The role of stem respiration in cold-acclimation of winter-dormant Quercus robur trees under large air temperature fluctuations

Abstract

Cold acclimation of leaf respiration often increases respiration rates at a given temperature after temperature declines. Whether a similar response occurs in stems and contributes to frost tolerance remains unknown. We evaluated stem CO₂ efflux (E_s) and cold acclimation in dormant Quercus robur trees in a climate chamber with day/night air temperatures progressively decreasing week-by-week from 20/10 °C to 5/−10 °C and then progressively increasing back to 20/10 °C over 9 weeks. Cold hardening was evidenced by lower cell damage at sub-freezing temperatures compared with trees kept in a greenhouse at 21.5/18.5 °C. A hysteresis between E_s and stem temperature (T_s) showed a time lag between E_s and metabolic activity during chilling and sub-freezing weeks. Stem CO₂ efflux was strongly related to T_s down to −10 °C. At a standardized T_s of 5 °C, E_s decreased quickly in response to decreases in temperature, suggesting a rapid down-regulation of respiratory metabolism to enhance cryoprotection. Accordingly, soluble sugar concentrations increased with decreasing T_s. However, upon subsequent warming from sub-freezing temperatures, stems continued dehydrating and did not recover previous respiration rates, suggesting freeze–thaw embolism and cell damage reduced the apparent ‘respiratory capacity’ of the stem. Moderate frost tolerance in the cold-hardened plants appeared to depend on stem dehydration and soluble-sugar accumulation, not increased stem metabolic activity.

Reduction in stem metabolic activity in response to decreasing temperatures enables the accumulation of soluble sugars and facilitates moderate frost tolerance in cold-hardened, dehydrated stems of oak.

Introduction

Low winter temperatures outside of the tropics constrain tree growth for several months. During winter dormancy, however, deciduous trees are not inactive. Sensing and responding to winter temperature fluctuations involves coordinated metabolic adjustments that allow trees to survive frosts, avoid premature cambial reactivation, and mobilize nutrients and carbon (C) reserves to resume growth in spring when the risk of frost damage is lower (Lavigne et al., 2004; Strimbeck et al., 2015; Pagter et al., 2017). Mitochondrial respiration plays a crucial role in orchestrating such metabolic adjustments and enabling trees to withstand cold and frost (Atkin and Tjoelker, 2003).

Rapid changes in respiration with temperature depend on the thermal sensitivity of respiratory enzymes. The intercept and slope of short-term respiration temperature-response curves—representing the basal respiration rate and its temperature sensitivity (Q₁₀), respectively—can adjust to sustained changes in ambient temperature. This process, known as thermal acclimation of respiration, responds mostly to changes in the respiratory demand and the availability of respiratory substrates occurring over warming/cooling periods (Atkin and Tjoelker, 2003; Atkin et al., 2005; Heidarvand et al., 2017). It can occur within hours to days and change reversibly over successive changes in temperature (Atkin et al., 2000; Bolstad et al., 2003; Lee et al., 2005; but see Armstrong et al., 2006). Such metabolic reorganization has significant implications for plant fitness, carbon dioxide (CO₂) release into the atmosphere, and climate–vegetation modeling (Atkin and Tjoelker, 2003; Atkin et al., 2008; Zhang et al., 2025). Thermal acclimation of respiration has been reported in leaves (e.g. Slot and Kitajima, 2015), roots (e.g. Loveys et al., 2003), and more recently in stems (Smith et al., 2019; Zhang et al., 2025). Smith et al. (2019) suggested that non-photosynthetic stems of different species have similar respiratory demands and similar rates of stem respiration (R_s) between 20 °C and 35 °C. At chilling and sub-freezing temperatures, lower respiratory demand and low-temperature restrictions to enzymatic activity usually result in low R_s rates (Zha et al., 2004; Gruber et al., 2009; Rodríguez-Calcerrada et al., 2014; Darenova et al., 2018; Zhao et al., 2018); however, the respiratory demand can increase near freezing temperatures in some cases. For example, in detached branches of several tree species (Sperling et al., 2015) and intact saplings of Pistacia integerrima trees (Sperling et al., 2017), R_s near 0 °C increased relative to warmer temperatures, a response that was related to energy-requiring frost protection mechanisms.

Most of what we know about respiratory metabolism in cold and frost conditions comes from studies on leaves. However, the demand and use of respiratory products vary among organs in relation to their different functions. In leaves, the cross-talk between mitochondria and chloroplasts coordinates respiratory metabolism with the maintenance of photosynthetic capacity (Loveys et al., 2003; Nunes-Nesi et al., 2008). In woody stems, cambial and parenchymatic cells mostly rely on mitochondrial energy to fuel secondary growth during the growing season and to maintain ion gradients, protein turnover, membrane stability, and antioxidant production during the winter-dormant season (Amthor, 2000; Sperling et al., 2017). Despite the fact that maintenance of these cellular processes might require the thermal acclimation of the stem respiratory metabolism to facilitate cold and frost tolerance (Amthor, 2000), few studies have measured R_s during the non-growing winter season and, to the best of our knowledge, none has specifically evaluated thermal acclimation of R_s in response to fluctuating winter temperatures. Based on the behavior of leaves (e.g. Larigauderie and Körner, 1995; Atkin et al., 2000; Talts et al., 2004; Heidarvand et al., 2017), cold acclimation of stem metabolism could manifest as a fast and reversible increase in R_s rates at a given measurement temperature in plants exposed to colder temperatures due to increased R_s basal rates and/or Q₁₀. While tissue dehydration and water redistribution between the symplast and the apoplast or between stem tissues passively favor cold acclimation (Zweifel et al., 2010; Charrier et al., 2013; Arias et al., 2017), other processes require increased respiration rates to enhance frost protection (Sperling et al., 2017). Cold-induced acclimatory shifts in gene expression lead to the accumulation of metabolites and proteins involved in osmotic adjustment and detoxification of reactive oxygen species (e.g. proline and soluble sugars), cell membrane stabilization (e.g. dehydrins), and ice-crystal binding (e.g. antifreeze proteins). These processes increase the degree of cold-hardening (García Bañuelos et al., 2008; Strimbeck et al., 2015; Heidarvand et al., 2017), but can also increase the demand for respiratory products. Similarly, cambial reactivation preceding cell division and growth during warming in spring can increase R_s rates (Lavigne et al., 2004; Gruber et al., 2009; Chan et al., 2018; Rodríguez-Calcerrada et al., 2019). The recovery of cell turgor and the accompanying transcriptional and metabolic transition from cell winter-dormancy to active growth can induce cold dehardening, rendering trees particularly vulnerable to frost (Vitasse et al., 2014; Pagter et al., 2017; Vitra et al., 2017; Vyse et al., 2019).

