Thermal de-acclimation: how permanent are leaf phenotypes when cold-acclimated plants experience warming?


O. Atkin. Fax: +01904-432-860; e-mail:


We quantified a broad range of Arabidopsis thaliana (Col-0) leaf phenotypes for initially warm-grown (25/20 °C day/night) plants that were exposed to cold (5 °C) for periods of a few hours to 45 d before being transferred back to the warm, where leaves were allowed to mature. This allowed us to address the following questions: (1) For how long do warm-grown plants have to experience cold before developing leaves become irreversibly cold acclimated? (2) To what extent is the de-acclimation process associated with changes in leaf anatomy and physiology? We show that leaves that experience cold for extended periods during early development exhibit little plasticity in either photosynthesis or respiration, and they do not revert to a warm-associated carbohydrate profile. The eventual expansion rate in the warm was inversely related to the duration of previous cold treatment. Moreover, cold exposure of immature/developing leaves for as little as 5 d resulted in irreversible changes in the morphology of leaves that subsequently matured in the warm, with 15 d cold being sufficient for a permanent alteration of leaf anatomy. Collectively, these results highlight the impact of transitory cold during early leaf development in determining the eventual phenotype of leaves that mature in the warm.


The majority of the earth's land surface experiences annual minimum temperatures near or below freezing (Larcher 2004). Beyond equatorial regions, temperature typically follows pronounced seasonal trends and can vary considerably within seasons, potentially compromising plant growth and survival (Loveys, Egerton & Ball 2006; Stuart et al. 2007; Gu et al. 2009). Understanding how sustained and transient changes in temperature affect plant function is therefore crucial for predicting the productivity of crops, forests and natural ecosystems, both now and in the future, warmer world (Woldendorp et al. 2008).

Many annuals and tropical plants are chilling-sensitive, with exposure to low temperatures leading to widespread necrosis. For example, cotton is irreversibly damaged by 48 h exposure to chilling temperatures (Sellschop & Salmon 1928), and chilling temperature during anthesis causes male sterility in rice (Imin et al. 2006; Mamun et al. 2006). However, during prolonged exposure, many other plants are able to alleviate unfavourable effects of low temperature by cold hardening/acclimating (Thomashow 2001; Stitt & Hurry 2002; Atkin et al. 2006). Associated with the cold hardening/acclimation process are large-scale changes in the transcriptome (Fowler & Thomashow 2002; Chinnusamy, Zhu & Zhu 2007; Robinson & Parkin 2008), metabolome (Cook et al. 2004; Gray & Heath 2005) and proteome (Goulas et al. 2006; Herman et al. 2006). Increases in enzyme activity (Guy, Huber & Huber 1992; Holaday et al. 1992; Strand et al. 1999) also occur. When first exposed to cold, pre-existing (PE) leaves of warm-grown plants often accumulate dry matter. This includes large pools of sugars and starch, which peak after a few days of cold and then subsequently decrease as the PE leaves cold acclimate (Ristic & Ashworth 1993; Cook et al. 2004; Kaplan et al. 2004). Importantly, sugars remain higher in cold-acclimated PE than warm-developed (WD) leaves (Koster & Lynch 1992; Ristic & Ashworth 1993; Strand et al. 1997; Gorsuch, Pandey & Atkin 2010) because of increases in the abundance and activity of synthetic enzymes (Guy et al. 1992; Holaday et al. 1992; Sasaki et al. 2001). Cold acclimation is also associated with dynamic changes in the rates of net photosynthetic CO2 assimilation (A) and respiratory CO2 release (Rdark) and O2 uptake in darkness, with both increasing in response to sustained cold treatment (Billings et al. 1971; Badger, Bjorkman & Armond 1982; Stitt & Hurry 2002; Atkin & Tjoelker 2003; Armstrong et al. 2008).

Subsequent formation of cold-developed (CD) leaves can lead to further changes in metabolic rates, and to additional changes in leaf morphology, anatomy and chemistry (Huner et al. 1981; Boese & Huner 1990; Strand et al. 1997; Hurry et al. 2000; Campbell et al. 2007; Gorsuch et al. 2010). For example, CD leaves are typically thicker than their WD counterparts because of a proliferation of cell layers (Boese & Huner 1990; Equiza & Tognetti 2002; Atkin et al. 2006; Koeda et al. 2009; Gorsuch et al. 2010). Additionally, inhibition of laminal growth, in combination with limited changes in cell initiation, results in high densities of stomata and epidermal cells of CD leaves relative to WD counterparts (Gorsuch et al. 2010). An important question, however, is how long plants have to experience sustained cold in order for newly developing leaves to develop such CD traits. On one hand, responding to a short period of cold could promote the formation of leaf anatomical, physiological and/or biochemical characteristics (i.e. phenotypes) that are optimal for the conditions present at the time. On the other hand, a rapid switch to the formation of CD-type leaves could result in a maladaptive phenotype, should this cold period be followed by a return to sustained warm conditions, as can happen in early summer (Loveys et al. 2006; Gu et al. 2009).

