Climate-dependent variations in leaf respiration in a dry-land, low productivity Mediterranean forest: the importance of acclimation in both high-light and shaded habitats

Authors


*Correspondence author. E-mail: oka1@york.ac.uk

Summary

  • 1Climate-driven changes in leaf respiration (R) in darkness have the potential to determine whether low productivity ecosystems exhibit positive or negative carbon balances.
  • 2We investigated whether sustained exposure to full sunlight, shade and seasonal drought alters the temperature response of leaf R of field-grown Quercus ilex subsp. ballota in a dry-land continental Mediterranean ecosystem. The plants studied, experience large diurnal and seasonal variations in temperature.
  • 3Whilst growth irradiance impacted on photosynthesis, it had little effect on the short-term temperature dependence of leaf R. Moreover, although basal rates of leaf R (i.e. rates of R at a common measuring temperature) were higher in sun-exposed than shade-exposed leaves, growth irradiance had little impact on the degree of acclimation to seasonal changes in temperature and/or moisture. Basal rates of leaf R were higher in winter than summer in both sun-exposed and shaded plants. Estimated Q10 values (i.e. proportional increase in R per 10 °C rise in temperature) for leaf R were greater in winter than summer; however, no seasonal variation was found in the apparent activation energy (E0) of leaf R. These observations were used to construct a simple Arrhenius model that fully accounted for both daily and seasonal variations in the temperature dependence of R in both sun-exposed and shaded plants. Crucial to the model was accounting for the seasonal and irradiance-dependent shifts in the basal rate of leaf R.
  • 4Although the balance between daily R and photosynthesis increased markedly in summer (particularly under full sun), the increase in this ratio was markedly less than would have been the case if leaf R had not acclimated to the high average day time temperatures in summer.
  • 5It is concluded that seasonal acclimation of leaf R plays a crucial role in determining the viability of tree growth in dry-land, low productivity forest ecosystems.

Introduction

Over half of the Earth's terrestrial surface is covered by ecosystems that exhibit low net primary productivity (NPP – the rate of net carbon gain by vegetation) (Larcher 2004). Under such conditions, plants may be expected to often operate near the threshold of positive carbon gain. An example of such ecosystems are dry-land forests in Mediterranean regions where rates of carbon loss by plant respiration (R) often equal, or exceed, the rate of carbon uptake by photosynthesis (P), except at the most favourable times of year (typically spring and autumn; Tenhunen et al. 1990; Rambal et al. 2003). Exposure to high air temperatures, drought, deep shade and/or nutrient deficiencies would be expected to further exacerbate any imbalance between plant R and P (Björkman 1981; Dewar, Medlyn & McMurtrie 1999; Walters & Reich 2000; Ciais et al. 2005; Bréda et al. 2006). Although it is widely accepted that plant R can play an important role in determining whether low NPP ecosystems exhibit a positive or negative carbon balance (Pereira et al. 1986), particularly following drought (Ciais et al. 2005; Bréda et al. 2006), the quantitative impact of plant R on the net carbon balance of low NPP ecosystems is unclear.

Under favourable conditions approximately half of whole-plant R may take place in leaves (Poorter, Remkes & Lambers 1990; Atkin, Scheurwater & Pons 2007), with up to 35% of CO2 fixed each day by photosynthesis being released back into the atmosphere by leaf R (Loveys et al. 2003). This percentage could be even greater in field-grown plants, where net CO2 exchange is often limited by reduced stomatal conductance (e.g. due to drought), low irradiance (e.g. in the understorey of mature forests) and high leaf temperatures (where net assimilation is reduced due to high rates of photorespiration). The percentage of daily fixed CO2 respired by leaves is also dependent on the response of respiratory metabolism in leaves to abiotic factors such as water availability (Flexas et al. 2006), growth irradiance (Wright et al. 2006), and short- and long-term changes in temperature (Atkin et al. 2005). Although initial stages of water stress can inhibit leaf R, sustained exposure to severe water stress often results in leaf R increasing, probably in response to an increase in energy demand as leaves cope with drought (Flexas et al. 2006). Growth irradiance also impacts on leaf R, with respiratory rates being lower in shade leaves than in their high-light grown counterparts (Boardman 1977; Sims & Pearcy 1989, 1991, 1994; Turnbull, Doley & Yates 1993; Noguchi & Terashima 1997; Noguchi, Nakajima & Terashima 2001). In their compilation from 20 sites around the world, Wright et al. (2006) found that rates of leaf R were highest at warm sites, reflecting the temperature dependence of leaf R. Leaf R was also highest at the high-irradiance, low-rainfall sites (Wright et al. 2006).

It has long been recognized that rates of leaf R are sensitive to short-term changes in temperature (Wager 1941; Forward 1960), with the temperature sensitivity of leaf R [quantified using the Q10 (i.e. the proportional change in R per 10 °C rise in temperature)] often assumed to be constant (Ryan 1991; Cox et al. 2000; Cramer et al. 2001). However, there is growing evidence that Q10 values can be highly variable (Breeze & Elston 1978; Ryan 1991; Azcón-Bieto 1992; Dewar et al. 1999; Tjoelker, Oleksyn & Reich 2001). For example, several studies have reported within-canopy variability in the Q10 of leaf R in some trees (Bolstad, Mitchell & Vose 1999; Griffin, Turnbull & Murthy 2002a; Turnbull et al. 2003). Moreover, Q10 values decrease as measuring temperatures increase (Tjoelker et al. 2001) and may vary in response to changes in the underlying factors regulating respiratory flux (e.g. maximum enzyme activity, substrate availability and/or the turnover of ATP to ADP; Atkin & Tjoelker 2003). This suggests that the temperature dependence of leaf R may vary in response to environmentally induced changes in substrate supply that are photosynthetically dependent.