Likewise, stem photosynthesis (P_s) might contribute to cold acclimation of stem maintenance metabolism during winter (see Smith et al., 2019). Young stems with thin bark layers can have considerable rates of P_s, which contribute to growth during the vegetative period, hydraulic integrity under drought, and preservation of C reserves when leaf photosynthesis is limited or absent (Pfanz et al., 2002; Ivanov et al., 2006; Steppe et al., 2015; De Baerdemaeker et al., 2017; De Roo et al., 2020). However, it is not known how P_s changes during winter dormancy as temperatures and R_s change. The responses of P_s and R_s to temperature during winter dormancy can affect the concentration of C reserves and the fractions of soluble and insoluble non-structural carbohydrates (NSCs), especially in winter-deciduous trees. Nonetheless, winter NSC dynamics are largely driven by stem sink strength and energy demand, with higher respiration rates during warmer winters leading to greater depletion of C reserves (Ögren et al., 1997) and reduced C availability for spring earlywood growth (Tixier et al., 2019; Petrik et al., 2024). Storage compounds such as starch and lipids in fat-storing species often accumulate in stem tissues by autumn or early winter (Kramer and Kozlowski, 1979; Ashworth et al., 1993). The subsequent accumulation of soluble sugars (SSs) needed for cold acclimation of stem parenchyma cells often comes from starch hydrolysis (Ashworth et al., 1993; Kasuga et al., 2007; Charrier et al., 2013; Strimbeck et al., 2015; Sperling et al., 2017; Furze et al., 2019). Similarly, changes in concentrations of starch and free amino acids such as asparagine, arginine, and proline run opposite in response to cold, suggesting mobilization of C reserves supplies hexoses for osmoregulation and biosynthesis of cryoprotective nitrogenous compounds in leaves and stems (Sperling et al., 2017; Zanotto et al., 2023). In contrast, other studies have reported fairly constant concentrations of SSs and starch in woody tissues of winter-deciduous trees (e.g. Quercus petraea; Hoch et al., 2003).

The objective of this current study was to better understand the coordinated role of stem energy and C metabolism in cold acclimation. To this end, leafless Quercus robur trees were grown in a climate chamber for 11 weeks at air temperatures between 20 °C and −10 °C during the winter and the beginning of the spring season. During this time, stem diameter variations and stem CO₂ efflux (E_s) were continuously monitored. On a weekly basis, we quantified E_s rates at 5 °C (E_{s_5}) and at the mean weekly air temperature (E_{s_amb}), as well as Q₁₀ and P_s. In addition, we measured NSC concentrations, and electrolyte leakage and malondialdehyde levels as indicators of frost-induced cell membrane damage (e.g. Yu et al., 2017; Ding et al., 2024). Our first hypothesis was that E_{s_5} increases as respiratory energy demand increases; that is on exposure to chilly and sub-freezing temperatures over the first weeks, and to progressively higher temperatures accompanying the resumption of spring growth later on. Our second hypothesis was that periods of ‘spring’ frost following relatively warm periods cause more cell membrane damage than periods of ‘winter’ frost following a progressive decline in temperatures. Finally, our third hypothesis was that concentrations of low-molecular-weight compatible solutes involved in cryoprotection and osmoregulation, such as soluble sugars and proline, increase at the cost of starch reserves as temperatures decline.

Material and methods

Experimental design

The experiment was conducted using the facilities of Ghent University, Belgium. In late January 2020, 12 Quercus robur trees cultivated outdoors under an open-sided rain shelter were transferred to a walk-in climate chamber (Wekk 10.40.8L, Weiss Technik, Tiel, the Netherlands) with a 10/14 h photoperiod (08.00 h–18.00 h) and subjected to controlled variations in air temperature over an 11 week period. Half of the trees were used for continuous and non-destructive monitoring of stem temperature, stem diameter variations, and stem CO₂ efflux (E_s; Group 1; Supplementary Fig. S1), whilst the other half were used for destructive harvesting of branches once a week for biochemical measurements (Group 2). Thus, effects of wounding on E_s were avoided in Group 1. At the same time, an additional third group of four control trees were moved to a greenhouse and kept at 20–23 °C during the day and at 17–19 °C during the night (Group 3).

At the beginning of the study, the mean (±SE) plant height was 1.81 m (±0.06) and the stem-base diameter was 1.94 cm (±0.1), with no significant differences among the three groups (P>0.25). The day/night temperatures in the climate chamber were changed weekly, progressively decreasing from 20/10 °C to 5/−10 °C during the first 5 weeks, then increasing back to 20/10 °C over the following 4 weeks, after which they were suddenly decreased to 5/−10 °C for 1 week, and finally increased back to 20/10 °C for a week (Fig. 1A). The rate of temperature change between the day and night periods was 2 °C h⁻¹, with temperatures starting to decrease at the end of the photoperiod (18.00 h) and starting to increase at 3.00 h (5 h before the end of the night). On one night each week, the night-time air temperature was programmed to remain the same as the daytime temperature in order to calculate the rate of stem photosynthesis (see below). Photosynthetic photon flux density was measured continuously at 0.5 m height (the same at which stem CO₂ efflux cuvettes were installed) using a SQ-110-SS quantum sensor (Apogee Instruments, Inc., UT, USA); it varied from 78–95 μmol m⁻² s⁻¹ in colder and warmer weeks, respectively (Fig. 1B). The concentration of CO₂ ([CO₂]) inside the chamber averaged 598 ppm (Fig. 1C); it was higher than ambient [CO₂] due to the plant and soil respiration under the ventilation conditions inside the chamber. Day/night relative humidity varied from 75/90% to 80/95% over the 11 weeks of the experiment.

Fig. 1.

Ambient conditions inside the climate chamber and stem CO₂ efflux (E_s) at hourly intervals over the 11 weeks (w1–w11) of the experiment. (A) Air and stem temperature (T). Stem T was slightly higher than air T. (B) Photosynthetic photon flux density (PPFD). (C) Air CO₂ concentration. (D) E_s, expressed as per unit stem volume. Gaps in the data reflect temporary failures in data logging.

Sampling for biochemical variables and electrolyte leakage (see below) was conducted at the end of each week, between 09.00–9.30 h. One branch per tree was cut with pruning shears, and branches were selected at different heights across trees and weeks to avoid any effect of sampling position over time.

All plants were regularly watered throughout the experimental period.