To what extent is the cold acclimation phenotype reversible when acclimated plants are shifted back to higher growth temperature? Do all traits exhibit de-acclimation (i.e. the loss of cold hardiness upon exposure to warm temperature; Levitt 1972; Chen & Li 1980)? Although the majority of research into the effects of cold on plant tissues has focused on the consequences of a continuous cold treatment, several studies have quantified the changes associated with subsequent de-acclimation (Kalberer, Wisniewski & Arora 2006). For example, sugar concentrations of cold-acclimated PE leaves (that had initiated under warm conditions before cold exposure) have been found to return to warm levels within 5 d of de-acclimation (Sasaki, Ichimura & Oda 1996; Wanner & Junttila 1999; Sasaki et al. 2001), partly because of decreases in synthetic enzyme activity (Sasaki et al. 2001). Starch returns to concentrations found in warm-grown plants even more quickly (Savitch, Gray & Huner 1997), although the metabolite profile in general does not indicate a simple reversal of a cold acclimation phenotype, with a few metabolites specifically regulated by de-acclimation (Kaplan et al. 2004). Freezing tolerance, which is correlated with sugar concentration (Hannah et al. 2006), appears to return to warm levels within 1 d in some species (Chen & Li 1980; Sasaki et al. 1996), but much more slowly and only partially in others (Gay & Eagles 1991). Leaf expansion at warm temperatures in previously CD leaves has been suggested to be associated with a loss of freezing tolerance, leading Rapacz (2002a,b) to conclude that de-acclimation and developmental progression are interconnected. However, the association between de-acclimation and most other phenotypes that are affected by cold, such as anatomy, morphology and rates of A and Rdark, has not been explored. Discrimination between different carbon isotopes can be used to infer the workings of these processes, and past studies report that discrimination against 13C is lower (i.e. the δ13C is higher) in CD tissues than in their warm-grown counterparts (Smith, Herath & Chase 1973; Smith, Oliver & McMillan 1976; Körner, Farquhar & Wong 1991; Yamori et al. 2006; Gorsuch et al. 2010). However, it is not known whether the de-acclimation process affects leaf δ13C. Understanding the extent to which the carbon balance and leaf structure change during de-acclimation may help in making predictions of plant growth and performance in periods of warming that follow sustained periods of cold.

The extent to which individual leaf characteristics respond dynamically to growth conditions appears highly variable, depending on the leaf trait in question and the stage of development during which a leaf experiences a change in environment. For example, as noted above, temperature-related carbohydrate profiles appear to be ‘phenotypically flexible’ (Piersma & Drent 2003), responding dynamically to growth conditions. By contrast, other characteristics of mature leaves (e.g. leaf anatomy) exhibit near permanence, borne through development [‘developmental plasticity’ (Rapacz 2002a,b; Piersma & Drent 2003; Atkin et al. 2006)], with the mature leaf phenotype being largely dependent on environmental conditions experienced during the early stages of development when many leaf cells are still dividing and/or expanding. Examples of this include systemic leaf responses to atmospheric CO2 concentration and growth irradiance (Lake et al. 2001; Yano & Terashima 2004). Although observations have been made of the phenotypic determination of monocotyledon leaf structure by cold exposure during development (Peacock 1975; Woodward 1979; Harrison, Nicot & Ougham 1998), equivalent data for dicotyledon leaf structure are lacking.

In this study, we quantified a broad range of Arabidopsis thaliana (Col-0) leaf characteristics for initially warm-grown plants that experienced cold for periods of several hours to weeks, before being transferred back to higher growth temperatures, where leaves were allowed to mature. Using this approach, we aimed to investigate the extent to which cold acclimation characteristics endure following transfer back to warm conditions. Specifically, we addressed the following questions: (1) For how long do warm-grown Arabidopsis thaliana plants have to experience sustained cold before developing leaves become irreversibly cold-acclimated? (2) To what extent is the de-acclimation process associated with changes in leaf anatomy, chemical composition and carbon economy? The experiments for our current study were performed alongside those of Gorsuch et al. (2010); in that study, we reported on the effects of continuous cold on a comprehensive range of physiological, structural and chemical composition characteristics exhibited by Arabidopsis leaves following shifting of warm-grown plants to sustained cold for up to 2 months. Where appropriate, comparisons are made with data reported in Gorsuch et al. (2010).


Treatment methodology and ‘reference leaf’ system

As described in Gorsuch et al. (2010), wild-type Arabidopsis thaliana (Col-0) were sown in peat-based compost (Levingtons, F2) in a growth chamber (Snijders Microclima 1750; Snijders Scientific BV, the Netherlands). An 8 h day/16 h night temperature regime of 25/20 °C was imposed, with 60–70% relative humidity (RH) throughout and 150 µmol m−2 s−1 photosynthetic photon flux density (PPFD) provided by fluorescent tubes. When approximately 20 leaf insertions had emerged from the apical bud (visible at close inspection), the plants were transferred to 17 L hydroponic tanks containing fully aerated modified Hoagland's nutrient solution at pH 5.8 (Poorter & Remkes 1990). The plants were held on foam discs in pairs from this point.

