls. Secondly, the relative chlorophyll content measurements in summer and autumn completed the picture of the autumnal leaf senescence process together with the visual observations.

The RCC_spring model accounted for the measurements in the first week of the spring drought treatment (May 31) and the two measurements after the treatment (July 4 and 18):

RCC_spring = β₀ + β₁ H_spring + β₂TR_spring + β₃DAY_cat + β₄TR_springDAY_cat

By including the interaction term between DAY_cat and TR_spring, the model allowed for the possibility that the change in RCC_spring over time (DAY_cat) differed among the treatment groups (TR_spring).

The relative chlorophyll content at the beginning of (August 1) and after (September 26 and October 24) the summer treatment (RCC_summer) was modeled with the same model structure as above but with H_summer and TR_full.

2.4.4. Autumnal Leaf Senescence

To examine the effect of the treatments on autumnal leaf senescence (SEN), the probability of having attained—on a given day—a given leaf senescence score or a lower score (p_SEN) was modeled by employing a generalized linear mixed model (cumulative logistic regression):

log(p_SEN/(1 − p_SEN)) = β₀ − β₁TR_full − β₂DAY

The timelapse between autumnal leaf senescence for the control–control group and the different spring and summer treatments was calculated based on the coefficients of the fitted model.

2.4.5. Change in Sapling Heights and Diameters

To evaluate the effects of the drought in each treatment on sapling growth, we modeled the change in sapling height and diameter for all the saplings, taking into account the initial height and diameter. We calculated the height and diameter increment by subtracting the height and diameter on May 23 from those on August 1 for the spring treatment (HI_spring and DI_spring), and by subtracting the height and diameter on August 1 from those on April 25, 2023 (height) or December 26, 2022 (diameter) for the summer treatment (HI_summer and DI_summer). We applied a linear model for the change in height caused by the spring treatment:

IH_spring = β₀ + β₁ H_spring + β₂TR_spring

For the modeling of the change in height caused by the summer treatment (IH_summer), the same model structure was applied with H_summer and TR_full. For the changes in diameter (DI_spring and DI_summer), the initial heights were replaced by initial diameters (D_spring and D_summer).

3. Results

3.1. Spring Treatment: Drought Symptoms, Recovery and Relative Chlorophyll Content

In the spring treatment, the reduction in weight of the pots demonstrated the water scarcity experienced by the saplings (Figure 2a). In the group of saplings that had experienced drought, saplings for which more than half of the leaves had desiccated (categories 51%–75%, 76%–95% and 96%–100%) resprouted significantly more after the drought in comparison to the saplings without leaf desiccation due to the drought (Figure 2b, Table S5). The categories of 51%–75%, 76%–95% and 96%–100% post-drought desiccated leaves are grouped and designated as D>50, whereas the categories of 0% and 1%–50% desiccated leaves are grouped and designated as D<50.

The relative chlorophyll content increased less or diminished in the saplings of the drought group of the spring treatment when compared to the controls (significant interaction terms between the date, i.e., July 4, and treatment groups are shown in Table S6, Figure 3). Later on in July, the relative chlorophyll content in the saplings with more than half of their leaves desiccated due to the drought (D>50) diminished relatively quickly (significant interaction terms between the date, i.e., July 18, and treatment groups are shown in Table S6, Figure 3). The noticeably different reaction of the relative chlorophyll content in the leaves of the D<50 and D>50 groups further justified the division of the group that experienced spring drought in the two subgroups, i.e., with more or less than half of their leaves desiccated due to the drought.

3.2. Summer Treatment: Drought Symptoms, Recovery and Relative Chlorophyll Content

The weight loss of the pots during the summer treatment demonstrated the water scarcity experienced by the saplings (Figure 4). The pot’s weight showed variability that could be related to the spring treatment. The pots of saplings with more than half of their leaves desiccated due to the spring drought (D>50-D) lost weight less quickly than the pots of saplings with less than half of their leaves desiccated (D<50-D) and the control plots of the spring treatment (C-D). On August 22 (the first saplings were taken out of the drought to avoid mortality on August 24), the mean relative weight loss of the pots in the D>50-D treatment was 30.8% ± 5.3%, which was significantly higher than the relative weight loss of the pots in the D<50-D (53.9% ± 5.4%) or C-D (60.5% ± 5.6%) treatment groups.

The control–drought saplings (C-D) and the drought–drought saplings with less than half of their leaves desiccated due to the spring drought (D<50-D) developed symptoms before the end of the summer treatment (Figure S1 and Figure 5). The difference in the onset of visual drought symptoms between C-D and D<50-D was significant (Table S7), with the D<50-D group displaying visual drought symptoms 3.3 days later. The saplings on which more than half of the leaves had desiccated due to the spring drought, as well as those in the drought group of the summer treatment, did not display any drought symptoms by the end of the summer treatment (D>50-D in Figure 5). Therefore, it could be deduced that the onset of visual drought symptoms in this group would have taken place more than 17 days later than in the C-D group.