Stem CO₂ efflux and temperature

Stem CO₂ efflux (E_s) was measured as a proxy of stem respiration (R_s). Transparent cuvettes were placed around the stem at ∼5 cm above the soil level (Supplementary Fig. S1). The cuvettes were ∼16 cm long and ∼4 cm wide, and were custom-made by sticking a transparent flexible polycarbonate film to two foam gaskets encircling the stem at both cuvette ends. The film was sealed along its length to form a cylinder and non-caustic silicon sealant was applied to the foam ends to prevent CO₂ leakage (tested by breathing onto the chamber and detecting no fluctuation in measured CO₂). Two plastic tubes tightly sealed with screw-nuts and O-rings to two holes punched in the film were connected to a recently calibrated infrared gas analyser (IRGA; LI-7000 CO₂/H₂O Analyzer; LI-COR), allowing measurement of the difference in CO₂ between air entering and leaving the cuvette. A pump was used to achieve an airflow rate of 1.3±0.4 ml s⁻¹ across the cuvettes. Before entering the cuvette, air from the climate chamber was mixed in a 50 l buffer container to ensure a stable CO₂ concentration. A custom-made multiplexer switched CO₂ measurements sequentially among the six cuvettes every 5 min. After each 30 min cycle, air coming from an additional cuvette enclosing a cylinder of PVC was analysed to account for potential measurement drift of the IRGA. Data were stored every 2 min in a Campbell CR1000 data logger. Technical problems with one stem cuvette reduced the final data set to five trees.

Stem temperature (T_s) was continuously measured with a thermocouple inserted 5 mm into the stem, at ∼10 cm above the cuvette. Immediately before starting the experiment, a 1 mm diameter hole was drilled to ensure a tight fit of the thermocouple, assuming this minor injury would not affect R_s.

Stem diameter variations

A linear variable displacement transducer (LVDT; model DF5.0; Solartron Metrology, Leicester, UK) was installed on each tree, at ∼5 cm above the stem cuvette, using custom-made stainless steel frames (Steppe and Lemeur, 2004; Supplementary Fig. S1). The stem diameter variation (ΔD) was recorded continuously, and the weekly ΔD was calculated as the difference in the maximum value in one week compared with the previous week.

Non-structural carbohydrate and proline concentrations

We measured branch non-structural carbohydrates (NSCs) without separating the xylem and bark tissues to estimate C reserves following a modified protocol of Maness (2010). First, 50 mg of ground dried tissue was incubated in 1 ml of hot 80% ethanol to extract soluble sugars (SSs). This step was repeated three times to guarantee complete extraction from the tissues. Then, starch present in the insoluble material left after centrifugation was extracted and transformed into glucose monomers using the enzymes amyloglucosidase from Aspergillus niger at 50 U ml^–1 (Sigma-10115) and α-amylase from Bacillus licheniformis at 500 U ml^–1 (Sigma-A4582) in 0.1 M sodium acetate, pH 4.5. The SSs and digested starch were quantified using anthrone–sulfuric acid colorimetric microassays, measuring the absorbance at 630 nm in 96-well microplates (Laurentin and Edwards, 2003). We used a standard curve with known concentrations of glucose/fructose/galactose (50:30:20) for the quantification of SSs (10–1000 ppm) or glucose (10–2000 ppm) in the case of starch. The concentration of total NSCs was calculated by summing the concentrations of SSs and starch.

For the quantification of proline, the protocol of Carillo and Gibon (2011) was followed using the ninhydrin assay adapted to a 96-well format. First, 50 µl of ethanolic extract from the SS extraction was mixed with 100 µl of 1% ninhydrin, 60% acetic acid, and 20% ethanol and incubated at 90 °C for 30 min in the dark. Then, the absorbance of the reaction mix was measured at 520 nm. A standard curve of L-proline (0.2–5 mM) was used for the quantification of the total free proline content in the samples.

Assessment of cell membrane integrity

We measured malondialdehyde (MDA) concentration and electrolyte leakage to evaluate cell damage in the same branches used for the NSC and proline measurements.

Malondialdehyde is a by-product of lipid peroxidation and is used as an indicator of oxidative stress (Ortega-Villasante et al., 2005). For the determination of MDA concentration, 100 mg of ground dried tissue was transferred to a screw-capped 1.5 ml Eppendorf tube and homogenized with 1 ml reaction solution [15% (w/v) trichloroacetic acid, 0.37% (w/v) 2-thiobarbituric acid, 0.25 M HCl, and 0.01% (w/v) buthylated hydroxytoluene]. After incubation at 90 °C for 30 min, the samples were centrifuged at 12 000 g for 10 min. The absorbance of the supernatant was measured at 535 nm and 600 nm (the latter to correct for non-specific turbidity) and the concentration of MDA was calculated using an extinction coefficient of 1.56×10⁵ M⁻¹ cm⁻¹ (Ortega-Villasante et al., 2005).

As an index of cell damage, electrolyte leakage was measured in 2 cm branch segments, including bark and xylem, placed in a sample tube containing 20 ml of distilled water. After 24 h at constant room temperature, the initial conductivity of the solution (C_i) was measured with a multirange conductivity meter (Hanna Instruments HI 9033; probe HI 76302). The solution was then returned to the tubes with the samples and boiled for 1 h to kill all cells and ensure complete electrolyte leakage. When the solution reached room temperature, the resulting maximum conductivity (C_max) was measured. The relative electrical conductivity (RC) was calculated as $\mathrm{RC}=(C_{\mathrm{i}}/C_{\max })\times 100$.

As a comparison with potentially non-cold-acclimated trees, MDA and RC were also measured on the four control trees of Group 3 in weeks 1, 4, and 5 (Fig. 1A). In the first week, these trees were kept in the greenhouse at relatively constant air temperatures of 20–23 °C during the day and 17–19 °C during the night. At the end of the third week, they were moved to the climate chamber and exposed to the same day/night temperature regime as the rest of the trees, namely 5/−5 °C in week 4 and 5/−10 °C in week 5. Thus, the measurements in weeks 4 and 5 were made in trees that had not been cold-acclimated. (At the end of week 5 these control trees were moved back to the greenhouse with the intention of sampling them again in the climate chamber in weeks 10 and 11; however, this part of the study was curtailed by the outbreak of the COVID-19 pandemic.)

Data analyses

Online visualization and preliminary analyses of data were performed using the PhytoSense software (http://www.phytosense.net/). Stem CO₂ efflux (E_s, µmol m⁻² s⁻¹) was calculated following the standard equation:

$$E_{\mathrm{s}}=\frac{\mathrm{\Delta }\mathrm{C}{\mathrm{O}}_2\times P}{\mathrm{R}\times T}\times \frac{f_{\mathrm{c}}}{S}$$

where ΔCO₂ is the difference in [CO₂] (ppm) exiting and entering the stem cuvette, P is the ambient pressure (in atmospheres, atm), R is the universal gas constant (82.057 atm ml mol⁻¹ K⁻¹), T is air temperature (K), f_c is the measured flow through the cuvette (ml s⁻¹), and S is the axial surface area of the stem segment (m²) (Salomón et al., 2021). The length and diameter of the stem enclosed in the cuvette were measured to calculate S. The E_s rates were then calculated on a stem volume basis (µmol m⁻³ s⁻¹) to account for differences in stem diameter among trees, considering the absence of heartwood in the xylem of these small-sized oaks.