When 10 further insertions had emerged, experiments were initiated (‘day 0’ of the temperature treatments); on this day, the next insertion to visibly emerge from the central meristematic region of each plant was tracked throughout using small acrylic paint spots on this ‘reference leaf’ (hereon called ‘Leafref’) and other insertions. Thereafter, a subset of plants were shifted to a growth chamber with identical conditions, but with 5 °C maintained throughout, for a predetermined treatment period (with a group of control plants kept in the warm cabinet). After completion of the cold treatment, the plants were transferred back to the initial warm cabinet; here, the Leafref was allowed to fully mature in the warm (defined as a slowing in the rate of growth – see below), at which point the plants were harvested. The Leafref typically took 1½–3 weeks of growth in the warm to mature (with no significant relationship between cold exposure duration and warm maturation time; data not shown). Warm control plants (Supporting Information Fig. S1a) were treated similarly, but with no cold treatment. CD plants were kept in the cold throughout, and harvested when the Leafref had reached maturity (Supporting Information Fig. S1b). The plants were assigned randomly in their foam disc pairs to each treatment, with 4–6 plants per treatment. The discs were kept adjacent to each other, but care was taken to avoid overcrowding.

After the Leafref had fully matured (for WD, CD and cold-treated plants that were returned to the warm where developing leaves matured), we then quantified either rates of CO2 exchange, or a range of structural and chemical composition characteristics. Separate plants were used for the CO2 exchange and structural/chemical composition measurements. To ensure that leaves were harvested at a comparable developmental stage, we monitored leaf expansion and sampled leaves when the growth of the Leafref had started to slow as it approached its maximum area, as described in Gorsuch et al. (2010). This leaf was then used for analysis wherever possible (gas exchange, area/mass relationships, carbohydrates), although some traits were quantified for the leaf insertions immediately adjoining the Leafref.

Net CO2 exchange measurements and subsequent analyses

CO2 exchange measurements were conducted on WD, CD and cold-treated plants that had been returned to the warm to allow initially CD leaves to mature. We used a Li-6400 Portable Photosynthesis System (Li-Cor Inc., Lincoln, NE, USA) in situ to quantify rates of photosynthetic CO2 assimilation (A), respiratory CO2 release (Rdark) and the balance between A and Rdark at two temperatures (25 and 5 °C), as described in Gorsuch et al. (2010). Initial measurements were made at an irradiance of 150 µmol photons m−2 s−1, and a CO2 concentration of 400 µmol mol−1, to give the assimilation rate at ambient CO2 conditions (A400). Intercellular CO2 concentration (Ci), stomatal conductance (gs), rate of transpiration and leaf vapour pressure deficit (VPDL) were also measured at these conditions, following which the irradiance was increased to a level predetermined as saturating (500 µmol photons m−2 s−1) to enable measurements of the light-saturated rate (Asat). Thereafter, the CO2 level was increased to 1000 µmol mol−1 (also predetermined as saturating) to give a CO2-saturated/light-saturated rate (Amax), and finally the plant was darkened for 30 min under a CO2 concentration of 400 µmol mol−1 for dark respiration (Rdark) readings. Data were corrected for CO2 diffusion across the gasket (Bruhn et al. 2002). At the completion of the gas exchange measurements, the leaves were harvested, and the fresh mass and area of leaf parts contained in the photosynthesis system quantified with a LI-3000A leaf area meter (Li-Cor Inc.). The dry mass was recorded following freeze-drying under vacuum (Edwards Modulyo Freeze Drier, York, UK).

Anatomical, epidermal and chemical composition characteristics

The parameters described below were measured on separate plants to those used for gas exchange analysis. Details on the protocols used to quantify leaf anatomy, and epidermal and chemical composition characteristics are provided in Gorsuch et al. (2010). Briefly, leaf dry mass per unit area (LDMA), fresh mass per unit leaf area (LFMA) and dry matter content (DMC, the ratio of dry mass to fresh mass) were calculated. Sugars and starch were extracted (using hot ethanol) from the freeze-dried, ground leaves, and the concentrations of sucrose, fructose, glucose and starch were analysed according to Loveys et al. (2003) and Campbell et al. (2007). To determine carbon concentration and carbon isotope ratio (δ13C), one freeze-dried leaf per plant was ground, with 2.5 mg subjected to continuous flow gas chromatograph isotope ratio mass spectrometry (GC-IRMS, Provac Services Ltd, Crewe, UK).

To investigate internal leaf anatomy, proximal halves of individual mature leaves were fixed in paraformaldehyde/ethanoic acid/ethanol, dehydrated in ethanol and infiltrated and embedded in L.R. White resin. Microtome sections (Ultracut UCT; Leica Microsystems, Wetzlar, Germany) were cut approximately midway between the midrib and margin, stained with toluidine blue and imaged (Optiphot 2 with DS-L1; Nikon, Tokyo, Japan). ImageJ (v. 1.38, NIH, Bethesda, MD, USA) was used to analyse enhanced images. Using the distal halves of the same leaves, samples of epidermal peels were made (Kagan, Novoplansky & Sachs 1992), after which epidermal cells and stomata were counted under 25× magnification. Stomatal index (SI) and ratio (SR) were calculated according to:


where SD is the stomatal density, ECD is the epidermal cell density, Ab is abaxial (lower) surface and Ad is adaxial surface.

Statistical analysis

Where necessary, analysis of variance (anova), analysis of covariance (ancova) and linear regressions were used. LSD post hoc tests were employed where one-way anovas indicated significant differences among the factors. Further tests are described in the text. Proportions were arcsine transformed before analysis. Depending on the characteristics being quantified, means were based on three to six independent replicates, with n = 4 being the most common.


In this study, we describe the impact of increasing duration of cold treatment (5 °C) on the characteristics of leaves that subsequently were allowed to mature under warm (25/20 °C) conditions. Where appropriate in graphs, we express values relative to those of WD controls. For comparison, absolute values of the WD and CD leaves are shown in Supporting Information Table S1, using data reported in Gorsuch et al. (2010).