In many global climate models, predictions of future CO2 exchange are based on the assumption that the temperature dependence of leaf R remains constant following long-term changes in temperature (i.e. Cox et al. 2000). However, in most plant species, leaf R is able to acclimate to long-term changes in temperature. As a result, rates of R measured at any given temperature are higher in cold-acclimated plants than their warm-grown counterparts (Atkin & Tjoelker 2003). This acclimation can result in a homeostasis of leaf R (i.e. constant rates of leaf R following long-term changes in temperature, when R is measured at the prevailing growth temperature) (Larigauderie & Körner 1995; Atkin, Holly & Ball 2000). Acclimation can be rapid, with substantial adjustments in leaf R occurring within 2 days of a change in growth temperature (Rook 1969; Billings et al. 1971; Atkin et al. 2000; Bolstad, Reich & Lee 2003). Although the degree of acclimation is developmentally dependent in some short-lived herbaceous species (e.g. Armstrong et al. 2006a; Armstrong, Logan & Atkin 2006b), substantial acclimation also occurs in mature leaves of long-lived species (Atkin et al. 2000; Bolstad et al. 2003; Bruhn et al. 2007) and irrespective of the growth irradiance (Zaragoza-Castells et al. 2007). Thermal acclimation in long-lived evergreen leaves could, therefore, play a pivotal role in allowing the maintenance of positive carbon balances for low NPP forest ecosystems such as in Mediterranean dry-land regions (in full sun and/or in the shaded understorey of large trees). This may especially be the case in summer when extreme leaf temperatures occur and when photosynthesis may simultaneously be limited by severe soil water deficits (Osborne et al. 2000).

This study quantifies the role of leaf respiratory CO2 release plays in determining foliar net carbon balances for a Mediterranean dry-land ecosystem in central Spain. We measured diurnal and seasonal variations in leaf R, and net photosynthesis (Pnet) of a dominant long-lived evergreen tree species (Quercus ilex subsp. ballota) over a 12-month period in 2004–2005. Photosynthetic and respiratory rates were measured at the ambient temperatures experienced in the field. Measurements were made at two sites characterized by contrasting growth irradiances [sun-exposed (open forest) and shade (understorey forest)]. The data were then used to model daily and seasonal variations in leaf R. Three hypotheses were tested: (1) due to climate-mediated changes in photosynthesis and accumulation of sugars, the temperature sensitivity of leaf R will be lower in plants growing in shade; (2) the degree of acclimation of leaf R to seasonal changes in the environment is similar in sun-exposed and shaded plants; and (3) the balance between R and Pnet in field-grown plants is not constant, varying seasonally, and differing between sun-exposed and shaded plants.

Materials and methods

site description

The field site was located 175 km east of Madrid in the Iberic System Mountain Range (950 m a.s.l.) at Los Cerrillos Biological Station, Villar de Cobeta (40°48′ N 2°12′ W), within the Alto Tajo Natural Park (Guadalajara, Spain). The experimental area was located on a steep (20°), south-facing slope, with Q. ilex subsp. ballota dominating the vegetation. The area has a continental Mediterranean climate, with hot, dry summers and cold winters, and is characterized by large diurnal and seasonal variations in air temperature (diurnal variations in temperature in the order of 25 °C are common, with leaf temperatures reaching near 50 °C in summer and –15 °C in winter) (Fig. 1). The site was characterized by limestone outcrops and a limited capacity of the soil to retain water. Selected individuals of Q. ilex were small shrubs 20–110 cm height growing under two contrasting growth irradiances [full sun and in the shaded (40%–75% of full sunlight) understorey of mature trees].

Figure 1.

(a) Variations in ambient total daily irradiance (mol photons m−2 d−1) open dots, and precipitation (mm) black bars, throughout the field study (from August 2004 to August 2005). Black arrows represent months upon which measurements of leaf gas exchange took place in Quercus ilex. (b) Variations in air temperature (°C) throughout the field study (from August 2004 to August 2005). Open triangles represent the maximum temperature, closed dots the mean temperature, while the open dots represent the minimum temperature.

Meteorological stations were installed in the full sun and shaded habitats. Each meteorological station was installed in June 2004 and included sensors for air temperature and relative humidity (Hobo H08-032-08, Onset, Pocasset, MA), soil moisture (ECH2O EC-20, Decagon Devices, Pullman, WA) and solar irradiance (Apogee quantum sensor QSO-SUN, Logan, UT), cross-calibrated with a Li-Cor SA Li-190 quantum sensor (Li-Cor, Lincoln, NE). Readings of each sensor were recorded every 10 min with a Hobo H08-006-04 data logger. Precipitation was recorded in the open with a Rain-o-matic small rain gauge (Pronamic Co. Ltd. Silkeborg, Denmark) attached to a Hobo H7 event data logger. Soil volumetric water content was estimated with a portable moisture device using the Time Domain Reflectometry (TDR probes) method, (TRIME-FM, Imko Micromoduletechnik GmbH, Ettlingen, Germany) at 20 cm soil depth, next to where the tree replicates were growing. Figure 1 shows the precipitation events, photosynthetic active radiation (PAR), and mean, maximum and minimum temperatures observed during the period of the study. Table 1 shows soil water content on four of the six visits to the field site over a 12-month period (August 2004 to August 2005).

Table 1.  Comparison between full sun and deep shade sites for water potential (MPa), maximal photochemical efficiency of PSII (Fv/Fm) and soil water content (%volume) on, or next to, the same trees used for gas exchange measurement over 4 months
Time (month)Full sun – predawnShade – predawn
Water potential (MPa)Soil water content* (% Vol.)Fv/FmWater potential (MPa)Soil water content* (% Vol.)Fv/Fm
  1. Measurements took place predawn (0630–0900 h). Mean values ± 1 SE are shown (n = 4). Within each site and measurement variable, months labelled with different letters differ significantly (P < 0·05). *The soil water content was measured around 1000–1100 h.

