Three parameters comprehensively describe the temperature response of respiratory oxygen reduction

Authors


M. A. Adams. Fax: +61 2 9385 2120; e-mail: Mark.Adams@unsw.edu.au

ABSTRACT

Using an exponential model that relies on Arrhenius kinetics, we explored Type I, Type II and dynamic (e.g. declining Q10 with increasing temperature) responses of respiration to temperature. Our Arrhenius model provides three parameters: RREF (the base of the exponential model, nmol g−1 s−1), E0 (the overall activation energy of oxygen reduction that dominates its temperature sensitivity, kJ mol−1) and δ (that describes dynamic responses of E0 to measurement temperature, 103 K2). Two parameters, E0 and δ, are tightly linked. Increases in overall activation energy at a reference temperature were inversely related to changes in δ. At an E0 of ca. 45 kJ mol−1, δ approached zero, and respiratory temperature response was strictly Arrhenius-like. Physiologically, these observations suggest that as contributions of AOX to combined oxygen reduction increase, E0(REF) decreases because of different temperature sensitivities for Vmax, and δ increases because of different temperature sensitivities for K1/2 of AOX and COX. The balance between COX and AOX activity helps regulate plant metabolism by adjusting the demand for ATP to that for reducing power and carbon skeleton intermediates. Our approach enables determination of respiratory capacity in vivo and opens a path to development of process-based models of plant respiration.

Abbreviations
AOX

alternative oxidase

COX

cytochrome oxidase

Ea

activation energy for a distinct chemical reaction

 

The activation energy for a metabolic reaction in living cells depends on the catalytic properties of the respective enzyme (either AOX or COX in the present study)

Eo(RO2)

overall activation energy for mitochondrial oxygen reduction (as influenced by both AOX and COX activities)

K1/2

temperature-dependent half-saturation ‘constant’ of one distinct enzyme (either AOX or COX in the present study)

SHAM

salicylhydroxamic acid

Vmax

temperature-dependent maximum velocity of a distinct, enzyme-catalysed metabolic reaction

INTRODUCTION

Plant respiration is strongly temperature dependent, and often shows exponential behaviour in response to temperature. On a different timescale, the mean global surface temperature is expected to rise within the next couple of decades. At present, it is not clear if this temperature rise will increase rates of respiration that cannot be compensated for by photosynthesis (Woodwell & Mackenzie 1995; IPCC 2001). On the other hand, plant respiration is flexible, and plants adjust their metabolism to the thermal environment. Better knowledge of the potential of plants to acclimate to changing environmental conditions is thus of broad significance (Atkin et al. 2005). In particular, it is important that we obtain a better mechanistic understanding of the temperature response of respiration, as this should enhance our capacity to predict the effects of temperature and other environmental variables on gas exchange processes in plant growth models.

Current theory identifies three phenomena that help describe the temperature response of respiration. Firstly, the shape of the exponential function flexibly adjusts to altered environmental conditions. This mechanism is sometimes referred to as ‘Type I’ acclimation, and is a seemingly swift response to changes in environmental conditions (Atkin & Tjoelker 2003; Atkin et al. 2005). In the longer term, plants can alter their respiratory capacity (Type II acclimation), leading to faster rates of respiration at low measurement temperatures. For example, plant development at lower temperature can increase mitochondrial density (Klikoff 1966; Armstrong et al. 2006) and, generally, concentrations of N in foliage – these then contribute to faster rates of basal respiration and to the frequently observed homeostasis in respiration with growth temperatures (Larigauderie & Körner 1995). It is to be expected that plant nutritional status acts in concert with seasonal and developmental factors in influencing basal respiration (e.g. Gough, Seiler & Maier 2004; Reich et al. 2005; Turnbull et al. 2005).

These two kinds of acclimation are thought to act independently from one another, whereby ‘Type II’ acclimation affects the base of the exponential function describing plant respiration, and ‘Type I’ acclimation is reflected in alterations of the exponent (Atkin et al. 2005).