Leaf expansion

When developing leaves were exposed to 5 °C for increasing durations, leaf growth was compromised during subsequent maturation in the warm (Fig. 1a). The final area of the first leaf to visibly emerge from the centre of the rosette following the commencement of the cold treatment (i.e. Leafref) exhibited a negative relationship with the duration for which plants had been exposed to cold (data not shown). The maximal leaf expansion rate in the warm became more compromised as the duration of the preceding cold treatment (Fig. 1b). Comparison with the WD control plants revealed that this effect was only significant in leaves that have been exposed to 5 °C for 30 or 45 d (as indicated from a one-way anova and LSD post hoc test for each cold duration treatment; Table 1). When shifted back to the warm, the duration of leaf expansion was similar for plants that had been cold treated for short (e.g. 5–10 d) and long periods (e.g. 30–45 d); consequently, the final leaf area attained was lower for leaves that had been cold treated for long periods (data not shown but see Supporting Information Fig. S1).

Figure 1.

Expansion of specific leaves, estimated from length and width measurements as described in Gorsuch et al. (2010). (a) Leaf area over time in leaves at constant 5 °C (filled triangles), and in leaves transferred from 5 to 25 °C at 22 d (arrow, open squares) and 30 d (arrow, open circles). (b) Maximum warm expansion rate of leaves as related to duration of cold experienced from the start of their development. See Table 1 for results of the statistical analyses.

Table 1.  Statistical tests of cold treatment effects in structural and metabolic analyses of leaves exposed to a transient 5 °C stimulus early in development but subsequently allowed to mature in the warm
VariableCold time effectData shown
  1. Area, fresh mass (FM) and dry mass (DM) statistics were cold exposure covariate effects from analyses of covariance (ancovas) (with experimental run as a factor). All other tests shown are regressions within one experimental run. Further tests are described in the text. Arrows denote the direction of observed changes as transient cold treatment time increased.

  2. Asterisks denote significance levels: *P < 0.05, **P < 0.01, ***P < 0.001.

  3. NS, not significant; LDMA, leaf dry mass per unit area; LFMA, leaf fresh mass per unit area; Ad, adaxial (upper) surface; Ab, abaxial surface; ECD, epidermal cell density; SD, stomatal density; SI, stomatal index; ECs, epidermal cells; SR, stomatal ratio.

Leaf expansionFinal leaf area1, 795.5*Fig. 1a
Maximum rate1, 8525.7***Fig. 1b
Area, FM and DM relationshipsLDMA1, 7555***Fig. 2
Absolute thicknessWhole leaf1, 1316**Fig. 3a
Ad epidermisNS
Palisade mesophyll25***
Vascular tissueNS
Spongy mesophyll6.2*
Ab epidermisNS
Proportion of total thicknessMesophyll cell1, 13NSSupporting Information Fig. S2a
Vascular tissueNS
Air spaceNS
Epidermal cell52***
Cell layers1, 1310.3**Supporting Information Fig. S2b
Epidermal dataAbECD1, 159.1**Fig. 4
Ecs leaf−1NS
Stomata leaf−14.7*
AdECD1, 1517.8***
ECs leaf−1NSTable 2
Stomata leaf−1NS
 SR (Ab/Ad)1, 1513**
Chemical compositionSugars 7.0*

Leaf biomass allocation and anatomy

In leaves that matured under warm conditions, leaf structural characteristics varied depending on the duration of cold experienced at the early stages of their expansion. For example, LDMA (Fig. 2a) and LFMA (Fig. 2b) both increased in response to increasing duration of cold treatment during early development. Relative to their respective WD controls, mature leaves exhibited significantly higher LDMA and LFMA values after 10 and 5 d of earlier 5 °C treatment, respectively (Fig. 2a,b; Table 1). DMC also significantly increased with the duration of cold treatment (Fig. 2c; Table 1), but the magnitude of this change was negligible compared with that between WD and CD controls (Supporting Information Table S1).

Figure 2.

Area and mass relationships in leaves subjected to a cold treatment applied for increasing durations early in development, after which leaves were transferred to the warm to mature. Cold treatment duration is shown for each parameter. (a) Leaf dry mass per unit area (LDMA), (b) leaf fresh mass per unit area (LFMA), (c) dry matter content (DMC). For each time-point, data are shown relative to their respective warm-developed (WD) control (i.e. 0 d treatment) mean, with data from six experiments in total. Mean (±SE) absolute values for warm control leaves: LDMA: 15.5(±2) g DM cm−2, LFMA: 194(±0.2) g FM cm−2, DMC: 8.0 (±0.1)%. See Table 1 for results of the statistical analyses.

Leaves that experienced prolonged cold before expanding in the warm were noticeably thicker than WD controls (Fig. 3; Table 1). After 15 d of exposure to 5 °C (and subsequent warm expansion), leaves exhibited a thickness 58% above that of counterparts that did not experience cold, although they were 17% below cold control levels. The increase in total thickness was driven by both mesophyll layers (Fig. 3a) – spongy and palisade depths increased by 64 and 79%, respectively, between the warm control and the 15 d cold treatment. The proportion of leaf section area that comprised of epidermis decreased significantly with exposure time (Supporting Information Fig. S2a), reflecting the increase in mesophyll cell area. The relative contributions of vascular tissue and airspace (as if spread in an even layer across the section) were unaffected by cold exposure (Supporting Information Fig. S2a). The leaves had one more cell layer following early exposure to the cold treatment (Supporting Information Fig. S2b). Importantly, however, they were still more than a layer short of the CD mean (which exhibited an average of near eight cell layers; Supporting Information Table S1).