March 2005–1·33 ± 0·70a8·40 ± 1·23a0·38 ± 0·03a–1·05 ± 0·69a9·49 ± 1·08a0·66 ± 0·04a
May 2005–1·24 ± 1·44a3·94 ± 0·67b0·82 ± 0·01b–1·66 ± 1·83b4·28 ± 0·79b0·84 ± 0·00b
July 2005–3·56 ± 1·31b1·33 ± 0·61c0·86 ± 0·01b–3·75 ± 0·80c1·41 ± 0·55c0·86 ± 0·02b
August 2005–4·60 ± 2·61c3·45 ± 0·86b,c0·75 ± 0·02c–5·22 ± 2·03d2·00 ± 0·70c0·79 ± 0·03b

sampling regime and physiological measurements

In August 2004, four individual trees were identified at both the full sun and shaded sites, with each tree representing an independent replicate. For each field campaign, leaf physiological measurements were made using a single south-east facing, attached, fully-expanded, mature leaf from each replicate tree. Unless otherwise stated, measurements were made using single leaves from four replicate trees. Different leaves were used on each sampling month field campaign, with the measured leaves being harvested for determination of chemical–structural characteristics at the end of each sampling month campaign.

Predawn stem water potential was measured with a self-built Scholander chamber using pressurized nitrogen stem water potential and soil moisture content decreased significantly throughout the experiment at both sites (Table 1). Estimates of the predawn, dark-adapted photochemical efficiency of photosystem II (Fv/Fm; dark adapted for 30 min) were made using a portable PAM-2000 (Heinz Walz GmbH, Effeltrich, Germany); these measurements were made at the same time as the water potential measurements took place, using leaves adjacent to those used for gas exchange measurements. Fv/Fm ratios range from 0·75 to 0·85 for healthy plants (Corcuera et al. 2005).

Measurements of leaf gas exchange were carried out using a Licor 6400 portable photosynthesis system (LI-COR). Leaves were exposed to an atmospheric CO2 concentration of 400 p.p.m. (using the built-in CO2 controller), and measurements made at the prevailing ambient air temperature (Atkin et al. 2000; Griffin et al. 2002b) and at ambient relative humidity (typically 35%–60%).

Measurements of leaf R took place at regular intervals during the night, starting 1 h after sunset to avoid post-illumination transients (Azcón-Bieto & Osmond 1983; Atkin, Evans & Siebke 1998) with measurements continuing through the night to just before dawn. To expand the temperature range over which leaf R in the dark was measured, from March 2005 onwards we continued measuring R on the same leaves during subsequent daylight hours, with leaves darkened for 30 min prior to measurement (due to time constraints, these measurements could only be made using three of the four replicates used in the night-time measurements). As it was found that there were no differences between leaf R measured during the day and night at the same temperature (see Fig. 3), temperature response curves were fitted to data collected both during night and day. Leaves for which R in darkness had been determined during the day were subsequently re-exposed to ambient light to allow the plants to photosynthesize and prevent the decline of substrates (Azcón-Bieto 1992). After 1 h exposure to ambient irradiance, measurements of net photosynthesis (Pnet) were made. The measurements were carried out at regular intervals (approximately every 2 h) through the day (interspersed between measurements of R on the same leaves), with Pnet being measured at the prevailing ambient irradiance. On each sampling month, measurements were made on three to four replicate leaves.

Figure 3.

Rates of leaf respiration in darkness (R) of fully-expanded, mature Quercus ilex leaves plotted against ambient temperature (n = 3, ± SE), both for plants grown in (a) full sun and (b) deep shade, from March 2005 to August 2005. In both (a) and (b), closed symbols show the rates of leaf R at night time, whilst open symbols show the subsequent rates of R exhibited after 30 min in darkness during the day-time. Fitted lines were calculated using eqn 9. Star symbols indicate the R rate calculated at the mean temperature experienced during each month.

Following CO2 exchange measurements on each sampling month, leaves were harvested, fresh mass and area of the leaf sections used for CO2 exchange measurements quantified, after which they were submerged in liquid N2. Once back in the laboratory, leaves were oven-dried for 3 days (65 °C) and then weighed. Leaves were then pooled for each month and each site, and ground to a fine powder using a hammer mill (31–700 Hammer Mill; Glen Creston, Stanmore, UK). Soluble sugars and starch were extracted and measured as described previously (Loveys et al. 2003).

modelling respiration rates

To account for observed variation in rates of leaf R at any given temperature (T), we compared the applicability of two approaches based on the Arrhenius and Q10 functions.

The following equation provides a simple Arrhenius function standardized to 20 °C:

image(eqn 1)

where R20 is leaf R at 20 °C, E0 has the significance of an activation energy, r is the universal gas constant (8·314 J mol−1 K−1) and TK is the leaf temperature expressed in Kelvin. Taking natural logarithms for each side, eqn 1 becomes:

image(eqn 2)

which allows loge(R) to be modelled as a simple linear regression function with loge(R20) being the intercept, the independent variable being (1/TK–1/293·15) and with E0/r being the slope.

The second function was the more common Q10 equation:

image( eqn 3)

where A is a parameter which describes the overall rate and B is a constant describing the temperature sensitivity of the reaction rate, which can be expressed in terms of a Q10, as follows:

image( eqn 4)

For example, for a Q10 of 2, then B = 0·0693. As for the Arrhenius function, eqn 3 can be expressed relative to the modelled respiration rate, in this case 20 °C, as shown:

image( eqn 5)

which can also be expressed in a linear form:

image( eqn 6)

The transformations associated with eqns 2 and 6 also served to stabilize the variances for the data set collected as part of this study.