A possible third phenomenon concerns the dynamic response of respiration to temperature. Reviews of the literature suggest that Q10 (the proportional change of respiration for every 10 °C increase in temperature) decreases as the measurement temperature increases (Tjoelker, Oleksyn & Reich 2001). Subsequent mechanistic interpretations for this phenomenon rely essentially on two components (Atkin & Tjoelker 2003). At low temperatures, respiratory oxygen reduction is thought to be restricted by the activity of the respiratory enzymes involved (Vmax), whereas flux at higher temperatures is increasingly limited by the availability of adenylates (ADP) or substrate (Atkin & Tjoelker 2003). At present, it is not clear whether such a proposed mechanism is reversible and dynamic, or if it will lead to sustained changes of respiration, i.e. an acclimation of Type I. Again, the two latter plant responses to changing thermal conditions appear to be relevant at different timescales. But are they separate phenomena, or does ‘Type I’ acclimation entail the dynamic response that becomes irreversible after prolonged exposure to a different temperature regime? Current experimental and theoretical approaches often make it difficult to answer this question unambiguously and gain insight into the mechanisms causing ‘Type I’ acclimation. This can be traced, at least partly, to the description of the two latter phenomena by the same means, namely a Q10 approach.

We thus need experimental techniques with adequate sensitivity to detect more subtle deviations from strictly exponential behaviour, i.e. the dynamic response of respiration to temperature.

Calorimetry is a potent tool for investigating the temperature response of oxygen reduction in vivo. Measurements can be made relatively quickly and with considerable precision. Heat rate measurements can be transformed into rates of O2 reduction, because the ratio of heat produced to oxygen consumed is nearly constant regardless of the substrate for respiration (Ordentlich, Linzer & Raskin 1991; Criddle & Hansen 1999; Kemp 2000). The detection limit of the instruments is about 5 µW, or 10–15 pmol O2 g−1 s−1. Despite its precision, calorimetry is less used than alternatives such as measuring rates of oxygen consumption via oxygen electrodes (e.g. Armstrong et al. 2006) or fuel cells (e.g. Davey et al. 2004).

We employed calorimetry in a companion study, where we also introduced a modified Arrhenius equation to describe the temperature response of respiration (Kruse, Hopman & Adams 2008),

image(1)

where RREF is the respiration rate at a low reference temperature and Eo (kJ mol−1) is the overall activation energy for respiratory metabolism that defines the shape of the exponential function (R is the gas constant, see also Kruse et al. 2008). In comparison with Q10 approaches, Arrhenius approaches have the clear advantage, that the shape of the exponential function (as defined by Eo) can be directly related to metabolic processes. This is particularly true for the processes associated with mitochondrial oxygen reduction. In higher plants, most oxygen reduction is catalysed by two distinct enzymes: (1) COX, which is coupled to proton transfer across the inner membrane and by generating a proton motive force (PMF), to ATP production or (2) AOX, which does not contribute to the build-up of PMF and thereby, to ATP production (Buchanan, Gruissem & Jones 2000, pp. 696–706).

We demonstrated in a companion study (Kruse et al. 2008) that partial inhibition of either COX (with cyanide) or AOX (with SHAM) did affect the average activation energy for overall oxygen reduction [Eo(RO2)], as calculated by the linearization of Eqn 1. Inhibition of COX significantly decreased, and inhibition of AOX slightly increased Eo(RO2) (compare with Kruse et al. 2008). The temperature response of respiration can, therefore, be used as an indicator of the contributions of COX versus AOX to overall oxygen reduction, as affected by developmental and environmental influences. This approach is an indirect method to estimate activities of the respective enzyme in vivo. However, it became apparent that inhibition of COX or AOX was followed by the partial re-adjustment of respiration rates (as compared with the control treatment) with increasing measurement temperature and duration of the experiment, or both (Kruse et al. 2008). That is, we observed a strong dynamic response of respiration to alterations of the measurement temperature.

The objectives of the current study were, first, to further develop an Arrhenius approach, as given by Eqn 1, in order to include a parameter that describes the dynamic response of respiration as a possible separate and third ‘driver’ of the temperature response of respiration. Secondly, we sought to induce and investigate a variety of different temperature responses (see Atkin et al. 2005, and references therein). In the present study, we explored the temperature dependency of oxygen reduction as affected by a number of environmental and developmental factors. We studied needles from Pinus radiata, differing in age and taken from different canopy positions during the growth season in spring, at the end of the summer and during midwinter. We also included plantation sites that have been treated with N-fertilizer. Thirdly, we expected to assess whether ‘Type I’ acclimation and the dynamic temperature response are independent phenomena. We also sought to evaluate if alterations of the base of the exponential function (i.e. RREF in Eqn 1) – as affected by environmental and developmental factors – is a useful indicator of ‘Type II’ acclimation. Taken together, we focused on an improved mechanistic understanding of the temperature response of respiration. In doing so, we aimed to derive parameters that could reliably describe the developmental and environmental control of respiration, and that could be used for modelling purposes (Gifford 2003).