For each tree and each week, E_s was calculated at a standard temperature of 5 °C and at the mean weekly temperature. For this, we used the following equation:

$$E_{\mathrm{s}\mathrm{\_}i}=E_{\mathrm{s}\mathrm{\_}0}{Q}_{10}^{\frac{T_{\mathrm{s}\mathrm{\_}i}}{10}}$$

where E_{s_i} is E_s at either 5 °C (E_{s_5}) or the mean weekly air temperature (E_{s_amb}), E_{s_0} is E_s at a reference temperature of 0 °C, Q₁₀ is the temperature sensitivity of each tree at each week, and T_{s_i} is either 5 °C or the mean weekly temperature. Only night-time values were considered for analyses to exclude the effect of P_s on E_s (De Roo et al., 2020). We also discarded the first night of the week when temperatures were changed, and the night-time of the third day, which was programmed to equal the daytime temperatures to calculate P_s. As there was a hysteretic relationship between T_s and E_s during night-time (Supplementary Fig. S2), we compared the coefficient of determination (r²) of the linear relationship between the natural logarithm of E_s and T_s recorded at the same time and up to 6 h earlier (at 30 min intervals), and used the (lagged) T_s that gave the highest r² to calculate Q₁₀, E_{s_5}, and E_{s_amb} (see Rodríguez-Calcerrada et al., 2014). The magnitude of the night-time hysteresis was calculated as:

$$\mathrm{Hysteresis}=\frac{E_{\mathrm{s}\mathrm{\_}\mathrm{amb}\mathrm{\_}\mathrm{early}}-E_{\mathrm{s}\mathrm{\_}\mathrm{amb}\mathrm{\_}\mathrm{late}}}{E_{\mathrm{s}\mathrm{\_}\mathrm{amb}\mathrm{\_}\mathrm{late}}}\times 100$$

where E_{s_amb_early} and E_{s_amb_late} are the E_{s_amb} values calculated from decreasing temperatures (early night-time, 18.00–23.00 h) and increasing temperatures (late night-time, 03.00–08.00 h), respectively.

Stem photosynthesis (P_s) was estimated using data from the third day each week, when temperatures were set as constant over the 24 h period:

$$P_{\mathrm{s}}=E_{\mathrm{s}\mathrm{\_}\mathrm{night}}-E_{\mathrm{s}\mathrm{\_}\mathrm{day}},$$

where E_{s_night} and E_{s_day} were E_s at night-time and daytime, respectively, calculated as the mean E_s over the central 9 h of each period. Thus, weekly P_s was calculated at daytime ambient temperature and light conditions across the 11 experimental weeks.

We compared the means of all variables among weeks using a repeated-measures general linear model, accounting for the non-independence of trees repeatedly sampled over time, and considering week as a fixed factor. We also tested the relationship between week number (1–11) and weekly means of variables to explore their variation over time. Linear, quadratic, and exponential regression models were fitted to assess the relationships among variables (i.e. r² and significance). We used ANCOVA to test for differences in slopes between the periods preceding and following week 5 of severe frosts (5/−10 °C). Analyses were carried out using the Statistica 8.0 software (StatSoft Inc.). The raw data obtained in this study are available in Supplementary Dataset S1.

Results

Diel changes in E_s across weeks

Stem CO₂ efflux (E_s) varied by almost 40-fold across weeks (Fig. 1D), mostly as an exponential function of T_s (Fig. 2). On a daily basis, there was a hysteresis between E_s and T_s, with higher E_s at the beginning of the night than at the end (Supplementary Fig. S2). The magnitude of the hysteresis was higher during the coldest weeks (up to 50–60%) and was barely noticeable during the warmer weeks (down to 10%) (Fig. 3A, D). The hysteresis disappeared when E_s was represented against T_s recorded before E_s measurements (Supplementary Fig. S2). The time lag between E_s and T_s was highest during the fourth and sixth weeks, reaching up to 160 min, when trees were exposed to sub-freezing temperatures of −5 °C (Fig. 3B, E). At week 5, when the minimum T_s reached −10 °C, the time lag decreased to ∼80 min. The pattern of variation of the increase in the goodness of fit observed when using lagged T_s instead of simultaneous T_s (computed by the change in r²) was similar to that of the time lag (Fig. 3C, F). Variability in all these metrics (i.e. SE) was higher in the weeks with sub-freezing temperatures.

Fig. 2.

Relationship between stem CO₂ efflux (E_s) and stem temperature (T_s) at hourly intervals during 11 weeks of varying air temperatures (w1–w11). (A) Values of E_s versus T_s recorded at the same time. (B) Values of E_s versus the value of T_s recorded either 30, 60, 90, 120, 150, or 180 min earlier depending on the best fit between the two variables (see text). The inset in panel (B) illustrates the proposed hypothetical relationship between E_s and T_s with decreasing ambient air temperatures (T_amb) and increasing T_amb and growth relative to initial conditions at week 1. Symbols represent the means of five replicate trees (error bars not shown for clarity).

Fig. 3.

Parameters describing the hysteresis of stem CO₂ efflux (E_s) against stem temperature (T_s) during 11 weeks of varying air temperatures. Different colours correspond to values observed at decreasing air temperatures (first weeks) and increasing air temperatures (last weeks). (A) Magnitude of the hysteresis of E_s, representing the difference between early- and late-night E_s at a common T_s (see text). (B) Lagged response of E_s to T_s (see text). (C) Improvement in the goodness of fit of the relationship of E_s versus T_s after considering the lagged response. Dashed lines show the weekly minimum T_s. Different letters indicate significant differences among means as determined using Tukey’s HSD test, when factor time was significant at P<0.05. (D–F) The relationships of the variables shown in (A–C) with the weekly minimum T_s; **P<0.01, ***P<0.001. Linear and quadratic regression models were adjusted when significant at P<0.1. All data are means (±SE), n=5.