Estimates of R20 were obtained using multi-level modelling techniques (Snijders & Bosker 1999) allowing us to take into account that measurements were repeatedly made on the same leaves over any 1 day. We also examined potential effects of both growth irradiances and time of year on overall estimates of R20 and also if the temperature sensitivity of respiration also changed according to time of year. That is to say, the fit of the Arrhenius equation the model applied was:

image(eqn 7)

where Rtp denoted a respiration measurement made at time t on plant p, bl is a categorical variable allowing for respiration rate to differ between sun and shade leaves; c2, c3 and c4 are categorical variables which allow for respiration rates to vary according to time of year (May, July and August 2005, respectively) and d2, d3 and d4 are categorical variables which allow for the temperature sensitivity of respiration rates to vary according to time of year. All categorical variables (2–4) are expressed relative to the first measurement date in March 2005, and likewise bl described the respiration rates of shade leaves relative to those in the sun. Similarly, for the Q10 model:

image(eqn 8)

In both eqns 7 and 8, U0p represents a random effect which allows R20 to vary from plant to plant and U1p is a random effect allowing the slope of the temperature–respiration relationships to also vary between plants on any given day. The remaining residual variance is designated Rtp.

Equations 7 and 8 represent a reduced form of an initial model which also included the possibility of differing temperature sensitivities of respiration for sun vs. shade leaves. Nevertheless, as no significant of growth irradiance on the temperature sensitivity of leaf R was found when applying either the Arrhenius or Q10 equations (data not shown), we employ and show results from the reduced versions only here.

statistical analyses

Other than for the statistical modelling of leaf R described above which were undertaken with mlwin (Rabash et al. 2004), statistical analyses were conducted using spss version 11 (SPSS Science, Birmingham, UK). Kolmogorov–Smirnov and Levene's tests were used to test for normality and homogeneity of variances, respectively. When the variances were found to be not equal, log10 transformations were preformed. If these transformations still failed to produce equal variances then the nonparametric test equivalent was used. Two-way anovas were carried out to investigate the effects of time (month) and site (sun vs. shade) on SLA, leaf thickness, dry matter content (DMC), Pnet and R. Within a site, one-way anovas were used to investigate how the same variables changed over time (between months). Regression analyses were used to investigate the relationship between stomatal conductance and Pnet using the F-ratio method (Sokal & Rohlf 1981) to determine whether the relationships differed between sites. For average day-time Pnet and mean night-time R, significant interactions between month and sun-exposed vs. shade-grown leaves were observed. For this reason, separate one-way anovas were carried out for sun-exposed or shade-grown leaves to investigate inter-month differences.

Results

average night-time rates of leaf r

Figure 2 shows average night-time rates of leaf R on each sampling month over the 12-month period, irrespective of the prevailing temperature. Rates were significantly higher in sun-exposed plants than at the shade site (P < 0·001) and a significant interaction term was observed between months and sun-exposed vs. shade environments (P = 0·028). This suggests that patterns of seasonal variation differed between the sun-exposed and shade sites for average rates of leaf R. For this reason, one-way anovas were performed to distinguish between months within both sites. At the sun-exposed site, leaf R varied between months (Fig. 2a; P = 0·04), with R being significantly lower in March. Similarly, leaf R varied through time at the shade site (Fig. 2b; P = 0·001) with rates in January and July being significantly higher than the other 4 months (P < 0·001).

Figure 2.

Average rates of night-time leaf respiration in darkness (R) and day-time net photosynthesis (µmol CO2 m−2 s−1) exhibited by fully-expanded mature Quercus ilex leaves each month, both for sun-exposed plants (a and c) and shade-grown plants (b and d) measured at the prevailing temperature experienced by leaves on the day of measurement. Values of respiration represent rates of R measured at regular intervals during the night-time (n = 3–4, ± SE). Different letters indicate significant differences between months within each light treatment (P < 0·05).

modelling daily and seasonal variations in leaf r

Figure 3 shows leaf R for both sun-exposed and shade leaves of Q. ilex; rates measured during night-time are shown as closed symbols, with rates measured during day-time (after leaves were covered for at least 30 min before measuring) as open symbols. At both sites, leaf R was temperature sensitive, generally being lower at any given temperature in shade plants than in plants at the sun-exposed site.

We used data in Fig. 3 to model daily and seasonal variations in leaf R using the Arrhenius and Q10 approaches described in eqns 7 and 8. In preliminary modelling exercises, incorporation of different temperature coefficients (i.e. Q10 or E0) for sun- and shade-grown plants was observed to have no impact on the goodness-of-fit between models and data (data not shown; thereby contradicting Hypothesis 1); because of this, we simply assigned common temperature coefficients to sun- and shade-grown plants in subsequent modelling exercises. Table 2 shows that both models yielded similar results in terms of model fit (as seen by the similar deviances) but with one important difference – although a single parameterisation of E0 (45·13 kJ mol−1) was adequate to explain the observed daily and seasonal variation in R using the Arrhenius model, allowing Q10 values to vary between months resulted in a better fit (as shown by the significantly different d values in Q10 model; Table 2). Modelled Q10 values in March, May, July and August 2005 were: 2·02, 2·15, 1·71 and 1·77, respectively. By contrast, a more parsimonious model could be fitted ignoring the possibility of a changing E0 with time of year. This is detailed in Table 3, which shows only an insignificant loss in goodness-of–fit model (increase in deviance of 3·9 associated with the loss of three terms). Also worthy of note in the above analysis is that all models suggested significant plant-to-plant variation in the rate of R at common temperature (R20) across the four periods, as evidenced by the significant variance associated with the U0p term (Tables 2 and 3), but with no significant variation in apparent responses to temperatures (as evidenced by the insignificant variances associated with the U1p terms).