MATERIALS AND METHODS

Experimental design

The experiments were conducted in 22-year-old pine plantations (planted 1982), located 10 km east of Mt. Gambier, South Australia. The climate of the region is Mediterranean, with cool wet winters and warm dry summers. Average annual rainfall is 720 mm of which about 70% falls in winter. Monthly mean maximum and minimum air temperature vary from 25.2 and 10.8 °C in January (midsummer) to 13.0 and 4.9 °C in July (midwinter, Australian Bureau of meteorology). The soil can best be described as a deep, podsolized sand, and the plantation was established on a flat terrain at a density of ~1200 stems ha−1. Research plots were established in 1995 as a 4 × 4 Latin square design with different combinations of fertilizer. In the current study, we chose, in addition to control plots, four plots treated with 300 kg N ha−1 applied as (NH4)2SO4. Both control and fertilizer-treated plots were thinned to the same basal area prior to fertilizer application in 2002. We selected four of the ca. 40 trees per plot (i.e. 10%) in a semi-random approach: The first tree between numbers 1 and 10 was selected at random, and subsequently, every 10th tree counting from the first choice was taken, thus ensuring a sufficient coverage of the plot area. During the 8- to 10-day-long field trips, control and fertilizer-treated plots were studied in an alternating fashion.

Sampling was scheduled and conducted at times representing the most contrasting environmental conditions, i.e. in the midst of the growing season when temperatures and water availability were optimal for growth, at the end of summer with more frequent temperature and drought stress, and towards the end of winter when temperatures were suboptimal (Fig. 1). In order to study the effect of the canopy position on the temperature response of needle respiration, canopies were stratified into three different layers of equal depth. Each morning during sampling, three branches per tree (one randomly selected from each layer), from each of the selected four trees, were shot down. Branches long enough to bear three needle age classes (a fourth needle age class is usually shed during the growing season until the end of summer, compare Kruse & Adams (2008) and with an unscathed apical bud were re-cut under water and carried to the calorimetry room that was in close proximity (200 m) to the research plots.

Figure 1.

Average weekly temperatures and relative humidity (a) and average monthly rainfalls compared to the long-term average (b) during the year of the study. Temperature, humidity and rainfall were recorded at a weather station in close proximity (200 m) to the field site. Sampling trips within each season lasted 8–10 d.

The age effect was inferred from the acclimation response of the youngest [current year (c) needles] to the oldest needles (c + 2 years old). Three fascicles, ranging from the lower to the upper end of the respective needle age class were collected and mixed with the needles of the same age classes from the remaining three branches of the same canopy layer. In summary, measurements were made using a mix of four independent trees per plot, so that at the end of each field sampling four replicates per treatment, canopy layer and age class were available. This approach was necessary to restrict sampling and measurement to 8 d (without extension into the next season), and to ensure maintenance of the necessary statistical power.

Respiration measurements

Respiration measurements were conducted with two multi-cell, differential-scanning calorimeters (CSC 4100, MC-DSC, Calorimetry Sciences Corporation, Provo, UT, USA) that allow heat rate measurements over a range of different temperatures, operating in the isothermal mode. The calorimeters are sensitive to alterations of the surrounding air temperature and were calibrated prior to measurements. Baselines were linearly related to room temperature in the laboratory (data not shown) and at the field site. During field studies, room temperatures were recorded at the same time as heat rates, in order to derive the respiration parameters described below. Without calibration, room temperature fluctuations would have caused an error of 1–4%.