3.3. Timing of Autumnal Leaf Senescence

The timing of leaf senescence was studied based on two methods: measurements of relative chlorophyll content in the leaves and visual observations of leaf decoloration. When comparing the relative chlorophyll content before and after the summer drought period, the two groups of saplings with more than half of their leaves desiccated due to the spring drought (D>50-C and D>50-D) showed a rise in relative chlorophyll content whereas the control–control group showed a decline (significant interaction terms between the date, i.e., September 26, and treatment groups are shown in Table S6, Figure 6). When looking at the change in relative chlorophyll content after the summer treatment, the saplings that had more than half of their leaves desiccated due to the spring drought (D>50-C and D>50-D) displayed a diminishment that was less steep in comparison to the C-C group (significant interaction terms between the date, i.e., October 24, and treatment groups are shown in Table S6, Figure 6).

Among the drought–control saplings, the group with less than half of their leaves desiccated due to the spring drought (D<50-C) decolored 3.3 days earlier than the control–control (C-C) group, while the saplings on which more than half of the leaves had desiccated due to the spring drought decolored 7.6 (D>50-C) and 9.3 (D>50-D) days later (Table S8, Figure 7). It should be noted that the D<50-D group lost the majority or all of their leaves due to the summer drought and did not resprout afterwards, and no relative chlorophyll content measurements nor leaf decoloration observations could be performed on this group of saplings.

3.4. Changes in Height and Diameter

The spring drought caused a smaller height increment between May 23 and August 1 in the group of saplings with less than half of their leaves desiccated due to the spring drought (D<50) compared to the controls (C), while the height of saplings with more than half of their leaves desiccated (D>50) was reduced due of the top parts dying off (Figure 8a, significant treatment groups are shown in Table S9). During the summer drought, we observed no change in height between August 1 and April 25, 2023 in the control–control group (C-C) and several treatment groups (D<50-C, D>50-C and D>50-D) (no significant treatment groups in Table S9), while sapling height decreased in the control–drought group (C-D) and in the drought–drought group with less than half of their leaves desiccated due to the spring drought (D<50-D) due to the top parts dying off (Figure 8b, significant treatment groups are shown in Table S9).

The spring drought did not cause significantly larger or smaller diameter increments (between May 23 and August 1) for the different treatment groups in comparison to the control (no significant treatment groups in Table S9). Interestingly, all the sapling groups that experienced drought in the summer treatment (C-D, D<50-D and D>50-D) displayed a larger diameter increment (between August 1 and December 26) in comparison to the C-C group (Figure 8c, significant treatment groups are shown in Table S9). The modeled diameter increment during the second part of the growing season, i.e., after August 1, was 17% of the diameter increment of the entire growing season for the control saplings (group C-C), 47% for the saplings that only experienced the summer drought (group C-D), 31% for the saplings with less than half of their leaves desiccated after the spring drought and subjected to the summer drought (group D<50-D) and 65% for the saplings that were deeply affected by the spring drought and subjected to the summer drought (D>50-D).

Six out of the seven saplings that were not alive anymore in the spring of 2023 belonged to the control–drought group (C-D). The deceased saplings displayed a larger diameter increment than the C-C group between August 1 and December 26 (Figure S2, significant treatment group in Table S10), similar to the surviving saplings in the C-D group.

3.5. Summary Modeling Results

The extent of leaf desiccation following the spring drought treatment (D<50 vs. D>50) significantly affected subsequent processes, including the likelihood of new shoot formation during the recovery phase, the onset of visual drought symptoms in the summer treatment, changes in relative chlorophyll content throughout the growing season, the timing of autumnal leaf senescence and the occurrence of partial stem dieback and growth reduction (Figure 9). Notably, diameter growth was stimulated by the summer drought treatment, regardless of whether saplings had experienced drought in the spring treatment (Figure 9).

4. Discussion

Leaf wilting and shedding are commonly observed responses of F. sylvatica to water shortage during the growing season [16,32]. In our experiment, the wilting and curling of the first leaves indicated the onset of a rapid progression towards wilting in the majority of sapling leaves, as evident in the spring and summer drought treatments.

4.1. Post-Drought Resprouting

The phenomenon of post-drought resprouting has been well-documented across various woody species [20,24]. Resprouting, whether after drought or fire, appears to be influenced more by the severity of the stress rather than by the type of stress [17,18,20]. In our experiment, we observed a correlation between the formation of new shoots in saplings following the spring drought and the degree of leaf loss due to desiccation. Saplings with more than half of their leaves desiccated exhibited significantly higher rates of resprouting. Interestingly, in the recovery phase after the summer drought treatment, we observed no resprouting. This underscores that a sapling’s resprouting capability not only hinges on the severity of drought stress but also on the timing of the drought within the growing season.

