A size-mediated effect can compensate for transient chilling stress affecting maize (Zea mays) leaf extension


  • Gaëtan Louarn,

    1. INRA, UMR1281 Stress Abiotiques et Différenciation des Végétaux Cultivés, BP 136, F80203 Péronne, France
    2. INRA, UR4 Pluridisciplinaire Prairies et Plantes Fourragères, BP6, F86600 Lusignan, France
    Search for more papers by this author
  • Bruno Andrieu,

    1. INRA, UMR1091 Environnement et Grandes Cultures, F78850 Thiverval – Grignon, France
    Search for more papers by this author
  • Catherine Giauffret

    1. INRA, UMR1281 Stress Abiotiques et Différenciation des Végétaux Cultivés, BP 136, F80203 Péronne, France
    Search for more papers by this author

Author for correspondence:
Gaëtan Louarn
Tel: +33 (0)5 49 55 60 63
Email: gaetan.louarn@lusignan.inra.fr


  • In this study, we examined the impact of transient chilling in maize (Zea mays). We investigated the respective roles of the direct effects of stressing temperatures and indirect whorl size-mediated effects on the growth of leaves chilled at various stages of development.
  • Cell production, individual leaf extension and final leaf size of plants grown in a glasshouse under three temperature regimes (a control and two short chilling transfers) were studied using two genotypes contrasting in terms of their architecture.
  • The kinetics of all the leaves emerging after the stress were affected, but not all final leaf lengths were affected. No size-mediated propagation of an initial growth reduction was observed, but a size-mediated effect was associated with a longer duration of leaf elongation which compensated for reduced leaf elongation rates when leaves were stressed during their early growth. Both cell division and cell expansion contributed to explaining cold-induced responses at the leaf level.
  • These results demonstrate that leaf elongation kinetics and final leaf length are under the control of processes at the n − 1 (cell proliferation and expansion) and n + 1 (whorl size signal) scales. Both levels may respond to chilling stress with different time lags, making it possible to buffer short-term responses.


Chilling temperatures affect the growth and functioning of chill-susceptible species such as maize (Zea mays) (Blondon et al., 1980; Miedema, 1982; Long, 1983; Pollock & Eagles, 1988; Dolstra et al., 1994; Verheul et al., 1995, 1996; Sowinski et al., 2003). In terms of crop yield, it has been shown that traits related to leaf size, rather than the transient impairment of leaf functioning, can explain the reductions in crop yield associated with early chilling exposure in the northern fringes of the maize cultivation area (Louarn et al., 2008). However, little is known about the mechanisms that affect single-leaf and whole-plant leaf area expansion in response to cold (Greaves, 1996).

In grass species, the expanding region of leaves is confined to the base of the organ, near the leaf insertion point (Volenec & Nelson, 1981; Schnyder et al., 1987; Skinner & Nelson, 1994; Muller et al., 2001). Because growth is mainly unidirectional, the process of leaf elongation is often regarded as the dominant component of leaf area expansion. In the first stage after initiation, the entire expanding primordium is a homogeneous division zone (DZ). Cell division and elongation are coordinated so that mean cell length remains constant. In the second stage, the more distal cells stop dividing and start elongating, giving rise to an ‘elongation only’ zone (EOZ) at the top of the division zone. As a result of the continuous increase in the size of DZ and EOZ, the leaf undergoes an extension which is nearly exponential until EOZ reaches its maximum length (Gallagher, 1979; Lafarge & Tardieu, 2002). When the distal cells stop elongating, they define a new compartment, the mature zone (MZ). The differentiation of MZ marks the beginning of a quasi-linear phase of elongation during which a quasi-steady influx of cells into the mature zone is established. Finally, the growth zone regresses quickly and gives rise to the fade-out phase of leaf elongation (Fournier et al., 2005). Because of its particular organization, which spatially but not temporally separates cell division and elongation, this system has become a model used to unravel the cellular processes of leaf growth (Volenec & Nelson, 1981; Fiorani et al., 2000 for genotypic comparisons) and its response to environmental stresses, including cold (Rymen et al., 2007).

This convenient system is also remarkable in that the expanding region of leaves is enclosed within the sheath tube of previous leaves (Kemp, 1980; Klepper et al., 1982). This structure has long been identified as having a critical impact on leaf ontogeny (Borill, 1961; Yu et al., 1975; Davies et al., 1983). Cutting the whorl, even well above the growth zone, typically results in shortened leaves (Begg & Wright, 1962; Volenec & Nelson, 1983; Casey et al., 1999). Similarly, artificial (Wilson & Laidlaw, 1985) or natural (Andrieu et al., 2006) lengthening of the whorl generates leaves of greater ultimate size. Despite no definitive mechanism having been identified to explain it (Skinner & Simmons, 1993; Chaffey, 2000; Fiorani et al., 2002), this whorl size-mediated effect is one of the few organismal traits that clearly participate in regulating cell division and expansion in the growth zone (Beemster et al., 2003; Kavanova et al., 2006). Assuming a positional (Eulerian) control of cell differentiation, it may act by delimiting zone lengths within the growth zone (DZ and EOZ) and by triggering phase changes (Fournier et al., 2005 and the references therein).