Table 2.  Estimates for the Arrhenius and Q10 models, using eqns 7 and 8 and data shown in Fig. 3
Fixed effectsArrhenius modelQ10 model
CoefficientSECoefficientSE
Intercept = loge (R20)–0·30950·0931–0·29470·0999
E0 = slope (Arrhenius model)46·953·31  
B = slope (Q10 model)  0·07030·0052
bl = shade effect0·52280·06950·51720·0723
c2 = date effect (intercept)–0·51350·1076–0·52880·1161
c3 = date effect (intercept)–0·87980·1151–0·87950·1230
c4 = date effect (intercept)–0·90350·1207–0·90480·1280
d2 = date effect (slope)–7·4576·9670·00610·0102
d3 = date effect (slope)5·7754·732–0·01670·0071
d4 = date effect (slope)3·5375·7540·01310·0083
Random effectsVar. Comp.SEVar. Comp.SE
  1. For the fixed effect terms estimates of the coefficients are shown, along with their standard errors (SE). For the partitioning of the variance for the random (error) terms, the variance component (Var. Comp) is given, along with the standard error. Terms in bold are statistically significant (P = 0·05). U0p refers to the variations in overall respiration rates (i.e. intercept) between the various trees sampled not accounted for by the model, and U1p refers to the variation in temperature response (i.e. slope) between plants not accounted for by the model. R0p is a residual term describing the remaining variance after the model fit and systematic plant-to-plant variability have both been considered.

Level two (plant) random effects
Variance (U0p)0·026490·009170·033280·01109
Variance (U1p)3·29719·860·16230·2841
Covariance (U0p, U1p)0·16230·28410·00030·0004
Level one variance
Variance (R0p)0·098910·0744370·099460·007502
Deviance254·3255·2
Table 3.  Estimates for the Arrhenius model with no effect of time of year on the activation energy (E0) using eqn 7 and data shown in Fig. 3
Fixed effectsArrhenius model
CoefficientSE
Intercept = loge (R20)–0·3350·085
E0 = slope45·132·07
bl = shade effect0·52370·0659
c2 = date effect–0·49040·1000
c3 = date effect–0·89990·1094
c4 = date effect–0·89750·1093
Random effectsVar. Comp.SE
  1. For the fixed effect terms estimates of the coefficients are shown, along with their standard errors (SE). For the partitioning of the variance for the random (error) terms, the variance component (Var. Comp) is given, along with the standard error. Terms in bold are statistically significant (P = 0·05). U0p refers to the variations in overall respiration rates (i.e. intercept) between the various plants sampled not accounted for by the model, and U1p refers to the variation in temperature response (i.e. slope) between plants not accounted for by the model. R0p is a residual term describing the remaining variance after the model fit and systematic plant-to-plant variability have both been considered.

Level two (tree) random effects
Variance (U0p)0·025440·009105
Variance (U1p)8·39221·53
Covariance (U0p, U1p)0·11720·2841
Level one variance
Variance (R0p)0·099690·0750
Deviance258·2

A reduced Arrhenius model was thus applied which did not include the possibility of changes in the temperature sensitivity of R with time of year (but which did still allow for overall rate changes with season and differences between sun and shade) namely:

image(eqn 9)

the results of which are shown in Table 3 with the fitted lines (model predictions) being shown in Fig. 3. From Table 3, for the first measurement date used in the model (March 2005), R20 was modelled to be 0·72 µmol CO2 m−2 s−1 for leaves on plant growing in the shade and 1·69 times greater for leaves on the sun-exposed plants (i.e. 1·21 µmol CO2 m−2 s−1). For the subsequent three dates (May, July and August 2005), these rates needed to be modified by the multiplication factors (0·61, 0·41 and 0·41, respectively). In the above, all fitted parameters were highly significant and the coefficients associated with c3 (July 2005) and c4 (August 2005) were significantly different to c2 (May 2005), the parameter estimate for which was, itself, significantly different to the first measurement date (March 2005). That is, R20 (March 2005) > R20 (May 2005; shade and sun leaves = 0·44 and 0·74 µmol CO2 m−2 s−1, respectively) > R20 (July 2005; shade and sun leaves = 0·29 and 0·49 µmol CO2 m−2 s−1, respectively) = R20 (August 2005). Across all time periods the proportional difference between sun and shade-grown plants did not change significantly: R20 (sun) = 1·69 R20 (shade) and, as stated above, there was no evidence of E0 changing with time of year.

When compared at the prevailing average daily temperature during each measurement period (derived using the model fit of eqn 9), rates of leaf R were relatively constant in both sun-exposed and shade plants (Fig. 3), despite the large seasonal variation in temperatures (Fig. 1). We interpret this as a near full homeostasis of average daily leaf R being achieved across the year.

photosynthesis: fluorescence and gas exchange

When maximal photochemical efficiency of PSII (Fv/Fm) was measured predawn there were very low values in winter time, (Table 1; sun site: 0·38, shade site: 0·66). Nevertheless, Fv/Fm values had recovered by spring (Table 1; sun-exposed: 0·82, shade: 0·84) and did not change significantly throughout the summer, with the exception of a small but significant decrease in the Fv/Fm values in the sun plants between July and August.

Net photosynthesis (Pnet), measured at the prevailing ambient irradiance (sun-exposed site, ranging from 610 to 1490 µmol photons m−2 s−1 from winter to summer, respectively; shaded site: ranging from 120 to 290 µmol photons m−2 s−1 from winter to summer, respectively). Pnet was significantly higher in sun-exposed plants than in the shade plants (Fig. 2c,d; P < 0·001). Further, Pnet changed significantly between months (P < 0·001), and a significant interaction term was detected between sampling time and growth irradiance (P = 0·005) indicating that the patterns of change through time were not identical at the sun and shade sites. Taken together, these results demonstrate that: (i) there were differences in Pnet between sun-exposed and shade-grown Q. ilex; and (ii) that the magnitude of these differences changed through the year, both in an absolute and proportional sense.