Weekly changes in stem respiratory and photosynthetic parameters

From the lagged responses of E_s to night-time T_s we calculated weekly values of Q₁₀, E_s at a standard measurement T_s of 5 °C (E_{s_5}), and E_s at prevailing ambient conditions (E_{s_amb}) as close surrogates of stem respiratory characteristics (Fig. 4). The Q₁₀ of E_s did not differ significantly over time, and it was not related to weekly ambient T_s (Fig. 4A, D). At weeks 5 and 10, when minimum air temperatures reached −10 °C, Q₁₀ tended to be (non-significantly) higher than in the other weeks, and it was more variable among trees. Stem CO₂ efflux standardized at 5 °C decreased over time, and this was faster over the first 5 weeks of the experiment when T_s was progressively decreasing. The relationship of E_{s_5} with T_s was strongly linear over these first weeks, and weak and non-linear after week 5 (interaction time × ‘severe frost period’ P<0.05; Fig. 4B, E). Thus, the highest values of E_{s_5} were observed in the first weeks and the lowest in the final weeks (time P<0.001). Stem CO₂ efflux at prevailing ambient conditions also varied over time (P time<0.001) in close association with T_s. The slope of the exponential relationship between E_{s_amb} and T_s was higher before week 5 than afterwards (interaction mean T_s × ‘severe frost period’ P<0.01; Fig. 4C, F). For comparison with other studies, E_{s_amb} values expressed per unit stem surface area rather than volume ranged from 1.81±0.19 μmol m⁻² s⁻¹ in week 1 to 0.23±0.06 μmol m⁻² s⁻¹ in week 10.

Fig. 4.

Respiratory parameters during 11 weeks of varying air temperatures. Different colours correspond to values observed at decreasing air temperatures (first weeks) and increasing air temperatures (last weeks). (A) Sensitivity of stem CO₂ efflux (E_s) to stem temperature (T_s), expressed as the Q₁₀ value. (B) Stem CO₂ efflux calculated at 5 °C. (C) Stem CO₂ efflux calculated at the mean T_s. Dashed lines show the weekly minimum T_s. Different letters indicate significant differences among means as determined using Tukey’s HSD test, when factor time was significant at P<0.05. (D–F) The relationships of the variables shown in (A–C) with the weekly minimum T_s; *P<0.05, **P<0.01, ***P<0.001. Linear and quadratic regression models were adjusted when significant at P<0.1. All data are means (±SE), n=5.

Stem photosynthesis (P_s) was low (<0.1 μmol m⁻² s⁻¹), showed much greater variability among trees than E_s, and did not vary significantly over time or with minimum T_s (P>0.15; Fig. 5). P_s was positively correlated with E_{s_amb} (r²=0.51, P<0.05) and with mean weekly T_s (r²=0.46, P<0.05). However, it was methodologically unavoidable that PPFD decreased slightly but significantly with decreasing air temperatures, from 95 μmol photons m⁻² s⁻¹ in warmer weeks to 78 μmol photons m⁻² s⁻¹ in the coldest weeks (r²=0.88, P<0.001; see Fig. 1A, C). Thus, the positive relationship between P_s and mean T_s could partly reflect a positive effect of PPFD on P_s. For comparison with E_{s_amb}, P_s per unit stem volume ranged from 14.1±4.1 μmol m⁻³ s⁻¹ in week 1 to 3.8±10.8 μmol m⁻³ s⁻¹ in week 5, excluding weeks 3, 4, and 6 where mean P_s was negative. On average, P_s accounted for 3.8% of E_{s_amb}, with a maximum of 5.7% in week 5.

Fig. 5.

Stem photosynthesis (P_s) during 11 weeks of varying air temperatures. Different colours correspond to P_s values observed at decreasing air temperatures (first weeks) and increasing air temperatures (last weeks). (A) Variation of P_s and the minimum stem temperature (T_s; dashed line) with time, and (B) the relationship of P_s with T_s (not significant). All data are means (±SE), n=5. No data were obtained for week 7 because of technical problems in the climate chamber.

Weekly changes in stem diameter

Stem diameter decreased over the course of the experiment (Fig. 6A). By the end of the experiment, stems had shrunk by 0.3 mm relative to initial values, with trees showing increasing cumulative variability in ΔD over time. The greatest decreases in diameter occurred in weeks 1 and 9, the weeks with the warmest air temperatures. Thus, weekly diameter shrinkage was positively related with T_s (Fig. 6B), P_s (r²=0.88, P<0.001) and E_{s_amb} (r²=0.87, P<0.001).

Fig. 6.

Stem diameter changes (ΔD) during 11 weeks of varying air temperatures. Different colours correspond to values observed at decreasing air temperatures (first weeks) and increasing air temperatures (last weeks). (A) Cumulative ΔD relative to the initial diameter, showing shrinkage of the stems. The dashed line shows the weekly minimum stem temperature (T_s). Different letters indicate significant differences among means as determined using Tukey’s HSD test, when factor time was significant at P<0.05. (B) Relationship of weekly ΔD (i.e shrinkage relative to preceding week) and minimum T_s. A quadratic regression model was adjusted for this relationship; ***P<0.001. All data are means (±SE), n=5.

Weekly changes in NSCs and proline

The concentrations of starch and total non-structural carbohydrates (NSCs) did not vary significantly over time. However, the concentration of soluble sugars (SSs) did vary significantly (P<0.05) and so did the fraction of NSCs represented by SSs (i.e. SS/NSC; P<0.001; Fig. 7), even though the restrictive Tukey’s HSD test did not separate the means among weeks significantly in the case of SSs. The concentrations of SSs were negatively correlated with minimum T_s after excluding values from the tenth week, when trees were suddenly exposed to sub-freezing temperatures (i.e. presumably did not have time to acclimate; Fig. 7A, D). The concentration of SSs increased by ∼35% from the warmest to the coldest weeks. A marginally significant relationship was also found between SS/NSC and minimum T_s (P<0.1), indicating a larger fraction of SSs in the NSC pool at colder T_s. The concentration of proline did not vary significantly over time (P>0.15; Supplementary Fig. S3). However, weekly changes in proline and SSs were similar, with both variables showing a marginal positive correlation across weeks (r²=0.31, P=0.075).

Fig. 7.

Concentrations of non-structural carbohydrates (NSCs) in branches during 11 weeks of varying air temperatures. Different colours correspond to values observed at decreasing air temperatures (first weeks) and increasing air temperatures (last weeks). (A) Soluble sugars (SS), (B) starch, and (C) the fraction of SS relative to total NSCs (SS/NSC); total NSC=SS+starch. Dashed lines show the weekly minimum stem temperature (T_s). Different letters indicate significant differences among means as determined using Tukey’s HSD test, when factor time was significant at P<0.05. (D–F) The relationships of the variables shown in (A–C) with the weekly minimum T_s. Linear regression models were adjusted when significant at P<0.10. The correlation in (D) excludes the values in week 10 (‘spring frost’). ^†P<0.10, *P<0.05. All data are means (±SE), n=5.