Priming, i.e., the process by which a phenotype adapts to environmental pressures, instigates stress memory, which enhances a plant’s response to future stress [33,34]. Mild drought stress typically prompts reversible stomatal closure and osmotic regulation without significantly impacting growth or other essential processes [18], whereas intense drought stress, triggering active repair mechanisms and profound metabolic changes, can serve as a priming event, leading to stress memory [17,18]. In our experiment, losing less than half of their foliage to desiccation due to the spring drought may be categorized as mild stress, although it still resulted in a few days’ delay in the onset of visual drought symptoms during the summer drought compared to the control–drought group (i.e., D<50-D vs. C-D). Losing more than half of their foliage due to the spring drought caused a delay in symptom onset that was at least five times longer during the summer drought (D>50-D vs. C-D). The ability to withstand the summer drought hence depended on the level of stress experienced in the spring drought. It can be hypothesized that saplings with extensive post-drought leaf desiccation in spring (D>50) achieved stomatal closure during the summer drought earlier than the saplings that were in the control group during the spring treatment (C-D). Previous studies on beech saplings have indicated that stomatal conductance remains reduced following severe drought, thereby enhancing intrinsic water use efficiency [35].

4.2. Timing of Autumnal Leaf Senescence

Following the summer treatment, the relative chlorophyll content in saplings with more than half of their leaves desiccated due to the spring drought (D>50-C and D>50-D) remained lower than in the double control group (C-C). As these saplings resprouted, newly formed leaves most probably did not have sufficient time to develop higher levels of chlorophyll content. However, the rate of decrease in chlorophyll content was significantly slower in these saplings compared to the double control group. This trend continued in autumnal leaf decoloration. Saplings heavily impacted by the spring drought (D>50-C and D>50-D) exhibited delayed autumnal leaf senescence. This post-drought delay in leaf senescence has already been documented in F. sylvatica [21,22], as well as in other species such as Quercus petraea (Matt.) Liebl. and Frangula alnus Mill. [30,36]. Delayed leaf senescence may prolong photosynthesis, potentially compensating for carbon losses caused by drought and aiding in a plant’s recovery [22]. In contrast, saplings in which less than half of the leaves desiccated due to the spring drought and which were in the control group during the summer treatment (D<50-C) exhibited significantly earlier leaf decoloration compared to saplings that experienced no drought (C-C). Both earlier and later timing of autumnal leaf senescence in the same experiment has already been documented in a controlled drought experiment on Cornus sanguinea L. [24]. In the study, C. sanguinea saplings with minimal to no visual drought symptoms exhibited earlier senescence compared to control saplings. Conversely, saplings displaying pronounced visual symptoms experienced a delayed senescence, while those with intermediate symptoms did not show any significant deviation from the control group. These contrasting trajectories of post-drought leaf senescence, observed in our experiment with F. sylvatica, corroborate the findings in the C. sanguinea study. It was already hypothesized that these observations may concern a trade-off between risk-minimizing acclimation (resulting in earlier senescence) and life-saving repair of drought-induced injuries (leading to later senescence) [18].

4.3. Drought Affects Sapling Size

The extent of leaf desiccation resulting from the spring drought had a significant impact on sapling height. Those with more than half of their leaves desiccated (D>50) were shorter after the spring drought due to the loss of the upper parts of their stems. The summer treatment impacted the height of the saplings in the D<50-D and C-D groups, similarly resulting in reduced height. Partial crown dieback following drought is common in adult F. sylvatica trees [16]. More interestingly, all saplings subjected to drought during the summer treatment (C-D, D<50-D and D>50-D) displayed a higher diameter increment in the later part of the growing season, from before the summer treatment until the season’s end, compared to the control group (C-C). In climatologically normal years, adult F. sylvatica tress typically exhibit peak radial growth around the end of May or the beginning of June [37]. By the end of June, these trees have completed approximately three-quarters of their annual radial growth [38]. Cambial cell divisions typically cease by the end of July or mid-August [38,39]. It can be hypothesized that in our experiment, the late-season radial growth represented a specific response to hydraulic impairment. Research has suggested that the primary response to hydraulic impairment in trees is the regeneration of the xylem through active vascular cambium [40] and not the refilling of embolized conduits [4]. Interestingly, the six saplings from the C-D group that did not survive the experiment also exhibited late-season radial growth, suggesting that they did not die from hydraulic impairment caused by the summer drought but rather from carbon starvation resulting from the demands of the post-drought radial growth repair [41].

5. Conclusions

The timing of the drought events within the growing season significantly influenced the recovery through resprouting, growth and the timing of autumnal leaf senescence in F. sylvatica saplings. Our results suggest stress memory, particularly when saplings experienced significant foliage loss due to spring drought. Earlier leaf senescence (e.g., 3 days earlier for a group of saplings that experienced spring drought) can be viewed as an acclimation strategy aimed at preventing impairments from putative new drought events. Delayed leaf senescence (e.g., 7 days later and longer for different groups of drought-treated saplings) likely provided essential time for recovery from drought-induced damage. This result likely reflects a trade-off between minimizing risks and prioritizing necessary life-saving repair processes. Saplings subjected to summer drought did not resprout but formed additional secondary xylem: they performed up to 65% of their diameter increment in the second part of the growing season compared to 17% for the control plants. Our findings contribute to a deeper understanding of forest responses to the anticipated increase in drought frequency.