In this water-limited ecosystem (see Table 1), it was thought that Pnet may be strongly influenced by stomatal conductance (gs). To determine whether this was the case, linear regressions were carried out between Pnet and gs for both sun-exposed and shade-grown plants (see Supplementary Fig. S1) indicating a significant relationships between Pnet and gs at the sun (P < 0·001) and shade (P = 0·020) sites. The F-ratio method revealed that the two regressions differed significantly (F2,36 = 17·933, P < 0·01).

ratio of leaf r to pnet

To assess whether the balance between CO2 release by night-time leaf R and day-time Pnet varied through the year and/or differed between the sites, R/Pnet ratios were calculated using average rates of R measured at night, and average rates of Pnet during the day-time (Fig. 4). As R/Pnet values were calculated from average rates, no statistical analysis was possible but the following patterns were observed. R/Pnet at both sites increased substantially in August 2005, but the increase in R relative to Pnet was particularly high in sun-exposed plants. There was also an increase in R/Pnet during March, again especially at the sun-exposed site. In August 2004 and July 2005, shade plants exhibited higher R/Pnet values than sun-exposed plants, due largely to the large reductions in Pnet that occurred during these months.

Figure 4.

Ratio of respiration in darkness (R) to net photosynthesis (Pnet) of fully-expanded mature Quercus ilex leaves plotted against month, both for full sun (open symbols) and deep shade (closed symbols) grown plants. R values are the average rates of leaf R exhibited by leaves during the night-time, whereas Pnet was taken as the average of all net photosynthetic rates exhibited during the day-time (see Fig. 2).

leaf biomass allocation and carbohydrate concentrations

Table 4 lists the SLA, leaf thickness (as estimated from the ratio of leaf fresh mass to leaf area), leaf DMC over the 12-month period; from August 2004 through August 2005. Seasonal variation in SLA values was observed by sun-leaves only (P = 0·037). To explore what underlying factors were responsible for the differences in SLA values observed, leaf DMC and leaf ‘thickness’ (as estimated from the ratio of leaf fresh mass to leaf area; Dijkstra 1989) values were calculated. The observed differences in SLA between sun and shade leaves were not due to differences in DMC (Table 4; P > 0·05). Rather, the inter-site differences in SLA reflected differences in leaf fresh mass to leaf area, with sun-grown leaves being ‘thicker’ than their shade-grown counterparts (Table 4; P < 0·001). Moreover, the seasonal variations in sun-grown SLA values reflected month-to-month variations in leaf thickness only (P < 0·010).

Table 4.  Effect of growth irradiance (sun- and shade-exposed) on leaf mass–area relationships and carbohydrate concentrations of Quercus ilex
Growth irradianceLeaf traitSampling month
August 2004January 2005March 2005May 2005July 2005August 2005
  1. Specific leaf area (SLA, ratio of leaf area to leaf dry mass), leaf thickness [as estimated from the ratio of leaf fresh mass to leaf area (Dijkstra 1989)], leaf dry mass content (DMC, the ratio of leaf dry mass to leaf fresh mass), concentrations of total soluble sugars (i.e. fructose + sucrose + glucose) and starch are shown. Total soluble sugar and starch concentrations were determined using pooled samples of four replicates harvested at each site and time. Biomass allocation parameters are the mean of four replicates (± SE).

Full sunSLA (m2 kg−1)5·2 ± 0·43·7 ± 0·13·4 ± 0·23·5 ± 0·14·3 ± 0·34·0 ± 0·2
Leaf thickness (g FM m−2)350 ± 20503 ± 36492 ± 18477 ± 18361 ± 39403 ± 18
Leaf DMC (%)55·9 ± 0·953·9 ± 2·159·5 ± 0·759·6 ± 0·566·4 ± 5·362·0 ± 1·0
[Sugar] (mg g−1)33·345·746·637·832·733·8
[Starch] (mg g−1)00014·600
ShadeSLA (m2 kg−1)6·4 ± 0·25·9 ± 0·35·7 ± 0·15·7 ± 0·15·9 ± 0·05·8 ± 0·2
Leaf thickness (g FM m−2)272 ± 8299 ± 14299 ± 05314 ± 16270 ± 04267 ± 08
Leaf DMC (%)57·3 ± 0·657·3 ± 1·459·1 ± 1·356·3 ± 3·563·2 ± 1·164·6 ± 0·8
[Sugar] (mg g−1)28·645·549·532·030·831·9
[Starch] (mg g−1)000000

Total soluble sugar concentrations (i.e. sucrose + glucose + fructose) and starch concentrations are also shown in Table 4 (the small sample size necessitated pooling of sugar per starch samples, with the result that statistical analysis was not possible). Nevertheless, the results suggest that sugar concentrations were higher in March 2005 (the coldest month) than the other months sampled (Table 4). No starch was detected in shade-grown leaves at any time of the year. Measurable starch was observed in sun leaves only in May 2005.

Discussion

daily and seasonal variations in leaf r

At the outset of our study, we hypothesized that the temperature sensitivity of leaf R would differ among plants growing under contrasting light regimes; this hypothesis was based on the observation that Q10 values can vary with substrate availability depending on the extent to which respiratory flux is limited by ATP turnover and/or enzymatic capacity (Atkin & Tjoelker 2003). To test this hypothesis, we have used two approaches to assess whether site-dependent variations in temperature sensitivity needed to be taken into account when modelling seasonal variations in leaf R. When using eqn 8, it was necessary to allow for seasonal variations in the Q10, reflecting the fact that Q10 values were higher in winter than summer, though this was not necessary when modelling the temperature sensitivity of respiration using the Arrhenius function. Nevertheless, neither modelling approach required a difference in the temperature sensitivity of leaf R according to light regime of growth. In Hartley et al. (2006) no correlation between the Q10 of leaf R and atmospheric [CO2]-induced changes in photosynthesis was observed; however, soluble sugar concentrations remained high in all treatments. Similarly, in Zaragoza-Castells et al. (2007) we found no correlation between the Q10 of leaf R and shade-induced changes in photosynthesis of controlled-environment grown Q. ilex, possibly because soluble sugar concentrations were insensitive to transient shade. Sugar concentrations were also similar in sun- and shade-grown plants at the dry-land field site used in our current study (Table 4), which may have been the reason why little difference was observed in the temperature sensitivities of leaf R for sun and shade plants. While such results do not preclude the possibility that the temperature sensitivity of leaf R will decline under conditions that severely reduce substrate supply, evidence to date suggests that climate-dependent changes in photosynthesis are unlikely to result in concomitant changes in the temperature sensitivity of leaf R under field conditions. Nevertheless, further controlled environment experiments that explicitly test the effect of changes in leaf carbohydrate levels on the temperature sensitivity of leaf R are required.