Weekly changes in cell-membrane integrity

Relative conductivity (RC) and MDA concentrations varied significantly over time (P<0.05; Fig. 8) and were not related to T_s (P>0.1). Unexpectedly, weekly mean RC values were generally high (>50%), suggesting that boiling the stems samples in water for 1 h was not enough to kill all cells and/or allow for electrolyte release through the bark surface area (Kovaleski and Grossman, 2021). The highest RC for plants continuously kept in the climate chamber (Group 2) occurred in the last week. In contrast, MDA was lowest in the last weeks, which resulted in a negative correlation between RC and MDA (r²=0.43, P<0.05).

Fig. 8.

Variation in indicators of cell-membrane damage in branches during 11 weeks of varying air temperatures. Different colours correspond to values observed at decreasing air temperatures (first weeks) and increasing air temperatures (last weeks). Open symbols correspond to trees kept in a greenhouse at roughly constant air temperatures (21.5/18.5 °C day/night) until the end of week 3 before being moved to the climate chamber with the rest of the trees for weeks 4 and 5. Dashed lines show weekly minimum stem temperature (T_s) in the climate chamber. (A) Relative electric conductivity (RC) and (B) concentration of malondialdehyde (MDA). Different letters indicate significant differences among means for the blue and red symbols as determined using Tukey’s HSD test, when factor time was significant at P<0.05. Significant differences between grey and blue symbols: ^†P<0.10, *P<0.05, ***P<0.001. All data are means (±SE), n=5. No data were obtained for RC in week 8 because of access restrictions due to COVID-19. For MDA, at week 5, two symbols are overlapped.

Compared with trees that were continuously kept in the climate chamber (exposed to progressively colder temperatures), control trees moved from the greenhouse (kept at roughly constant temperatures) to the climate chamber (Group 3) had significantly higher RC values and MDA concentration in the fourth week, when all trees were exposed to −5 °C. Differences disappeared the next week, when both sets of trees were exposed to −10 °C. Non-significantly higher MDA in greenhouse trees was observed in the first week, when they had not yet been moved to the climate chamber.

Discussion

It has been considered that variable E_s versus T_s response curves over time reflect thermal acclimation of R_s (Gansert et al., 2002; Rodríguez-Calcerrada et al., 2014; Darenova et al., 2018; Smith et al., 2019). However, previous studies have been conducted in conditions where variable growth rates, xylem sap flow rates, or soil water availability prevent the unequivocal attribution of changes in E_s versus T_s response curves to thermal acclimation of maintenance R_s. Our experiment, conducted on well-watered, leafless, undamaged dormant trees, aimed to assess the extent of cold acclimation of R_s and its contribution to cellular freezing tolerance.

Weekly variations in E_{s_5} indicated that R_s acclimated to changes in T_s quickly and partly reversibly, although not in the expected direction. Notably, although sub-daily temperature fluctuations were greater than and partly overlapped with weekly temperature changes, plants exhibited a change in E_{s_5} within one day after the new weekly thermal regime was applied (see Fig. 1A, D). This suggests that respiratory acclimation was driven by minimum rather than mean temperatures. Within the minimum temperature range of 10 °C to −10 °C, 5 °C changes can be critical for cellular metabolism due to the non-linear effect of temperature on cellular processes, particularly near the freezing point (Sperling et al., 2017). Atkin et al. (2005) concluded that leaf and root respiration track daily minimum/night temperatures more closely than daily mean temperatures, aligning with the broader view that plant physiology acclimates to extreme rather than average climatic cues (Reyer et al., 2013). The rapid change in stem respiration rates at a set measurement temperature (E_{s_5}) observed here (Fig. 4) matches previous results in leaves (e.g. Lee et al., 2005), even if some studies mention longer periods for acclimation (i.e. weeks to months; Armstrong et al., 2006) or no acclimation at all (Slot and Kitajima, 2015). However, two observations were unexpected when compared to previous respiration studies in leaves and roots: E_{s_5} decreased with decreasing T_s, and E_s deviated from R_s at a sub-daily scale.

E _s response to lagged T_s

Estimating R_s from E_s is not straightforward due to several biophysical processes affecting the radial diffusion of respired CO₂, including axial transport of CO₂ dissolved in xylem sap, recycling of CO₂ via woody-tissue photosynthesis, and light-driven axial diffusion of CO₂ (Ryan et al., 1995; Steppe et al., 2007; Teskey et al., 2008; De Roo et al., 2019; Salomón et al., 2024). These processes vary over time depending on environmental conditions such as stem hydration and prevailing irradiance. Here, the hysteresis between T_s and E_s (Supplementary Fig. S2) could hardly be attributable to these alternative fates of CO₂, as it occurred during night-time. Even if daytime conditions had exerted a carry-over effect on early-nighttime E_s—either via transient post-illumination respiration peaks (Werner et al., 2011) or via daytime respiratory depression (Tcherkez et al., 2017)—such effects were probably negligible because the daytime light intensity in the climate chamber was low. Thus, we consider two explanations for the hysteresis between E_s and T_s. First, and more likely, it could reflect a temporal uncoupling of E_s from R_s. The radial CO₂ efflux could change independently of R_s at sub-freezing temperatures due to ice formation in the stem, mostly in the larger xylem conduits and the extracellular spaces (Neuner et al., 2010; Lintunen et al., 2020). Ice propagation would force CO₂ out of the stem by diffusion as its concentration builds up in the ice front (Sevanto et al., 2012), which could have increased E_s early at night-time as T_s dropped below 0 °C. Once ice forms, it can slow down CO₂ radial diffusion and the rate of E_s before thawing occurs at dawn (Lintunen et al., 2014). These processes might explain the highest night-time hysteresis of E_s versus T_s observed in the weeks with freezing temperatures. However, the hysteresis was also appreciable in non-freezing weeks, especially in chilling weeks. Although resistance to CO₂ diffusion has been observed across a wide range of temperatures (Ryan et al., 1995; Steppe et al., 2007), low T_s reaching 0 °C and 5 °C might enhance such effects due to cold-induced reductions in cell-membrane fluidity, permeability, and CO₂ diffusion. Second, the fact that E_s was lower later in the night than earlier might indicate depletion of respiratory substrates, as reported for leaves (Jones et al., 2024). Nevertheless, stem metabolic activity appears to be more sink-driven than source-driven, at least seasonally (Salomón et al., 2022).

Unusual peaks in CO₂ efflux have been observed in leaves (Atkin et al., 2000) and stems (Lintunen et al., 2014; Sperling et al., 2015) at or near sub-freezing temperatures. Here, E_{s_amb} and T_s were strongly correlated both above and below 0 °C over time, although the mean and SE of Q₁₀ tended to increase at weeks 5 and 10 with −10 °C frosts (Fig. 4). Whether an anomalous release of CO₂ reflects a metabolic response or a biophysical process linked to expanding ice remains an open question (Lintunen et al., 2020).