Figure 3.

Thickness of each cell type layer (a) in leaf sections (b–d) of leaves exposed to a cold treatment for a determined period during early development, and then transferred to the warm to mature. (a) Thickness measurements taken from leaf sections made following regrowth in the warm, after initial cold treatment times of between 0 d [i.e. warm-developed (WD) control] (b) and 30 d (c). (d) Cold-developed (CD) control at the same physiological stage of development. Part (a) shows a representative run, displayed to reflect the anatomy of showing thicknesses of (in order from pale to dark grey, and from top to bottom) adaxial epidermis, palisade mesophyll layer, vascular tissue, spongy mesophyll layer and abaxial epidermis. Mesophyll thickness measurements include intercellular airspaces. In b–d, bars represent 100 µm; note that strength of lines in cross-section images does not accurately indicate cell wall thickness. See Table 1 for results of the statistical analyses. See Supporting Information Figure S2 for graphs of cell type and airspace proportions, and of the number of cell layers.

Regression analysis showed that several epidermal surface parameters were affected by increasing the duration of the cold treatment. ECD and SD increased with cold treatment time (Fig. 4a,b; Table 1), although the SI did not change (Fig. 4c). The total number of stomata decreased with time on the abaxial surface (Fig. 4e). The SR decreased with cold treatment time (Fig. 4f), indicating decreases in both the abaxial stomatal count relative to that of the adaxial surface, and the calculated ratio of abaxial to adaxial SI values (F1,15 = 7.0; P < 0.05, data not shown).

Figure 4.

Stomatal and epidermal cell characteristics of leaves exposed to cold treatment applied for increasing duration during early development, after which leaves were transferred to the warm to mature. Filled circles: abaxial (lower) leaf surface, open circles: adaxial. (a) epidermal cell density (ECD), warm-developed (WD) control (0 d) abaxial: 363 (±SE = 23) mm−2, adaxial: 259 (±18) mm−2; (b) stomatal density (SD), 0 d abaxial: 132 (±8) mm−2, adaxial: 108 (±7) mm−2; (c) stomatal index (SI), 0 d abaxial: 26.8 (±0.5)%, adaxial: 29.5 (±0.3)%; (d) estimated total epidermal cell number, 0 d abaxial: 218 000 (±19 000) mm−2, adaxial: 156 000 (±14 000); (e) estimated total stomata, 0 d abaxial: 80 000 (±7000) mm−2, adaxial: 65 000 (±6000; (f) abaxial : adaxial stomatal ratio (SR), 0 d: 1.27 (±0.06)%. Bars represent standard errors. See Table 1 for results of the statistical analyses.

Carbohydrates and δ13C

The duration of cold treatment, prior to leaf maturation in the warm, had little effect on the total concentration of soluble sugars (Tables 1 & 2). By contrast, starch levels showed a clear linear decrease with cold treatment duration, driving a decrease in total non-structural carbohydrate levels (Tables 1 & 2). δ13C showed a negative linear relationship with the duration of cold treatment (Tables 1 & 2). This contrasted with a higher δ13C in CD controls compared to the WD controls (Supporting Information Table S1). The proportion of carbon in leaf dry matter was unaffected by the preceding cold duration (data not shown).

Table 2.  Carbohydrate and carbon isotope composition of leaves subjected to a cold treatment applied early in development, prior to maturation of leaves in the warm
Cold treatment duration (days)Soluble sugars (mg g1)Starch (mg g1)TNC (mg g1)δ13C (‰)
  1. Cold treatment duration is shown for each parameter.

  2. Average (±SE) concentrations (expressed per unit dry mass) of total soluble sugars (i.e. glucose + fructose + sucrose), starch and total non-structual carbohydrates (TNC, starch + total soluble sugars) are shown, as are results of the mass spectrometry analysis (δ13C). See Table 1 for results of the statistical analyses.

 016.5 ± 0.536.6 ± 7.353.0 ± 7.3−29.7 ± 0.2
1519.1 ± 1.317.8 ± 7.036.9 ± 6.9−30.2 ± 0.2
2222.2 ± 2.311.3 ± 3.733.5 ± 3.4−30.3 ± 0.1
3022.6 ± 1.72.6 ± 2.125.2 ± 2.1−30.4 ± 0.2

Leaf gas exchange

On a leaf area basis and when measured at warm regrowth temperature (25 °C), photosynthesis did not show a significant relationship with cold treatment duration under ambient conditions (A400), saturating irradiance (Asat) or saturating irradiance and CO2 (Amax), as determined using linear regression (Fig. 5a–c). When expressed per unit dry mass, all measurement conditions yielded a marked decrease in A with exposure time (Fig. 5d–f; Table 3). The response of Amax was particularly strong. The photosynthetic response to a decrease in measurement temperature to 5 °C depended on its thermal history, with a significant reduction in temperature sensitivity as preceding cold treatment duration increased (the ancova measurement temperature by cold duration effect was significant; Table 3), eventually resulting in a lack of temperature response. When measured at 5 °C and expressed on a leaf area basis, photosynthesis exhibited a significant positive relationship with the duration of cold treatment, at all three irradiances and CO2 conditions tested (Fig. 5a–c; Table 3). No relationship was found when A was expressed per unit DM at 5 °C (Fig. 5d–f).