Before measurement, young and old needles were cut into ca. 1-cm-long pieces and left for 20 min to allow respiratory wound responses to dissipate. Preliminary experiments indicated that wound responses dissipated within 10 min of cutting (data not shown). From the pool of cut pieces, ca. 15–25 (depending on needle thickness) were selected at random and sealed (O-ring seal, small amount of silicon grease) in ampoules with a volume of 1 cm3. Ampoules were placed into measurement cells, pre-cooled to 13 °C, well above the dew point. The samples were left for 10 min at 13 °C. The first measurement was conducted at 15 °C, followed by measurements at 5 °C intervals. Stabilization of the signal at each new temperature took about 40 min, after which the heat rate and room temperature were recorded. After the 25 °C measurement, ampoules were opened briefly, flushed with air and returned to the wells. Preliminary experiments showed that provided oxygen concentrations were above 5%; heat rates were unaffected. After the final measurement (35 °C), the needles were microwaved for 20 s, dried at 70 °C for 48 h and, subsequently, stored at room temperature until further analysis.

Determination of nitrogen

After weighing, dried samples were ground to a fine powder with a ball mill. Nitrogen and carbon concentrations were determined by Dumas combustion. Aliquots of 50 mg of the finely ground foliage samples were combusted to N2 and CO2 in the presence of O2, and quantified by means of thermoconductivity (LECO CHN2000, Leco Corporation, St Joseph, MI, USA).

Calculation of respiration parameters

Heat rates of needles were calculated from the difference of the heat rate at any given measurement temperature and the respective baseline, after correcting the baseline for prevailing temperature. Heat rates were transformed into rates of O2 consumption according to Thornton's rule that the ratio of heat produced to oxygen consumed is nearly constant regardless of the substrate for respiration and is about −455 µJ nmol−1 O2 (Hansen et al. 1994).

Commonly, the temperature response of respiration and the dynamic adjustment of RO2 to increasing measurement temperatures are described by the dimensionless Q10, or the proportional change in R per 10 °C increase in temperature (Schulze, Beck & Müller-Hohenstein 2002),

image(2)

where T2 and T1 are temperatures, and R2 and R1 are respiration rates of the upper and lower boundaries of the temperature interval considered. This approach cannot distinguish between two putatively different kinds of acclimation (‘Type I’ acclimation and the dynamic response to temperature). Furthermore, Q10 depends on the temperature interval chosen for its calculation (compare Fig. 2). We therefore decided to elaborate on the modified Arrhenius equation (see Eqn 1 in the Introduction) and to include a parameter that describes the dynamic response of RO2 to temperature. The temperature response of respiration can be approximated by the Arrhenius model (Kruse & Adams 2008), whereby the overall activation energy for RO2 (Eo) defines the shape of the exponential function (compare Fig. 3a). This parameter can be calculated from the linearization of the Arrhenius equation (Criddle et al. 1994; Taylor et al. 1998). For this purpose, values of ln RO2 are either plotted against the reciprocal of the temperature (Fig. 3b), or against the temperature term described in Fig. 3c. The slopes of the resulting functions are the same, as is the calculated average activation energy [Eo(average) = 90 kJ mol−1 in our example depicted in Fig. 3]. Although linearly related with little deviation (R2 > 0.97), there was a clear deviation with increasing temperature from strict Arrhenius kinetics. The data shown in Fig. 3c are in fact better fitted by a second-order polynomial (Fig. 3d), and the factor δ (in units of 103 K2) describes the dynamic response of Eo to temperature (i.e. the deviation from strict linear Arrhenius kinetics). Hence, Eqn 1 was further modified,

Figure 2.

Q10 for different temperature intervals, calculated according to Eqn 2. In this example, we assumed that respiration can be described by the Arrhenius equation (i.e. Eqn 1 in the Introduction), and that RREF is 1.4 nmol g−1 s−1 and the overall activation energy is 50 kJ mol−1. These values are close to those calculated for the temperature response of oxygen reduction of the current year needles in a previous study (Eo = 49.3 kJ mol−1, experiment 1b in Kruse et al. (2008). Q10, proportional change of respiration for every 10 °C increase in temperature; RO2, oxygen reduction; T, absolute temperature; RREF, the respiration rate at a low reference temperature; Eo, overall activation energy.

Figure 3.