Despite this well-defined dual scale regulation of leaf elongation, very few studies have focused on the respective roles of processes at the cell and whole-plant levels in determining the final size of leaves (Beemster et al., 2003; Tsukaya, 2003). However, this distinction might be crucial to understanding the mechanisms that control growth when studying abiotic stresses (e.g. phosphorus deficiency; see Kavanova et al., 2006). Indeed, some part of the plant response is likely to result from the direct effect of an environmental cue on short-term growth, but there may be another, longer term effect that propagates from the first and is independent of any stress prevailing at the time at which it is observed.

To date, studies on the effects of low temperatures on maize growth have focused either on the cellular (Ben Haj Salah & Tardieu, 1995; Rymen et al., 2007, in a single-leaf extension analysis) or on the whole-plant scale (Louarn et al., 2008). At the cellular level, recent results have suggested that cell division, rather than cell elongation, may be responsible for reduced leaf length (Rymen et al., 2007). At the plant level, Louarn et al. (2008) observed that early chilling stress in the field led to a marked reduction in leaf size for all leaves emerging after the first occurrence of cold. Because cold usually occurs as a transient stress, this pattern could result either from a direct negative impact of cold on young expanding leaves at the time of the stress (as reduced cell proliferation might suggest), or simply from the size-mediated propagation of an initial reduction up to the topmost leaves.

To investigate these two alternative views, the aim of the present study was thus to examine the impact of transient chilling stress on cell production, leaf elongation kinetics and final leaf length in the whole plant. To assess the role of whorl length, we used two inbred lines contrasting with respect to their whorl formation and tested different timing of stress. Growth kinetics were expressed vs temperature-compensated time units in order to account for the nonstressing effect of temperature (Parent, 2009). At the leaf level, a five-parameter biphasic leaf extension model was used to compare the exponential and pseudo-linear phases of extension between treatments (Lafarge & Tardieu, 2002; Hillier et al., 2005). Cellular processes were inferred through estimations of overall cell production (final epidermal cell number in a file) and final cell length in mature leaves.

Materials and Methods

Experimental design

Experiments under controlled conditions were carried out at Estrées-Mons (49°N, 3°E, 85 m), France, from January to April 2008 using the cold-tolerant maize (Zea mays L.) inbred lines F286 and F2 (early flint lines of temperate origin differing in terms of their leaf architecture characteristics). The plants were grown in a temperate glasshouse (24°C : 20°C day : night, 16-h photoperiod) and moved into a cold compartment (14°C : 10°C, 16-h photoperiod) for 1 wk, either before (T1: 8 d after plant emergence; 4th leaf tip emerging; whorl length rapidly increasing) or shortly after tassel initiation (T2: 20 d after emergence; 8th leaf tip emerging; whorl length close to its maximum value). Control plants (T0) were kept in the temperate glasshouse. All plants were arranged in a split-plot design with four replication blocks. Randomization in the split-plot was restricted in that temperature treatments were grouped to facilitate transfer operations. Each block contained 16 9-l pots (two plants per pot) filled with standard potting soil spaced at regular intervals. This resulted in a density of c. 11 plants m−2. The plants received supplementary light (from a Philips son-T 400-W sodium lamp, Philips Lighting UK, Surrey, UK) when the solar light intensity at the top of the canopy dropped below 240 μmol m2 s−1 in the photosynthetically active radiation (PAR) domain. Fertirrigation with a nutrient solution (Hydrokani C2, 3.33‰ v : v; Hydro Agri, Neuilly sur-Seine, France) was applied twice daily (2 × 2 min at a flow rate of 2 l h−1). Supplementary water was applied twice weekly to ensure that the fraction of soil transpirable water remained above 0.8 (assessed by pot weighting of a subsample of pots).

Meteorological data

The air temperature and relative humidity in the glasshouse compartments were recorded using capacitive hygrometers (50Y; Campbell Scientific Ltd, Shepshed, UK) placed in a standard radiation shield. The apex temperature was estimated as being the soil temperature before tassel initiation (measured by thermistors placed 3 cm deep) and the sheath temperature after tassel initiation (monitored using copper-constantan thermocouples placed inside the sheath of the last liguled leaf). All data were stored in a datalogger (CR10X; Campbell Scientific Ltd, Shepshed, Leicestershire, UK), with measurements taken every 20 s and averaged over 900 s.

Calculation of temperature-compensated time units

The response of crop developmental processes to temperature is usually modeled using linear equations (Keating et al., 2003). However, such linear approximations do not apply to the lowest and highest temperatures in the range usually experienced by plants. This is particularly significant in chill response studies, as temperatures generally remain close to the base temperature. To account for this curvilinear response, we used the equation described by Johnson & Lewin (1946) which modifies the Arrhenius equation to account for the rates of biological processes that typically change at high and low temperatures:

image(Eqn 1)

where F (T ) is the theoretical rate F at temperature T. The numerator is a modified Arrhenius equation where k is a proportionality constant, EA is the average activation energy for the considered process, and R is the gas constant. The denominator accounts for the reversible denaturation of enzymes, determined by the enthalpy (ΔH) and entropy (ΔS) of the reactions. Temperature-compensated rates can be calculated as an equivalent rate at 20°C (J(20)) for any measured rate J(T) using the simple conversion:

image(Eqn 2)

The applicability of Eqns 1 and 2 to processes linked to plant growth and development was described by Parent et al. (2009). These authors showed how a unique set of parameters (EA = 76.8 kJ mol−1, ΔH = 285 kJ mol−1 and ΔS = 0.933 kJ mol−1 K−1) could account for different growth and developmental processes in different maize genotypes (Parent et al., 2009) and represent an improvement on the previously proposed curvilinear response curves (Yan & Hunt, 1999). Parameter k acts as a scaling factor for the temperature–response curve to the process and/or genotype considered. During the present study, this was estimated from the leaf emergence rates obtained during previous experiments carried out under controlled conditions with constant temperature regimes on the two genotypes F2 and F286 (Fig. 1). Durations in equivalent days at 20°C (d20°C) could therefore be calculated during this experiment using measured apex temperature and leaf emergence records.