Figure 5.

Photosynthesis (±SE) at growth levels of CO2 and irradiance (A400, top row), saturating irradiance (Asat, middle row) and saturating irradiance and CO2 (Amax, bottom row) of leaves exposed to cold treatment applied for increasing duration during early development, after which leaves were transferred to the warm to mature. Data are expressed per unit area (a–c) and dry mass (DM, d–f). Readings were carried out at 25 °C (open symbols) and 5 °C (filled symbols). Cold exposure durations refer to the period of cold during early development, before subsequent maturation in the warm. See Table 3 for results of the statistical analyses.

Table 3.  Results of statistical tests on gas exchange characteristics of leaves exposed to a transient 5 °C stimulus early in development but subsequently allowed to mature in the warm
VariableMeas. T (°C)Regressionancovas: Meas. T effectInteractionData shown
Cold duration effect
  1. Cold treatment time effect refers to linear regression analysis (respiration d.f. = 1, 24, other d.f. = 1, 26). Analyses of covariance (ancovas) tested the effect of an increase in measurement temperature from 5 to 25 °C on each parameter, with measurement temperature by cold exposure duration interaction terms also shown (d.f. = 1, 50 to 1, 52). Arrows denote the direction of observed changes as transient cold exposure time or measurement temperature increased.

  2. Asterisks denote significance levels: *P < 0.05, **P < 0.01, ***P < 0.001.

  3. See text for explanation of photosynthesis abbreviations. NS, not significant; meas. T, measurement temperature; DM, dry mass; R : A400, ratio of respiration to photosynthesis; Ci, internal CO2 concentration; VPDL, vapour pressure deficit; IWUE, instantaneous water-use efficiency.

Photosynthesis (µmol CO2 m−2 s−1)A40054.8*21.7***5.9*Fig. 5
Photosynthesis (nmol CO2 g DM−1 s−1)A4005NS16***4.5*
Respiration (µmol CO2 m−2 s−1)520***137***5.0*Fig. 6
Respiration (nmol CO2 g DM−1 s−1)58.0**139***8.3**
R : A400512**63***NS
Conductance (mol H2O m−2 s−1)5NS12**NSFig. S3
Ci (mmol mol−1)5NS11**NS
Transpiration (mmol H2O m−2 s−1)5NS52***NS
VPDL (kPa)518***152***6.0*
IWUE (mmol CO2 mmol H2O−1)5NS13***NS

Ci decreased with previous cold exposure time when measured at 25 °C, but not 5 °C (Supporting Information Fig. S3a; Table 3). Conductance to water vapour also decreased at the lower temperature (Supporting Information Fig. S3b), but transpiration did not show a consistent linear relationship with cold exposure duration at either temperature (Supporting Information Fig. S3c). Leaf vapour pressure deficit (VPDL) increased significantly with cold treatment duration at both measurement temperatures because of high readings in the 45 d cold plants (Supporting Information Fig. S3d; Table 3). No trend was observed in instantaneous water-use efficiency (IWUE; Supporting Information Fig. S3e). A decrease in measurement temperature resulted in significant increases in conductance, Ci and IWUE, and decreases in transpiration and leaf VPDL.

When measured at the regrowth temperature (25 °C), respiratory rates did not exhibit a relationship with the duration of cold experienced during leaf development; however, a positive relationship was observed when measurement temperature was reduced to 5 °C (Fig. 6a,b; Table 3). The response of the ratio of respiration to photosynthesis (R : A400) was similar, with an apparent increase with cold treatment duration at 25 °C, but not 5 °C (Fig. 6c).

Figure 6.

Dark leaf respiration (Rdark) (a,b) and the ratio of Rdark to photosynthesis (R : A400) (c) at ambient CO2, per unit (a) area and (b) dry mass (DM) of leaves exposed to cold treatment applied for increasing duration during early development, after which leaves were transferred to the warm to mature (±SE). Readings were taken following subsequent maturation in the warm, at 25 °C (open symbols) and 5 °C (filled symbols).


We sought to establish the impacts of de-acclimation on a broad range of phenotypes for leaves that had experienced different durations of transient cold during early development, before being transferred back to the warm to mature. Previous reports have demonstrated that the phenotypic responses of leaf gene expression, freezing tolerance and carbohydrate content and metabolome to several days' exposure to low temperature are largely reversible following transfer back to warm temperatures (Sasaki et al. 1996, 2001; Savitch et al. 1997; Wanner & Junttila 1999; Kaplan et al. 2004). By contrast, we have shown that Arabidopsis leaves that experience cold for extended periods during early development exhibit little plasticity in leaf structure (Figs 2–4) and gas exchange (Figs 5 & 6), do not revert to the carbohydrate profile of WD controls (Table 2) and display a permanently compromised growth rate when transferred back to a warm growth environment (Fig. 1). Sustained cold treatment of developing leaves for periods as little as 5 d resulted in irreversible changes in morphology of leaves that subsequently mature in the warm (Fig. 2), with 15 d of cold being sufficient to result in permanent changes in leaf anatomy (Fig. 3). Collectively, these results: (1) highlight the importance of the duration of cold treatment during early development for functional traits of leaves that subsequently mature under warm conditions; and (2) support suggestions of a close association between leaf expansion and phenotypic determination (Sims & Pearcy 1992; Rapacz 2002a,b).