Determination of the three parameters that comprehensively describe the temperature response of respiration. The exponential temperature response (a) is usually approximated by the average activation energy of respiration for the entire temperature range under consideration (b,c). This approach does not take account of the dynamic response Eo to temperature, i.e. deviation from strictly linear Arrhenius kinetics. The linearized data depicted in (c) can be better described by a second-order polynomial (d), where ln RO2(REF) is the natural logarithm of respiration rates at the reference temperature, Eo(REF) is the overall activation energy at the reference temperature and the factor δ describes the dynamic response of Eo to temperature, starting from the reference temperature. In the current study, the three parameters were related to the reference temperature of 288 K (15 °C). RO2, oxygen reduction; Eo, overall activation energy; R, gas constant; T, absolute temperature; RREF, the respiration rate at a low reference temperature; δ, defines the temperature-dependent change in Eo starting from the reference temperature.

image(3)

where RREF is the respiration rate at the reference temperature (nmol g−1 s−1), Eo is the overall activation energy at the reference temperature (kJ mol−1), R is the gas constant (8.314 J mol−1 K−1), δ(103 K2) defines the temperature-dependent change in Eo starting from the reference temperature and T is the absolute temperature (K). In our example depicted in Fig. 3d, ln RO2(15 °C) = −0.189 nmol g−1 s−1, Eo(15 °C) = 132 kJ mol−1 and δ(15 °C) = −23.8 103 K2.

This model comprehensively describes the observations obtained by calorimetry (average R2 of polynomial fits: 0.9985, n = 144) and yields three parameters with physiological meaning.

Statistical analysis

Statistical analysis was performed with the results of four independent replicates per season, canopy position, age and treatment. Note that each replicate already represented foliage from four different trees. The main effects (season, canopy position, age, treatment) and their interactions were assessed by analysis of variance (anova) (Statistica, version 6.0, StatSoft Incorporated, Tulsa, OK, USA). Categorical predictor variables were coded according to sigma-restricted parameterization, and sums of squares were calculated according to Type VI. For the dependent variables N-content and needle length, anova included only three factors, owing to a loss of samples and needle length was included between seasons for young needles only. Statistical significance of main effects and their interactions is indicated by *, if P < 0.05, **, if P < 0.01 and ***, if P < 0.001.

RESULTS

Nitrogen content and needle length

For the young needles, fertilizer treatment exerted the largest effect on N-concentrations, followed by canopy position and season (Table 1). Fertilizer application increased the N-concentrations of the young needles (and of the old needles) in November and particularly in February, but not in August (significant interaction between treatment and season, not shown). In November 2003, the young needles in lower, middle and upper sections of the tree canopies had reached ca. 30, 50 and 60% of their final length. Final needle length was reached earlier in the upper canopy as – in contrast to needles from the lower canopy – elongation stopped by the end of February (Table 1). Fertilizer treatment modified needle length significantly depending on canopy position, with the longest needles of fertilized trees in the upper canopy (significant second-order interaction, not shown).

Table 1.  Nitrogen concentration and length of needles at different times of the year
Canopy positionNeedle ageNeedle nitrogen (%)
NovemberFebruaryAugust
Control+FertilizerControl+FertilizerControl+ Fertilizer
  1. Data shown are the average ± SD of four independent replicates. Each replicate was a mixture of foliage from four different trees.

  2. n.a., not available.

LowerOld0.80 ± 0.151.40 ± 0.11n.a.n.a.0.88 ± 0.030.95 ± 0.04
Young1.06 ± 0.061.19 ± 0.120.98 ± 0.081.53 ± 0.091.08 ± 0.030.98 ± 0.05
MiddleOld1.03 ± 0.061.55 ± 0.14n.a.n.a.1.03 ± 0.071.11 ± 0.04
Young1.24 ± 0.041.48 ± 0.051.07 ± 0.091.71 ± 0.061.19 ± 0.061.13 ± 0.04
UpperOld1.28 ± 0.091.67 ± 0.07n.a.n.a.1.19 ± 0.021.31 ± 0.08
 Young1.47 ± 0.101.71 ± 0.041.21 ± 0.141.87 ± 0.181.30 ± 0.031.24 ± 0.08
  Needle length (cm)
NovemberFebruaryAugust
LowerYoung3.2 ± 0.93.7 ± 1.110.8 ± 1.911.2 ± 1.811.7 ± 0.912.3 ± 1.7
MiddleYoung5.7 ± 0.76.9 ± 0.811.5 ± 1.213.4 ± 1.511.9 ± 1.113.3 ± 1.5
UpperYoung7.9 ± 0.39.6 ± 0.613.3 ± 0.715.6 ± 1.413.3 ± 1.315.5 ± 1.5