Figure 1.

 Response curve of the leaf emergence rate to apex temperature for inbred lines F2 (open circles) and F286 (open squares). Vertical bars show standard deviations. The natural logarithms of parameter k (Eqn 1) determined from these phyllochron records are indicated. No significant differences were found between genotypes and a common value was used for thermal time unit calculations.

Plant measurements

For each treatment, 12 plants from each line (three per block) were chosen so that their development was close to the median for that treatment, and these plants were tagged at the stage of one liguled leaf. Two or three times a week, the numbers of visible and liguled leaves were counted and the exposed lengths of the last two emerging leaves (Ln and Ln+1) were measured. In order to ensure continuity in the description of the developmental process, a decimal leaf stage was calculated as described by Andrieu et al. (2006) using the linear relationship between Ln and Ln+1 to interpolate between discrete events of new leaf emergence.

In addition, destructive measurements were carried out c. once a week on three median plants per line under each treatment. Dissections were performed under a binocular microscope (Olympus SZ-CTV; Olympus, Tokyo, Japan; maximum magnification ×50) to count the number of initiated leaves and measure the dimensions of the apex and leaf primordia below 1 cm. A new leaf was considered to be initiated as soon as it reached half of the apex length. A decimal stage was attributed, ranging from 0 (primordium length equal to half of the apex length) to 1 (next primordium appearance), in a similar way to leaf emergence. The plastochron index equaled the number of primordia having reached half of the apex length plus the decimal leaf stage. Organ dimensions above the 1-cm threshold (blade and sheath lengths) and whorl length (distance between the apex and the V formed by the last two emerging leaves) were measured using a ruler.

Cell dimension analyses

The epidermal cell length was measured on three median plants per line for mature leaves of ranks 4, 5, 7 and 9. Each leaf blade was divided longitudinally into five zones of equal length. In the middle of each zone, a transparent negative film of the adaxial epidermis was obtained after evaporation of a varnish applied to the upper surface of the leaf. The films were removed from the leaf using a strip of Scotch tape and then fixed onto a glass slide for analysis under a microscope (Leica DM2000; Leica Microsystems, Wetzlar, Germany) coupled to an image analyzer. The lengths of c. 50 epidermal cells were determined using ImageJ software macros (http://rsbweb.nih.gov/ij/; Abramoff et al., 2004). Measurements were predominantly made on cell files parallel to the midrib that contained no stomata. The total number of cells in the longitudinal direction was calculated for each leaf by integrating average cell size over the five zones sampled.

Analysis of leaf extension kinetics

Leaf extension kinetics were analyzed by fitting a multi-phase model to the destructive measurements performed during the experiment at each leaf position. The model described leaf elongation as a simple biphasic process: the first phase represents the exponential development of an elongation zone from the leaf primordia, whereas the second phase determines a steady functioning of the established elongation zone until maximum leaf length is reached (Andrieu et al., 2006). The phases are given explicitly as follows:

image(Eqn 3)

(L(t), the leaf length L at temperature-compensated time t; t0, the time at which the leaf primordium is initiated (d20°C after emergence); L0, L1 and Lmax, are the initial primordia length, leaf length at the end of the exponential phase and the final leaf length, respectively; DEP, the duration of the exponential phase; DLP, the duration of the linear phase.) Rates of growth during the two phases are described by parameters r1 during the exponential phase (d20°C−1) and LER (leaf elongation rate, mm d20°C−1) during the linear phase. The following constraints ensured continuity of the model function at the two phase transitions:

image(Eqn 4)
image(Eqn 5)

An additional constraint on the continuity of the first derivative at the exponential-to-linear phase change (t0 + DEL) was imposed as suggested by previous studies (Andrieu et al., 2006):

image(Eqn 6)

Model parameters were estimated for lamina and total leaf extension kinetics using the maximum likelihood fitting procedure described by Hillier et al. (2005) and r software (http://www.r-project.org/). Parameter t0 was not estimated along with the other parameters. It was instead calculated from the linear regression between leaf initiation records and temperature-compensated time (d20°C after emergence) and kept fixed during model parameter identification. For the first five preformed leaves, t0 was set to zero, making the initial leaf length L0 correspond to its length at plant emergence.

Statistical analysis

Analyses of variance were performed using the r statistical software (http://www.r-project.org/) to test for significant differences between means. Analysis of covariance (ANCOVA) was used to compare slopes and intercepts of linear relationships (Statistica 6.0; Statsoft, Tulsa, OK, USA).