Cold acclimation of R_s and freezing tolerance

Cold-induced changes in the transcriptome, proteome, and metabolome imply wide readjustments of metabolic pathways that can demand an increase in respiratory products to enhance cell frost resistance (Xin and Browse, 2000; Kaplan et al., 2004; Welling and Palva, 2006; Strimbeck et al., 2015; Pagter et al., 2017). Thus, based on previous studies on thermal acclimation of leaf respiration (e.g. Atkin et al., 2000; Atkin and Tjoelker, 2003), our first hypothesis was that the respiratory sensitivity (Q₁₀) and/or respiration rates at a set T_s (E_{s_5}) would increase with colder temperatures. However, on the contrary, decreasing air temperatures over weeks 1–5 translated into roughly constant Q₁₀ and a progressive decrease in E_{s_5}. It must be noted that the trees were probably cold-hardened when the experiment started. Outdoor trees were moved to the climate chamber in late January, and Repo et al. (2008) found that boreal Q. robur populations can start hardening in September. Thus, our results suggest that once cold-hardened, Q. robur stems do not need to enhance their metabolic activity to face colder temperatures. Instead, cold-induced decreases in apparent ‘respiratory capacity’ at a set T_s (i.e. E_{s_5}  sensu  Atkin et al., 2000) enhanced cold-induced reductions in respiratory rates (i.e. E_{s_amb}), probably as a dynamic, passive mechanism of frost resistance linked to saving SSs (Ögren, 1996; Ögren et al., 1997; Zha et al., 2004). Reducing respiration (and the activity of the cytochrome pathway) would also enhance cold resistance by lowering the generation of reactive oxygen species that would potentially damage cell membranes, as discussed in fruits (Jiang et al., 2023).

On the other hand, we expected that the trees would resume growth as air temperatures increased from the fifth week onwards and that, because of metabolic costs associated with growth respiration, stems would then exhibit higher E_s at a given T_s by the end of the experiment. Again, this expectation was not fulfilled. We observed lower E_{s_5} and lower sensitivity of E_{s_5} and E_{s_amb} to increasing T_s during the last weeks of the experiment, when T_s increased after the severe frosts of week 5 (Fig. 4E, F). This could reflect unusually high respiratory activity during the first week in the climate chamber under spring-like temperatures (20/10 °C), consistent with de-acclimation in buds of woody species resulting in increased respiration rates (García Bañuelos et al., 2008; Kjaer et al., 2019). Another explanation for our observations is that the trees responded to progressively warmer temperatures (in weeks 6–9) by reducing stem maintenance respiration (Gansert et al., 2002; Rodríguez-Calcerrada et al., 2014). This downward shift in the long-term E_s versus T_s response curve would reduce the consumption of C substrates, thereby extending the cold-hardening period (Ögren, 1996) while preserving substrate availability for growth resumption in spring (Tixier et al., 2019; Petrik et al., 2024). Alternatively, and more likely, the attenuated response of E_s to increasing T_s after the repeated −5 °C and −10 °C frosts could reflect damage in the respiratory apparatus and the oncoming death of the stem. Two aspects of our observations support this possibility: the continuous decline in stem diameter (Fig. 6A) and the stem dieback observed after the experiment concluded (Supplementary Fig. S4). The diameter shrinkage suggests continuous dehydration unrelated to cold acclimation. Despite the recovery of air temperatures from the sixth week onwards, the stems did not rehydrate and resume growth. They dehydrated continuously over the experimental period, faster in warmer weeks probably because of higher evaporative demand and bark transpiration (Pfanz et al., 2002), and because of water redistribution to other organs, such as buds (Charrier et al., 2013) or roots that were probably spared from sub-freezing temperatures (Sakai and Larcher, 1987). The absence of growth and bud-break might also have resulted from the constant photoperiod and low light levels limiting warming effects in Q. robur growth and leaf phenology (Morin et al., 2010). Nonetheless, we think that the variable temperature regime during the experiment reduced frost hardiness and made the trees of this ring-porous oak particularly sensitive to accumulating frost spells. It is possible that trees partly de-acclimated over the first week in the climate chamber (20/10 °C), re-acclimated to colder conditions of the following weeks 2–4 (10/0 °C and 5/−5 °C), and de-acclimated again as temperatures increased back to 20/10 °C (Fig. 1). In accordance with this, de-hardened woody tissues of Q. robur and other temperate tree species show freezing damage at −10 °C (Morin et al., 2007; Charrier et al., 2013). One month after moving all the trees outdoors at the end of the experiment, we observed that some had not resprouted or had only resprouted from the base of the stem (Supplementary Fig. S4), suggesting that embolisms caused by dehydration and the repeated freeze–thaw cycles prevented xylem transport and import of soil water (Davis et al., 1999; Martínez-Vilalta and Pockman, 2002; Mayr et al., 2003; Cavender-Bares, 2005; Sevanto et al., 2012; Lintunen et al., 2018). In the absence of major CO₂ bursts below sub-freezing temperatures, it could form large bubbles during freezing, blocking the vessels on thawing (Lintunen et al., 2014, 2018). Cell death by freezing and impaired metabolic activity induced by stem dehydration in the remaining live cells (Xin and Browse, 2000; Charrier et al., 2018) would have reduced E_s towards the end of the experiment (Fig. 4C).

The trees exhibited some symptoms of cold hardening and de-hardening upon changing temperatures, as assessed via electrolyte leakage and MDA concentration (Fig. 8). According to our second hypothesis, we expected that the −10 °C winter frost (week 5) would cause less damage than the −10 °C spring frost (week 10) and more damage in trees that were kept in the greenhouse before the winter frost, and overall this was supported by our results. On one hand, the spring frost was more harmful than the winter frost (Charrier et al., 2013; Pagter and Arora, 2013), as reflected by the changes in RC experienced by trees in the climate chamber (Group 2). On the other hand, the control trees that were kept at roughly constant temperatures in the greenhouse (Group 3) suffered more from the impact of the winter frost than the trees exposed to progressively decreasing temperatures in the climate chamber (Fig. 8). Moreover, the higher MDA concentrations in the control Group 3 that were already apparent in the first week might reflect regulation of gene expression during early de-acclimation of trees in the greenhouse (Schmid-Siegert et al., 2012).