From the modelling exercise (eqn 9), we found that E0 remained constant through the year, despite the Q10 of leaf R being higher in winter than summer. This apparent contradiction can be explained by the fact that constant E0 values are inevitably associated with a Q10 declining with increasing temperature. This can be shown by the following (see Supplementary Material for a full derivation):

image( eqn 10)

Importantly, however, a constant E0 cannot account for scenarios where the temperature dependence of the Q10 is substantial, such as reported for Eucalyptus pauciflora growing in SE Australia (Atkin et al. 2000), or for diverse plant taxa and biomes sampled around the globe (Tjoelker et al. 2001; Atkin & Tjoelker 2003). Rather, changes in E0 are needed to account for the larger scale temperature-dependent changes in Q10, as shown in Fig. 5; in this example, a constant E0 value explains only half of the actual global temperature dependence of Q10 as derived by Tjoelker et al. (2001). Thus, while our analysis demonstrates that E0 of shade- and sun-exposed Q. ilex plants at our field site was constant across the year, considerable variability in E0 must occur in order to account for the large temperature-dependent changes in Q10 reported by Tjoelker et al. (2001). Although it is known that the temperature dependence of R is linked to shifts in the control exerted by maximum enzyme activity at low temperature and substrate limitations at high temperature (Atkin & Tjoelker 2003), little is known about how such factors vary among sites in a way that would explain large variations in E0. Establishing a processed-based understanding for why E0 varies within and among sites is necessary, if we are to more accurately incorporate variations in the temperature dependence of leaf R into large scale models (which at present assume either a constant Q10 or E0).

Figure 5.

Temperature dependence of the Q10 of leaf R as dependent on the assumed activation energy (E0) value (calculated using eqn 10). The temperature dependence of the Q10 reported as based on data set reported in Tjoelker et al. (2001) is shown for comparison.

In contrast to the constant E0 at our site, basal rates of R at a reference temperature of 20 °C (i.e. R20) varied substantially, decreasing from winter to summer, and being consistently higher in the sun-exposed plants (compared to their shade counterparts). By incorporating additive non-interacting functions of time of year and light regime (in eqn 9) that allowed for variations in R20, we were able to model seasonal and daily variations in leaf R in both sun- and shade-exposed plants. Previous studies have also reported shade-induced reductions in the basal rate of R, with reductions in respiratory capacity being largely responsible (Noguchi & Terashima 1997; Noguchi et al. 2005).

Importantly, both sun- and shade-grown plants exhibited seasonal changes in temperature response curves (Fig. 3) and predicted rates of R20, suggesting that acclimation occurs independently of the growth irradiance and that acclimation is underpinned by variations in the basal rate of R. While the season shifts in rates of leaf R (Fig. 3) could reflect respiratory responses to changes in factors other than temperature [e.g. decreases in water availability and/or foliar N content as the summer progresses (Xu & Griffin 2006)], we have recently observed similar shifts in the daily temperature response of leaf R in Q. ilex under controlled environment conditions, both under high- and low-light (Zaragoza-Castells et al. 2007) which suggests that seasonal changes in daily average temperature may have been paramount. Moreover, exposure to severe water stress typically increased rates of leaf R (Flexas et al. 2006; Wright et al. 2006), as opposed to the decreased leaf R that was observed in our study when soil moisture was lowest (Fig. 3, Table 1). Moreover, Bruhn et al. (2007) reported that thermal acclimation in an evergreen tree species (in response to artificial heating in the field) was underpinned by variations in the basal rate of R. Changes in the basal rate of leaf R likely reflect changes in maximal rates of mitochondrial O2 uptake per unit protein and alterations in the density of mitochondria per unit cell volume (Armstrong et al. 2006a).

photosynthesis in plants grown under full sun and shade

Overall, Pnet was higher at the sun site than at the shade site, especially in May 2005. However, in March and August 2005 (with coolest and hottest temperatures, respectively) the rates dropped sharply at both sites (Fig. 2). This seasonal response is similar to the results of several studies of species growing in Mediterranean ecosystems (i.e. Pereira et al. 1986; Faria et al. 1998; Peñuelas et al. 1998; Gratani et al. 2000; Llorens, Peñuelas & Filella 2003; Corcuera et al. 2005). The seasonal changes in Pnet also differed between sun and shade sites with the magnitude of the changes in the rate of photosynthesis between months being greater at the sun site (Fig. 2). Rates of Pnet during the summer drought were lower at the sun site than in the shade, suggesting that plants were experiencing greater environmental stress. However, plants at the shade site were experiencing a greater water stress during this period than the sun plants (Table 1). Despite this, shaded plants were capable of re-adjusting their rates of Pnet and R, which agrees with recent studies assessing the performance of drought-affected Mediterranean plants under shaded conditions (Sack 2004; Sánchez-Gómez, Valladares & Zavala 2006).