Impact of transient chilling stress on plant development and final organ size

Fig. 2 shows the time course of the plastochron index, leaf emergence and collar appearance after plant emergence under the different treatments applied. For each genotype, typical linear responses were found for the relationships between the numbers of initiated/visible leaves and the temperature-compensated time. No significant differences were found between temperature treatments regarding plastochron (3.18 and 3.37 d20°C for F286 and F2, respectively; ANCOVA; ≥ 0.34) and phyllochron (5.19 and 5.02 d20°C for F286 and F2, respectively; ANCOVA; ≥ 0.10) values. Collar appearance systematically showed a broken-line pattern with a break of slope located c. 45 d (d20°C) after emergence. Collar appearance rates were unaffected by the cold treatments during the first period (0.157 and 0.124 leaf d20°C−1 for F286 and F2, respectively; ANCOVA; ≥ 0.13), but were significantly higher in T2 and lower in T1 during the second period, in both the F286 (ANCOVA; P < 0.001) and F2 (ANCOVA; P = 0.023) lines. On the whole, the times between leaf initiation and collar appearance remained unaffected by chilling temperatures for leaf ranks up to leaf 9, and were either slightly shortened (T2) or lengthened (T1) by 1–4 d (d20°C) for leaf ranks higher than leaf 9.

Figure 2.

 Plastochron index (triangles), decimal leaf stage (circles) and timing of collar appearance (squares) as a function of time after the emergence of maize (Zea mays) inbred lines F286 and F2 under the three temperature treatments T0 (closed circles), T1 (open circles) and T2 (grey circles). Time is expressed in temperature-compensated units, as equivalent days at 20°C (d20°C). Hatched boxes represent the period of transfer to the cold glasshouse for treatments T1 and T2.

The two transient stresses (hatched boxes in Fig. 2) affected expanding leaves at different stages of their development. During T1, leaves 10–12 were initiated, 6–9 were in the early phase of growth and 3–5 had either already emerged or emerged during the stress period. During T2, leaves 10–15 were in the early growing stages, while leaves 6–9 had emerged or were emerging. Leaves up to ranks 1 and 4 had completed their growth before the beginnings of the T1 and T2 stress periods, respectively. Leaves 2 (T1) and 5 (T2) completed their growth during the stress period shortly after transfer and had reached their final lamina length before transfer.

The consequences of transient chilling stresses T1 and T2 on final organ dimensions are presented in Fig. 3. Under all treatments, the lowest leaf rank with a significant reduction in total length (sheath + blade) was the lowest from the bottom that was still in expansion during the stress period (i.e. leaves 3 and 6 for T1 and T2, respectively). Reductions in sheath length occurred one (T1) or two (T2) leaves before those on the laminae, so that the first reduced blade was on the first leaf emerging after the stress (Fig. 3a,b). The most marked decreases in length were seen to affect the following three to five leaves. These reductions propagated until the topmost laminae in F2, but not in F286. Conversely, the sheath lengths of upper leaves remained at the level of the control value under all treatments. Line F286 displayed a more rapid recovery than F2. Transfer to T2 induced a quantitatively more marked response in terms of final lamina length reduction than T1.

Figure 3.

 Final lamina (a) and sheath lengths (b) of maize (Zea mays) inbred lines F286 and F2 under the three temperature treatments T0 (closed circles), T1 (open circles) and T2 (grey circles). Horizontal bars show standard deviations. Letters indicate Newmann–Keuls homogeneous groups at a given leaf position (α = 0.05). ns, nonsignificant.

Kinetics of lamina extension

Fig. 4 illustrates the time course of lamina length for leaves at various positions along the stem under treatments T0 and T1. When expressed in calendar terms (Fig. 4a), the rate of lamina elongation decelerated drastically during the transfer, so that the kinetics appeared to be markedly out of phase with T0 from that moment on. To overcome this limitation and to determine the chill stress-specific responses of elongation, time was expressed in temperature-compensated units (Fig. 4b). Surprisingly, using this method, the most striking differences in the kinetics of lamina elongation between chill-stressed and control plants were observed in leaves that did not necessarily differ in their final length. Leaves that were stressed during the linear phase of expansion (i.e. ranked 4 and 5 in T1) displayed only slight differences in their kinetics as compared with T0, except regarding the final length reached. By contrast, leaves exposed to chilling temperatures during the early exponential phase (i.e. ranked 9 and 11 in T1) reached similar final lengths despite the kinetics of expansion contrasting with those in T0. Differences were quantified by fitting the two-phase model of extension to the data collected for each individual leaf rank and by comparing parameter values.

Figure 4.

 Examples of leaf elongation kinetics (a) and biphasic-model fit (b) for maize (Zea mays) laminae at different positions. Data are for the F286 inbred line under the temperature treatments T0 (closed circles) and T1 (open circles). Time is expressed in calendar time in the upper panel and in temperature-compensated units, as equivalent days at 20°C (d20°C), in the lower panel. The insert represents the same curves on a log scale. Hatched boxes represent the period of transfer to the cold glasshouse.

The model could be fitted appropriately to the extension of laminae 3–13 (Fig. 5). Phytomers 1–2 could not be fitted because of insufficient data, whereas phytomers 14–15 displayed too much interplant variability in their final lamina length to allow any unique fit across dissection dates to be considered meaningful. Independent of leaf rank and genotype, the initial primordium length (L0) remained unaffected by chilling temperatures. Higher values for leaves 3–5 reflected the fact that L0 corresponded to the length at emergence for preformed leaves (initiation time t0 set at zero). The following exponential phase of elongation was represented by parameters for relative growth rate (R1) and duration of the exponential phase (Eqn 3). Estimated parameter R1 did not vary significantly between temperature treatments. As a result, and because of constant L0 values, the kinetics in the very early growth stages did not differ between control and stressed plants (e.g. log-scale insert in Fig. 4). However, the estimated durations of the exponential phase were reduced significantly in both leaves exposed during their exponential phase to chilling temperatures and leaves initiated after the end of the transfer period (i.e. leaves 7 and above in T1; leaves 10 and above in T2). When entering the linear phase, these leaves were therefore shorter than in T0 (parameter L1; not shown). As expected, leaves that had completed their exponential growth before transfer all had parameter values that were similar to those of the control.