Prolonged cold during development affects leaf morphology and anatomy

Leaf dimension measurements revealed that maximal growth rate at 25 °C was compromised in leaves that had been cold treated for extended periods earlier in their development (Fig. 1). Cross-sections taken from 15, 22 and 30 d cold-treated leaves suggest that growth rate may have been constricted by structural factors, as these leaves were much thicker than WD controls because of an expansion of the mesophyll (Fig. 3), potentiated by a higher number of cell layers (Supporting Information Fig. S2b). This suggests that the thickness of cold-treated leaves is determined within the first 15 d of development, partly by the number of cell layers. This is consistent with previous affirmations that most cell division in Arabidopsis leaves occurs early in development (Dale 1992; Donnelly et al. 1999; Cookson, Van Lijsebettens & Granier 2005). Compared to the transiently cold-treated plants, fully CD control leaves were around a further 25% thicker, and had one further cell layer. Given that a leaf is unlikely to consume a cell layer during warm development, these data suggest that either the anatomy of CD leaves continues to change after more than a month of cold development, or the spongy mesophyll ‘unfolded’ during warm maturation, splaying the cells out to help the leaf to expand. More research is needed to determine which of these applies.

Leaf fresh and dry matter accumulation per unit area were permanently affected in leaves exposed to increasing periods of cold treatment during development (Fig. 2), presumably at least in part due to changes in internal cell architecture (Fig. 3). An LSD post hoc test suggested that leaves affected very early in development by treatments of 5 d or more at 5 °C – during which minimal expansion would be expected to have occurred (Gorsuch et al. 2010) – displayed a higher ratio of LFMA (Fig. 2b) than warm controls. As LFMA is correlated with leaf thickness (Dijkstra et al. 1989; Atkin, Botman & Lambers 1996; Vile et al. 2005), this may suggest that the measured changes in leaf thickness (Fig. 3a) may also have been present following relatively brief cold exposure. Differences between treatment and warm control in dry mass per unit area (LDMA; Fig. 2a) were significant from 10 d. DMC also showed a significant, if slight, relationship with cold exposure duration (Fig. 2c). The close association between dry matter and leaf structure (Kubacka-Zebalska & Kacperska 1999) further supports evidence of considerable permanent effects on leaf structure.

Increases in the densities of stomata and epidermal cells in leaves exposed to long term, relative to short term, cold early in development (Fig. 4a,b) were attributable to the limited final area of these leaves, as there was no concomitant increase in cell count (Fig. 4d,e). The initiation of stomata relative to epidermal cells also showed no relationship with cold exposure duration (Fig. 4c), so it appears that what little area-independent effect cold development has on stomatal phenotype (Gorsuch et al. 2010) is no longer evident following warm maturation.

Carbohydrates did not return to warm control levels during de-acclimation

Following the transfer of cold-treated plants to warm temperatures, previous reports have demonstrated a drop in leaf sugar levels to WD control concentrations in 1–5 d (Sasaki et al. 1996, 2001; Wanner & Junttila 1999). Consistently with these studies, concentrations of total soluble sugars did not differ significantly between a cold treatment of 15 d (followed by 13 d maturation at 25 °C) and WD controls (Table 2). However, leaves subjected to 22 or 30 d cold treatment during early development differed markedly from WD controls in levels of at least one sugar (Table 2), following warm regrowth periods of 12 and 18 d, respectively. The relative maintenance of sugar concentrations may be caused by the breakdown of starch (Table 2), which showed a striking, linear decrease with preceding cold treatment time. Starch was apparently exhausted in leaves that developed during 30 d cold treatment before warm maturation (Table 2). Starch levels increase during the photoperiod (Strand et al. 1997), but an artefact of harvest timing is unlikely to be responsible, as the WD control was analysed earliest and the 22 d cold treatment latest, with only 2 h between these extremes. It seems more likely that the balance between respiration and photosynthesis was responsible for determining the leaf carbohydrate profile.

CO2 exchange is permanently compromised when leaves experience cold during early development

Leaves treated with prolonged cold during their development exhibited little sign of photosynthetic de-acclimation to subsequent warming during maturation. Moreover, rates per unit dry mass were negatively correlated with preceding cold treatment duration when measured at 25 °C (Fig. 5d–f), despite only slight changes in DMC over the cold treatment duration range (Fig. 2c). The vestiges of cold acclimation were also apparent when photosynthesis was measured at 5 °C and expressed per unit area, with a positive relationship apparent between assimilation and cold treatment duration (Fig. 5a–c). The 45 d cold treatment showed the least evidence of de-acclimation despite 18 d regrowth at 25 °C – although its phenotype was not as extreme as that of CD control leaves, suggesting that some de-acclimation had most likely taken place. The net result of these contrasting responses to increasing cold treatment duration at each measurement temperature was a loss of photosynthetic temperature sensitivity, especially in 45 d cold-exposed plants, which showed no measurement temperature response of either A400 or Asat (Fig. 5) – indeed, these leaves also lacked sensitivity to irradiance, suggesting a fundamental shift in the balance of limitation of photosynthesis following cold exposure of this length. The low temperature sensitivity of these leaves also suggests that chloroplastic inorganic phosphate (Pi) supply does not limit carbon assimilation, as in cold-exposed tissues (Sage & Kubien 2007) – instead, these leaves appeared to be limited by CO2 (Fig. 5c). Taken together, the photosynthesis data presented here strongly suggest a fundamental and possibly permanent effect of long-term cold treatment on leaf physiology.