In both sun and shade plants, stomatal conductance (gs) values were low throughout the year (see Supplementary Fig. S1) and limitations in CO2 availability may have limited photosynthesis to some extent, particularly when soil moisture was lowest and air temperatures highest (Table 1, Fig. 1). The extent to which CO2 availability limited Pnet was, however, likely to have differed between sun and shade plants, particularly under conditions of increased gs. At the low range of observed gs values, rates of Pnet were similar at both sun- and shade-exposed sites but at higher gs values, Pnet ranges diverged, suggesting that Pnet becomes limited by factors other than CO2 availability at high gs values at the shade site.

balance between leaf r and pnet

Overall, it is expected that there will be a coupling between Pnet and leaf R given that photosynthesis provides the substrate supply for respiration, and respiration the energy required for the maintenance of photosynthesis (Hoefnagel, Atkin & Wiskich 1998). Moreover, respiration provides the energy for exporting assimilates (Farrar 1985) and oxidizes excess of redox equivalents from photosynthesis (i.e. Saradadevi & Raghavendra 1992). However, the temperature sensitivity of these two processes differs and temperature may, therefore, have an impact on the R/Pnet ratio. Although several studies have reported that R/Pnet remains more or less constant in plants experiencing contrasting growth temperatures (Gifford 2003; Loveys et al. 2003), less is known about the extent to which R/Pnet is invariant under the extreme conditions experienced by plants growing in dry-land, low NPP ecosystems such as was the case here. Indeed, our data show that estimates of daily R/Pnet (calculated using the daily average rates of night-time R and day-time Pnet) were not constant throughout the year (Fig. 4), with average night-time leaf R greatly exceeding average day-time Pnet during periods of severe water limitation and high day-time temperatures in summer, and during periods of freezing/photoinhibitory stress in winter (Table 1). The low predawn Fv/Fm values in the winter (Table 1) have been described as indicative of changes in chlorophyll pigments and associated with the presence of xanthophyll pigments in PSII, resulting in a down-regulation of PSII (Gilmore & Ball 2000) and lower photosynthetic rates (Martínez-Ferri et al. 2004; Corcuera et al. 2005). In both cases, the imbalance between night-time R and day-time Pnet was greatest in sun-grown plants, suggesting that leaves maintain a more positive carbon balance at shaded sites at hot, dry sites such as that used in our study. Balanced against this beneficial effect of shade is the fact that maximal rates of R and Pnet are considerably lower in the shade (Fig. 2), thus potentially reducing growth rates under low irradiance.

Although R/Pnet did increase markedly in summer (particularly under full sun; Fig. 4), the increase in this ratio was considerably lower than it would have been if leaf R had not exhibited large seasonal changes in basal rates (Fig. 3), which we suggest was the result of the thermal acclimation to the high average day time temperatures in summer (see above). If one assumes that the winter (March 2005) temperature response curve of leaf R was maintained all year round (i.e. assuming no acclimation of leaf R to seasonal shifts in temperature), then estimated R/Pnet values in summer would have been near double those shown in Fig. 4. Acclimation of leaf R thus contributes substantially to help both sun- and shade-grown Q. ilex leaves maintain a relatively positive carbon balance than would not be possible in the absence of thermal acclimation.

Conclusions

We found no difference in the temperature sensitivity of leaf R between sun- and shade-exposed grown Q. ilex growing under field conditions in a dry-land, Mediterranean ecosystem. Thus, while it is theoretically possible for the temperature sensitivity to vary as a function of substrate supply (Atkin & Tjoelker 2003), to date no study has found systematic variations in temperature sensitivity in response to light- or [CO2]-mediated changes in photosynthesis in leaves where substrate concentrations remain relatively high irrespective of the growth treatment (Hartley et al. 2006; Zaragoza-Castells et al. 2007). Our results add to previous studies showing that the Q10 of leaf R is greatest at low temperatures (Tjoelker et al. 2001); importantly, however, we show moderate temperature-dependent variations in the Q10 can be accounted for via application single E0 value, with the result that a Arrhenius-based model successfully accounts for daily and seasonal variations in R. Crucial to the formulation of such a model is accounting for large seasonal shifts in the basal rate of leaf R (that are consistent with thermal acclimation) that occur in both sun- and shade-grown plants, and ensuring that the irradiance-mediated differences in the basal rate of R are also taken into account.

Our finding that Q. ilex exhibits large seasonal adjustments in the basal rate of R over the year and in response to shading has important implications for our understanding of the underlying factors controlling net carbon uptake at this dry-land, low NPP ecosystem. Failure to adjust the basal rate of leaf R would result in respiratory CO2 release being excessively high in summer, with concomitant increases in the ratio of leaf R to photosynthesis and reductions in NPP. By downward adjusting basal rates of leaf R during hot periods, both sun- and shade-exposed grown plants are more likely to reduce daily R/Pnet ratios and maintain a positive overall annual carbon balance. Acclimation of leaf R to seasonal changes in the environment may thus play a crucial role in determining the viability of tree growth in dry-land ecosystems.

Acknowledgements

We thank to David Sherlock for technical assistance, to Rafael Ruiz López de la Coba and the staff of the Alto Tajo Natural Park and Delegación Provincial de Medio Ambiente (Junta de Castilla La Mancha) for their support to this research and to Beatriz Alonso, Oscar Godoy, Angelika Portsmuth, Jose Alberto Ramírez, Virginia Sanz, and many students who helped us with the logistics of each field campaign. We would like to thank the two anonymous referees for their useful suggestions. This work was funded by the Natural Environment Research Council (NERC) in the UK (NER/A/S/2001/01186; OKA), a University of York Innovation & Research Primer Grant to OKA, and a University of York PhD studentship to JZC. Financial support was also provided by the Spanish Ministry of Education and Science (grants RASINV, CGL2004-04884-C02-02/BOS, and PLASTOFOR, AGL2004-00536/FOR) to FV and a scholarship from the Spanish MECD (FPU fellowship, AP 2001-0193) to DSG.

Ancillary