NSC dynamics as related to frost hardiness and R_s

We currently know little about how stem NSCs, R_s, and P_s change during winter. Our results showed that P_s decreased during colder weeks (Fig. 5), corresponding with a reduction in E_{s_amb} (Fig. 4). This relationship is consistent with the well-documented correlation between R_s and photosynthetic capacity, both within and across species (Cernusak and Marshall, 2000; Berveiller et al., 2007). Overall, P_s was low due to the low temperatures, which were far from the temperature optimum at >20 °C (Cernusak and Marshall, 2000), and the continuously low light levels, which never exceeded 100 μmol m⁻² s⁻¹ PPFD (Fig. 1) and were probably near the light compensation point (Linder and Troeng, 1980; Pfanz et al., 2002; Dukat et al., 2024). Thus, P_s could only have a modest effect on mitigating the depletion of SSs over time (Fig. 7), as it only represented a maximum of 5% of E_s at ambient conditions. The low light values, representative of winter understorey conditions in high-latitude forests, were probably insufficient to saturate P_s, which under saturating light can attain higher rates and lead to higher CO₂ refixation rates (e.g. 31–90% in one-year-old stems; Pfanz et al., 2002; Berveiller et al., 2007; Teskey et al., 2008; De Roo et al., 2020). Thus, combined low temperatures and low light during winter can make P_s a secondary contributor to the plant C balance and NSC dynamics compared to R_s.

The concentration of SSs and their fraction within the total pool of NSCs both increased with decreasing T_s (Fig. 7), supporting our third hypothesis. This indicates a direct effect of low temperatures on carbohydrate metabolism and composition. The cold-induced up-regulation of SSs is a widespread response to enhance frost tolerance, for example by reducing the freezing point of sap, favoring extracellular ice formation, and stabilizing cell membranes (Cavender-Bares, 2005; Morin et al., 2007; Charrier et al., 2013, 2018). In dehydrated stems, the accumulation of SSs can have a synergic effect on frost hardiness (Charrier et al., 2013, 2018). Severe dehydration in shrunken cells accumulating SSs can reduce the water potential and limit the metabolic activity to very low, but still non-negligible values during sub-freezing periods (Charrier et al., 2018). Moreover, the strong relationship between T_s and E_{s_amb} (Fig. 4) supports the fact that cold temperatures have an indirect positive effect on SS concentrations by reducing the rates of sugar consumption, a relationship that has often been discussed (e.g. Ögren, 1996), but rarely tested.

Conclusion

There is very little information on E_s during sub-freezing periods (Gansert et al., 2002; Zha et al., 2004). The literature on plant respiratory behavior during cold and frost conditions is heavily skewed toward leaves. In part, this is due to the difficulty in attributing E_s directly to R_s. The need to account for the time-lag between R_s and E_s when studying stem physiology has been highlighted previously (e.g. Ryan et al., 1995; Teskey et al., 2008). We have extended this consideration to cold and sub-freezing conditions, under which resistance to CO₂ diffusion appears to increase.

Stem respiration rates (E_{s_amb}) decreased markedly from 20 °C to −10 °C within a few weeks, enhanced by a cold-induced decline in the apparent ‘respiratory capacity’ (E_{s_5}). Because the trees used for the experiment were probably already cold-hardened (Repo et al., 2008) and evidenced further hardening during the experimental period (Fig. 8), our results suggest that frost tolerance in cold-hardened plants is a passive process mediated by dehydration and accumulation of SSs that does not require increased metabolic activity. This conclusion can be reconciled with previous literature (on leaves) suggesting that acclimation to cold stress increases energy requirements and consequently R_s (e.g. Larigauderie and Körner, 1995; Atkin et al., 2000; Talts et al., 2004; Heidarvand et al., 2017). Cell metabolic reorganization occurring when trees are exposed to low autumn light and temperatures (not captured in our experiment) would firstly enhance the stem respiratory demand. Following this active process of cold acclimation, and once the trees were cold-hardened, passive acclimation to low temperatures would imply a reduction in metabolic activity, thereby saving SSs for use in frost resistance and future growth. Longer experiments extending from autumn to spring are needed to test this dual role of R_s in frost tolerance.

Continuous changes in R_s rates at a set T_s in cold-hardened trees complicate the prediction of winter R_s rates from short-term E_s versus T_s responses, especially when trees experience frost damage. Temperate oaks with large vessels such as Q. robur can be sensitive to repeated freeze–thaw events causing embolism formation, particularly when undissolved CO₂ becomes trapped in ice. Future temperatures will probably be higher and more erratic during winter in Northern Europe (Benestad, 2005; Tomczyk et al., 2019; Ruosteenoja et al., 2020). The fast and strong response of E_s to T_s observed here for Q. robur trees suggests that warm winter spells will increase the respiratory use of C reserves, feeding forward on CO₂-induced climate change and reducing frost resistance and next-spring growth in winter-deciduous species.

Abbreviations

ΔD

stem diameter variation

E _s

stem CO₂ efflux

E _{s_5}

E _s at 5 °C

E _{s_amb}

E _s at the mean weekly air temperature

MDA

malondialdehyde

NSC

non-structural carbohydrate

P _s

stem photosynthesis

Q ₁₀

temperature sensitivity of respiration

RC

relative electrical conductivity

R _s

stem respiration

SS

soluble sugar

T _s

stem temperature

Supplementary data

The following supplementary data are available at JXB online.

Fig. S1. Image of the experimental set-up and materials.

Fig. S2. Hysteresis between stem CO₂ efflux and current and lagged stem temperatures.

Fig. S3. Variations in proline concentrations across the 11 weeks of the experiment.

Fig. S4. Image of stem-dieback of some trees after the end of the experiment.

Dataset S1. Experimental raw data.

Author contributions

JR-C, RLS, and KS: conceptualization, methodology; JR-C and RLS: formal analysis; JR-C, RLS, JS-P, and KS: investigation, writing—review & editing; JR-C: writing—original draft; KS: resources and funding acquisition.

Funding

This work was supported by the Spanish Ministry of Science, Innovation and Universities through the mobility programme ‘José Castillejo’ (CAS19/00130). RLS acknowledges funding from the Research Foundation Flanders (FWO), European Union’s Horizon 2020 research and innovation programme (Marie Skłodowska-Curie grant agreement no. 665501), and the Spanish Ministry of Science, Innovation and Universities (Ramón y Cajal Programme, grant no. RYC2021-032467-I). KS acknowledges funding from the FWO (research project G063720N) and the Special Research Fund (BOF) of Ghent University, Belgium (basic infrastructure funding, grant 01B04515). This work was partly supported by project PID2022-137378NB-I00 funded by Spanish Ministry of Science, Innovation and Universities.

Data availability

The data that support the findings of this study are available from the corresponding author, Jesús Rodríguez-Calcerrada, upon request.