Figure 5.

 Biphasic-model parameters of leaf extension as a function of leaf position for maize (Zea mays) inbred lines F286 and F2 under the three temperature treatments T0 (closed circles), T1 (open circles) and T2 (grey circles). Vertical bars show standard deviations of parameter estimates. Note the logarithmic scale used in the L0 panel.

Leaf elongation during the linear phase was specified using two additional parameters: the leaf elongation rate (LER) and the duration of the linear phase (Eqn 3). Both were strongly affected by temperature treatments. It was possible to distinguish between the behaviors of a leaf cohort exposed to chilling temperatures during its linear phase of growth (i.e. leaves 3–5 in T1; leaves 6–8 in T2) and another cohort exposed during or before its exponential phase of growth (i.e. leaves 7 and above in T1; leaves 9 and above in T2). In the first case, LER remained unaffected but the duration of the linear phase was significantly shorter, resulting in a shorter final length (Lmax). Conversely, in the second cohort exposed to low temperatures during its exponential growth, LERs were severely reduced, although the durations of the linear phase were similar (leaves in their late exponential phase) or significantly longer (leaves initiated or in their early exponential phase) than in the controls. The two types of variation tended to compensate for each other. The reduction of the difference in final size was only partial in the first leaves but achieved completeness for later leaves. The leaves in this cohort all displayed strongly modified elongation patterns as compared with T0, with a shorter exponential phase and lower LER. Differences in final length tended to decrease with increasing leaf rank. It should be noted that, because of the continuity constraint at the exponential-to-linear transition, the duration of the exponential phase and LER are not independent parameters. Parameter estimates could be expected to reflect more accurately the LER value (integrating a longer time period and a larger subsample of data) because of the weighting procedure applied during the optimization process.

Differences were found between the two genotypes regarding the magnitude of parameter variations between the stressed and control treatments. In T0, inbred line F2 displayed higher LER values than F286 for almost the entire leaf profile (ranks 4–11). Other parameters had roughly similar values in F2 and in F286 for all leaf ranks. In T1 and T2, variations in LER values, the durations of the exponential phase and the duration of the linear phase in the leaf cohort exposed to chilling during its exponential phase were always lower in F2 than in F286. The duration of the linear phase in particular increased far less than in F286 and never compensated completely for LER reductions. As a result, the final length (Lmax) of the topmost leaves did not differ from that of the controls in F286 but was significantly lower in F2 for both T1 and T2.

Impact on cell size and cell numbers

The total numbers and average final length of epidermal cells were examined on laminae at four positions along the stem (Table 1). All the leaf blades sampled had been stressed during their elongation, except for leaves 4–5 in T2 which had reached their final size before transfer. All displayed a significant difference in lamina length vs the T0 control, except leaves 4–5 in T2 and F286 leaf 9 in T1 (Fig. 3a). Total cell numbers were significantly reduced in leaves exposed to chilling temperatures during the exponential phase of their growth (leaves 7–9 in T1; leaf 9 in T2) or the early linear phase (leaf 7 in T2). Decreases in cell numbers reached 9% in T1 and 26% in T2. Conversely, leaves exposed during the late linear phase of their growth (leaves 4–5 in T1 and 7 in T2) produced cell numbers similar to those in the controls. Average cell lengths followed almost the opposite pattern. Leaves exposed to chilling temperatures during the linear phase had shorter cells than in T0 (cell length reductions reaching 14%) whereas the cell length of leaves exposed during the exponential phase tended to be unaffected or longer (F286 leaf 9 in T1 and T2). As a result, reduced cell length accounted for the totality of lamina length reduction in leaves stressed during the late phase of extension. By contrast, increased cell length tended partly or totally to compensate for the reduction in total cell numbers in leaves stressed during early growth. One exception was the F2 line in T2, where leaf 9 shorter cells amplified the reduction in total cell number.

Table 1.   Final epidermal cell length and cell numbers for maize (Zea mays) inbred lines F286 and F2 under the three temperature treatments T0, T1 and T2
Leaf 4Leaf 5Leaf 7Leaf 9Leaf 4Leaf 5Leaf 7Leaf 9
  1. Letters indicate Newmann–Keuls homogeneous groups at a given leaf position (α = 0.05). Levels of significance at P < 0.05, P < 0.01 and P < 0.001 are indicated by *, ** and *** respectively. ns, nonsignificant. Standard deviations are indicated in brackets.

  2. 1N = 400–500.

  3. 2N = 3.