In general, respiration (Fig. 6) exhibited a similar response to photosynthesis, with an increase cold exposure duration eliciting a rise in rates measured at 5 °C (but not 25 °C), and resulting in a decrease in temperature sensitivity (i.e. Q10 of Rdark declined). What underlying factors might have been responsible for the asynchronous changes in Rdark measured at 5 °C versus those measured at 25 °C? Previous studies have reported that at low measurement temperatures, respiratory flux is typically limited by the enzymatic capacity of the respiratory apparatus (i.e. glycolysis, the TCA cycle and mitochondrial electron transport), either because of the inhibitor effect of cold on potential enzyme activity per se (both in soluble and membrane-bound compartments) and/or limitations on the function of enzymes embedded in membranes at temperatures (Covey-Crump, Attwood & Atkin 2002; Atkin & Tjoelker 2003). By contrast, respiratory flux is less limited by enzymatic capacity at moderate temperatures (Wiskich & Dry 1985). It has also been shown previously that associated with the increase in rates of O2 uptake per leaf mass in CD leaves are increases mitochondrial abundance, changes in mitochondrial ultrastructure and an increase in the capacity for O2 uptake in individual mitochondria (Armstrong et al. 2006, 2008). Such changes could explain the higher rates of Rdark measured at 5 °C exhibited by leaves that experienced sustained cold treatment during early leaf development, prior to subsequent leaf maturation in the warm. The question then arises as to why there was no concomitant increase in Rdark measured at 25 °C. Increased substrate limitation seems unlikely, given the homeostasis of total sugars (Table 2). Instead, it is likely that either adenylates were limiting, or that the temperature optimum of respiration shifted permanently, perhaps via irreversible changes to membrane or mitochondrial structure, such as via alteration in the crista to matrix ratio (Armstrong et al. 2006).

13C discrimination

Previously, we have reported that cold treatment progressively increases δ13C (Gorsuch et al. 2010). If the same is true for developing leaves that were cold treated for extended periods before being returned to the warm to mature, then it seems likely that initially δ13C values of cold-treated leaves would have been high, but then decreased following transfer to the warm. The potential reasons for such an increase in discrimination against 13C are unclear, especially given that most other parameters in the current investigation exhibited little de-acclimation to warm maturation temperature. The replacement of 13C-rich, cold-derived compounds is possible given that the half-life of total leaf protein is approximately 1 week (Simpson, Cooke & Davies 1981; Scheurwater et al. 2000), yet physiological traits showed little sign of concomitant optimization to the maturation temperature. The loss of starch in CD, warm-matured leaves (Table 2) may have been responsible for the decrease in δ13C below warm control levels, as starch is around 4‰ more enriched in 13C than bulk matter (Brugnoli et al. 1988; Gleixner et al. 1993). As fractionation reflects the integration of many processes, more investigation is required. Palaeoecological studies aiming to determine long-term climatic conditions often employ δ13C measurements (Gröcke 2002; Nguyen Tu et al. 2002). The results presented here provide a reminder that δ13C is not only associated with long-term climate, and that relevant studies should be robust enough that their validity is not compromised by the effect of transient changes in temperature on carbon isotope discrimination.

Concluding statements

Our results highlight the influence of transient cold exposure during leaf development on a wide range of functional traits when the leaves subsequently mature in the warm. Crucially, fundamental aspects of leaf anatomy (e.g. number of cell layers) are altered after plants experience transient cold for periods as short as 2 weeks. Moreover, the area and mass relationships of warm-matured leaves are irreversibly altered in plants that experience transient cold for 5–10 d. Finally, many of the responses to de-acclimation – particularly those of carbohydrate profile, gas exchange and stable carbon isotope fractionation – showed unexpected responses which are likely to reflect alterations in underlying processes not specifically studied here. The carbohydrate pool sizes observed may indicate long-term effects of cold on enzyme activity and abundance. Some aspects of the phenotype described may even act as a cold ‘memory’, to protect an individual plant from the harmful effects of later low-temperature events. Moreover, given that environmentally induced epigenetic changes might be inherited by future generations (Bossdorf, Richards & Pigliucci 2008), an intriguing possibility is that the cold tolerance of a plant's offspring might in part be dependent on the thermal history of the parent plant. Analysis of freezing tolerance is a priority in addressing such questions. Finally, the response of plant carbon flux to environmental conditions is central to predicting climate change, which will most likely alter current patterns of temperature fluctuation. It is therefore essential that we understand the interaction between temperature and plant development, and subsequent CO2 exchange, so that we can gain the clearest possible estimation of our future climate.


P.A.G. was the recipient of a UK Biotechnology and Biological Sciences Research Council (BBSRC) PhD studentship. The study was also supported by a Natural Environment Research Council (NERC) research grant to O.K.A. (NER/A/S/2001/01186). We would like to thank David Sherlock for expert technical assistance. Joana Zaragoza-Castells, Anna Armstrong, Jens Subke, Alex Dumbrell and Phil Ineson are thanked for their valuable technical and academic guidance.