Cell length (μm)1
 T0152a (44)142a (38)119a (27)112b (27)191a (58)168a (42)145a (35)137a (30)
 T1131b (39)134b (39)123a (28)120a (32)166b (47)155b (39)147a (35)136a (31)
 T2146a (43)142a (39)117a (30)118ab (26)185a (55)162a (42)135b (33)118b (33)
Cell number (leaves per file)2
 T02860a (124)4114a (96)7405a (107)8076a (394)3243a (87)4670a (172)6934a (92)7158a (171)
 T12969a (120)4052a (140)6809b (180)7372b (154)3291a (262)4489a (225)6586ab (315)6752a (384)
 T22947a (129)4090a (340)6806b (351)5980c (92)3395a (350)4892a (149)6408b (369)5612b (334)

Impact of transient chilling stress on whorl length

The impact of temperature treatments T1 and T2 on the time course of whorl length throughout the experiment is presented in Fig. 6. The genotypes differed with respect to whorl formation. Line F286 produced leaves that grew in a more rolled form than F2, and panicle initiation was delayed by a few days when compared with F2 (not shown). Consequently, whorl length, made up of the last sheath plus the tube formed by unfolded leaves, was significantly longer in this genotype (ANOVA; P < 0.05 at the peak). In contrast, temperature treatments exerted no significant effect on whorl length in either line. The slight negative impact observed on sheath lengths (Fig. 3b) was not significant at the whorl level when cumulated with the length of the stack of unfolding leaves.

Figure 6.

 Whorl height (distance between apex and whorl opening) as a function of time after the emergence of maize (Zea mays) inbred lines F286 and F2 under the three temperature treatments T0 (closed circles), T1 (open circles) and T2 (grey circles). Time is expressed in temperature-compensated units, as equivalent days at 20°C (d20°C). Standard deviations are indicated by vertical bars. Hatched boxes represent the period of transfer to the cold glasshouse for treatments T1 and T2.

Nonetheless, the temperature treatments applied allowed us to generate contrasting whorl length dynamics during and following transfers. Whorls increased in length long after T1 but quickly decreased after T2. Interestingly, a linear relationship was found between long-term leaf length reduction and whorl size at emergence for leaves in the early phase of growth at the time of the stress (Fig. 7). This relationship was able to account for the two genotypes independently of the transfer period.

Figure 7.

 Relationships between (a) whorl length and final lamina length and (b) whorl length and lamina length reduction as compared with the control treatment T0. Data are for the sixth leaf emerging after transfer, which was in the early exponential phase during the chilling stress. r2 is 0.96 and 0.92 for the first and second relationships, respectively.


Transient chilling temperatures affected the leaf extension dynamics of all leaves emerging after the stress but not all final leaf lengths

The effect of transient chilling temperatures was visible, with a short delay, in the final sheath length of the leaf emerging at the time of transfer and in the lamina length of the following leaf. This finding confirmed the relationship found in field experiments between the timing of the occurrence of cold and the rank of the first leaf affected (Louarn et al., 2008). Growing leaves that had reached a more advanced stage of development at the time of transfer did not display any symptoms of stress with respect to their final size, extension dynamics or cell characteristics. The one-leaf lag between sheath and blade reductions, and the nonsignificant effect on leaves in the late linear phase, are likely to be explained by a stage-dependent effect of cold on the functioning of the growing zone. Because of the temporal separation of cell proliferation and cell elongation between these grass leaf parts (Skinner & Nelson, 1994), the susceptible cellular process (respective roles of cell division and cell extension discussed in the last section) probably ended up in the most distal part of the leaf while still remaining active in the proximal part.

A stage-dependent effect was also observed in the extension kinetics of laminae. As widely reported previously (Ben Haj Salah & Tardieu, 1995 in maize; Lafarge & Tardieu, 2002 in sorghum (Sorghum bicolor (L.) Moench); Durand et al., 1999 in fescue (Festuca arundinaceae)), the time course of leaf elongation was strongly affected by temperature treatments when expressed in calendar time (Fig. 4a). However, when time was expressed in temperature-compensated units this did not erase the temperature effects at all leaf positions, as found with nonstressing temperatures (Lafarge & Tardieu, 2002). Leaf behaviors could be distinguished according to their stage of development at the time of cold transfer. The kinetics of those in the pseudo-linear phase of extension were modified less. LER (expressed in mm d20°C−1) remained unchanged relative to the controls but the duration of linear elongation was reduced. Leaves at this stage had already developed a growing zone (DZ + EOZ) close to its maximal length (Durand et al., 1999; Fournier et al., 2005; Parent et al., 2009). By contrast, leaves affected at a younger stage of development (DZ and EOZ still expanding) or initiated during the stress period displayed a shortened exponential phase that was associated with lower LER during the fast elongation period. These effects lasted long after removal of the temperature constraint, until final length was reached (Fig. 8a). Chilling stress could thus be distinguished from reversible stresses such as drought, the removal of which allows the leaf to recover growth characteristics similar to those of the control (Yeo et al., 1991; Durand et al., 1995). Despite the apparent sensitivity of early growth stages, this study did not reveal any significant differences in initial primordium length (as represented by the parameter L0) and relative growth rate during the exponential phase (parameter R1).

Figure 8.

 Time course of the changes in whorl length and individual leaf extension (a) and the corresponding lamina lengths reached at the exponential to linear phase transition (L1 parameter) (b). Data are for the leaves presented in Fig. 5 and Table 1. Time is expressed in temperature-compensated units from plant emergence. Horizontal dashed lines in the lower panel represent the maximum length of the growth zone as indirectly estimated using the model described by Durand et al. (1999), fitted to our data. This shows that L1 is unlikely to represent a meaningful estimate of the growth zone length.

Strikingly, these alterations in growth kinetics did not correlate with final leaf length reductions. While leaf elongation kinetics were affected for the whole leaf profile in both lines (Fig. 3), reductions in final leaf length were observed in a limited number of phytomers in F286 (ranks 3–7 and 6–8 for sheaths in T1 and T2, respectively; ranks 4–7 and 7–12 for blades in T1 and T2, respectively; Figs 3, 5). Significantly shorter laminae in T1 corresponded to leaves in the rapid extension phase with kinetics identical to those of the controls (Fig. 4), except for a shorter duration. However, upper leaves displayed markedly reduced LER values, but these were compensated by a longer duration of the rapid extension phase so that the final size reached was close to that of the controls. In T2, the trends were similar, but with less compensation. The effect of our transient chilling treatments thus disrupted the usual relationship, making grass leaves with higher LER longer (Assuero et al., 2004; Fournier et al., 2005), but shared many similarities with the effects of density as determined by Andrieu et al., 2006. The duration of the fast elongation period appeared to be critical to explaining the range of reduction in the final leaf length of the two lines. This period was initially shortened for leaves in the linear phase of growth, but then increased for subsequent leaves. This confirms that a second mechanism is involved in regulating leaf extension kinetics, probably triggering the end of leaf growth independently of the impact of cold.

The size-mediated effect did not propagate initial cold-induced leaf length reductions but on the contrary acted as a potential buffer against short-term stress

The impact of the whorl size-mediated effect (Wilson & Laidlaw, 1985; Casey et al., 1999) on leaf extension kinetics and final leaf size was examined in the case of controlled transient chilling stresses. In particular, we looked for a potential role of this mechanism in propagating initial cold-induced sheath length reductions to upper leaves. The experimental design enabled us to generate contrasting combinations of stressed expanding leaves and whorl lengths (Fig. 6). Our results show clearly that whorl size did not propagate initial cold-induced leaf length reductions. First, no impact of temperature treatments was found on whorl length during the experiment. The negative impact observed on sheath lengths (up to 10%) was not found at the whorl level when cumulated with the length of the stack of unfolding leaves (representing c. 1/3 of whorl height, with a larger plant-to-plant variation than maximum sheath length reduction). Second, and unlike the findings of field experiments (Louarn et al., 2008), initial sheath and blade length reductions did not propagate to the topmost leaves in F286 and only to a limited extent in F2. Severe blade length reduction was found only in a limited number of phytomers emerging just after cold occurrence.

We argue, however, that a size-mediated effect was exerted on leaf elongation during this experiment, but in a reverse manner. It did not propagate or amplify an initial stress, but it did limit its impact. Several lines of evidence strongly support this new facet of the whorl-size effect hypothesis. First, we found that leaf length reductions were smaller and visible in fewer consecutive phytomers when whorl lengths were greater during and following the period of stress. A ranking of the treatments applied according to whorl length was F286-T1 > F2-T1 > F286-T2 > F2-T2. This perfectly matched the buffering effect observed on final leaf length in these treatments (Fig. 7). Second, the mechanism of action on lamina length reported in the literature for such a size-mediated effect would explain the longer duration of the linear phase and the compensation in the lengths of leaves affected during the early growing stages. Indeed, because leaf tip emergence and ligule differentiation are highly coordinated processes in grass leaves (Girardin et al., 1986; Schnyder et al., 1990; Skinner & Nelson, 1994; Fournier et al., 2005; Andrieu et al., 2006), maintaining whorl length, even in stressed plants (as shown in Fig. 6), could be expected to increase the duration of rapid extension to balance any reduction in extension velocity. Such a size-mediated effect would not impact, or would impact positively, LERs during the linear phase of extension (Casey et al., 1999; Kavanova et al., 2006). Following this reasoning, the net decrease in the LERs of young stressed leaves thus appeared to result from a direct impact of chilling temperatures, whereas wheras whorl size contributes, as suggested in the previous section, to the mechanism that triggers the end of elongation. Similar results were reported by Kavanova et al. (2006), who demonstrated that phosphorus deficiency directly affects the tissue expansion rate while the size-mediated effect mainly controls the position in the growth zone at which expansion stops. Finally, our results concerning fully expanded cell size and final cell number also suggested a potential role for a size-mediated effect in regulating final leaf length (discussed in more detail in the next section).

To explain cold-induced leaf length modifications in the field, the present data set suggests a role for a direct impact of low temperatures (rather than only size-effect propagation) on the growth of young leaves. It was shown that transient chill could potentially affect the kinetics of all nonemerged growing leaves at the time of the stress. The magnitude of the actual plant size reductions caused by this stress depended on the timing of the stress during the vegetative growth period: within the range of timings tested, later stress had a stronger effect. In temperature-fluctuating environments, the integration over time of the short-term effect described here could explain most of the apparent propagation of initial reductions, if plants were repeatedly exposed to suboptimum temperatures. Such conditions are likely to occur several times during the vegetative period in the northern fringe of the maize cultivation area (Louarn et al., 2008). A two-step effect related to stress intensity (first a direct effect of stress on the extension of individual leaves, and then the addition of size-mediated and direct stress effects as soon as whorl size is significantly reduced), as proposed in the case of phosphorus deficiency (Kavanova et al., 2006), would also contribute to explaining the patterns reported under field conditions.

Cell division and cell extension both controlled leaf extension dynamics and final leaf size under cold stress

We examined total cell numbers (cells per line) and the final length of epidermal cells to enable a quantitative assessment to be made of the net impact of transient chill on cell division and cell expansion processes with respect to leaf size. In contrast to previous studies carried out on a single leaf rank (Ben Haj Salah & Tardieu, 1995; Rymen et al., 2007), we found that cold temperatures affected both total cell numbers and final cell lengths (as shown, for instance, in roots by Pahlavanian & Silk, 1988). However, the outcome of the stress was more visible on one or other component depending on the leaf developmental stage during the stress (Table 1). For leaves stressed during the linear phase just before tip emergence, cell expansion was reduced and accounted for a major part of the reduction of lamina length. Ben Haj Salah & Tardieu (1995) also reported a reduction of c. 12% in cell size in leaves at a similar stage at the time of transfer to a gentle chilling treatment (25°C → 13°C). At this stage, the ligule of grass leaves differentiates in DZ. From this time on, an increasingly large part of the DZ is dedicated to sheath cell production until the ligule finally flows out of the DZ a few days later (Skinner and Nelson, 1994). Because most lamina cells had already been produced, it is thus unsurprising that leaves at this stage did not display a significant impact of cell division on final lamina length. Similarly, because many of the blade cells were passing through the EOZ, the effect of chilling on cell wall properties and cell expansion (Shewfelt, 1992) was expected to be substantial. Our results confirmed such an effect.

These patterns were more difficult to analyze in younger leaves where both cell division and cell expansion were likely to contribute significantly to any lamina length modifications. In leaves that were just being initiated or in the early exponential phase (DZ + EOZ still expanding), total cell numbers, and thus cell division, were severely affected. Rymen et al. (2007) reported similar results for the fourth emerging leaf after transfer (night chilling; 25°C : 25°C → 25°C : 4°C) in which a slowing down of the cell cycle and reduced cell production explained chilling injuries. It is noteworthy that the cell length reductions demonstrated in the present study were only observed in this leaf cohort in F2-T2 (shorter whorls). Under other treatments they were either not observed (F286 leaf 7 in T1 and F2 leaves 7 and 9 in T1, as in Rymen et al., 2007) or resulted in an increase in the final size of cells (F286 leaf 9 in T1). Therefore, cell extension did not systematically amplify the reduction of cell proliferation (and sometimes even contributed to compensating for it). Interestingly, the changes in cell elongation matched those discussed in the previous section for the duration of rapid elongation and with the corresponding variations in whorl length (F286-T1 > F2-T1 and F286-T2 > F2-T2). This correlation suggests a possible impact of size-mediated effects on cell elongation, as shown, for instance, in artificial sheath lengthening experiments (Wilson & Laidlaw, 1985).

Indirect estimates of growth zone lengths, as represented by parameter L1 in this study (Fig. 8b), are unsuitable for inferring the relationship between cellular processes and growth zone functioning. Because of the continuity imposed on the first derivative for parameter identification, L1 is indeed simply a surrogate for the duration of the exponential phase (L0 and R1 being constant) that enables balancing of the average LER during the pseudo-linear phase of growth. A precise characterization of dynamic cellular processes in the growing zone, together with the direct measurement of DZ and EOZ lengths over time, is required to gain more insight into the respective effects of cell proliferation and extension on leaf elongation and thus to allow the proposal of a realistic mechanism for these effects (Gandar & Rasmussen, 1991; Silk, 1992; Durand et al., 1995). Nonetheless, the present whole-plant analysis was sufficient to invalidate the hypothesis of a purely cell-cycle-centered regulation of leaf elongation under cold temperatures (Rymen et al., 2007). It highlighted the fact that single-leaf analysis, even at the finer scale, is unlikely to produce the broad picture that is required to disentangle the effects of cellular processes on final leaf length in the case of long-lasting stress effects such as cold. In particular, the intricacy of rapidly responding processes at the n − 1 scale (cell proliferation and expansion; Tardieu et al., 2000) and more slowly responding processes at the n + 1 scale (whole-plant whorl size; Kavanova et al., 2006) needs to be taken into account with regard to effects on leaf elongation kinetics. To tackle this problem it would be wise to apply classical cell flux analysis to several leaves stressed at different stages of development and/or to plants of different whorl size (this information is lacking in currently published data sets). The consideration of nonsteady-state phases of growth could also improve the analysis of apparently stage-dependent stressing effects (Parent et al., 2009).


Our data suggest that the outcome of transient chilling stress in terms of leaf extension dynamics and final leaf length is related to both a direct negative impact of cold on the leaf elongation rate and a potential buffering effect of whorl length throughout the duration of leaf elongation. Such a buffering effect (varying in time; potentially cancelled as soon as whorl length itself is reduced) may be responsible for hysteresis in the response of individual leaves to stress, and explain stage-dependent whole-plant responses to some extent. From a modeling perspective, it implies that simple growth models that scale up single-leaf extension to the plant level by assuming that all leaves are independent (Chenu et al., 2008) are likely to fail to capture plant behavior in a fluctuating environment. A proper quantification requires the description of both structure and function in a plant growth model which accounts for specific organ growth responses and generic self-regulation rules (Verdenal et al., 2008).


This study was supported by the Conseil Régional de Picardie (http://www.crpicardie.fr) and INRA (http://www.inra.fr), France. We are grateful to J. Hillier for making the fitting procedure available and to V. Bäcker for providing tools for the cell dimension analyses. We would also like to thank D. Fiorillo, S. Hervieu, and J-F. Hû for their assistance with the experiments.