Effects of season, canopy position, needle age and fertilizer treatment on basal respiration rates (RREF) at a low reference temperature (15 °C)

Respiration rates at the reference temperature of 15 °C were highly variable and significantly dependent on age, season, canopy position, fertilizer treatment and their interactions (Tables 2 & 3, Fig. 4). Needle age had the largest impact on RREF– old needles averaged 1.82 nmol g−1 s−1 and young needles 5.33 nmol g−1 s−1. RREF also varied with season albeit dependent on needle age (significant second-order interaction, Table 3). From November to February, RREF declined by 40% in old and by 70% in young needles but then increased between February and August. The difference between the old and young needles declined from more than fourfold in November to less than twofold in August.

Table 2.  Respiration rates at the reference temperature of 15 °C (RREF), overall activation energy for oxygen reduction at the reference temperature [Eo(15 °C)] and the dynamic response of Eo to temperature (δ), starting from the reference temperature
SamplingCanopy positionNeedle ageRREF (15 °C, 288 K) (nmol g−1 s−1)Eo (15 °C, 288 K) (kJ mol−1)δ (15 °C, 288 K) (103 K2)
Control+FertilizerControl+FertilizerControl+Fertilizer
  1. The three parameters RREF, Eo and δ were calculated from a modified Arrhenius equation, as described in the Materials and Methods (Fig. 3d). Data shown are the average ± SD of four independent replicates. Each replicate was the average of foliage taken from four different trees.

NovemberLowerOld1.1 ± 0.21.3 ± 0.3122 ± 24104 ± 6−22.5 ± 7.9−13.9 ± 5.9
Young2.7 ± 0.63.6 ± 1.4123 ± 23117 ± 6−25.6 ± 8.2−22.0 ± 4.4
MiddleOld2.1 ± 0.13.0 ± 0.386 ± 680 ± 4−11.2 ± 1.7−10.1 ± 1.2
Young8.2 ± 2.310.4 ± 1.132 ± 429 ± 73.3 ± 2.24.2 ± 1.4
UpperOld2.2 ± 0.22.9 ± 1.095 ± 889 ± 7−14.4 ± 1.5−12.3 ± 3.2
Young12.4 ± 3.919.8 ± 4.919 ± 419 ± 65.2 ± 2.95.1 ± 2.0
FebruaryLowerOld0.8 ± 0.11.2 ± 0.7147 ± 16150 ± 17−28.6 ± 6.6−30.2 ± 3.7
Young3.1 ± 0.83.0 ± 0.547 ± 1240 ± 10−1.0 ± 3.92.9 ± 1.3
MiddleOld1.3 ± 0.11.4 ± 0.198 ± 697 ± 13−11.3 ± 3.1−13.1 ± 6.6
Young2.5 ± 0.82.6 ± 0.570 ± 1670 ± 8−7.3 ± 5.3−6.3 ± 2.9
UpperOld1.0 ± 0.21.5 ± 0.2151 ± 11127 ± 17−30.8 ± 2.6−23.5 ± 4.9
Young2.6 ± 0.92.9 ± 1.2105 ± 2292 ± 19−18.2 ± 8.9−12.7 ± 5.6
AugustLowerOld1.8 ± 0.21.6 ± 0.190 ± 790 ± 6−12.5 ± 3.1−12 ± 2.9
Young3.1 ± 0.63.1 ± 0.363 ± 363 ± 6−4.3 ± 1.5−4.6 ± 2.2
MiddleOld2.2 ± 0.62.1 ± 0.175 ± 280 ± 7−5.6 ± 1.7−8.9 ± 3.9
Young3.7 ± 0.73.4 ± 0.765 ± 568 ± 2−5.9 ± 2.2−7.2 ± 1.1
UpperOld2.7 ± 0.52.7 ± 0.582 ± 880 ± 9−9.9 ± 2.9−8.8 ± 3.2
Young4.8 ± 0.44.1 ± 0.564 ± 263 ± 3−6.9 ± 0.7−6.4 ± 2.1
Table 3.  Results of the multi-factorial analysis of variance (anova), showing the significance of the main effectors and their interactions for RREF, Eo, δ and IA
anovaRREF (nmol g–1 s–1) R2: 0.92Eo (kJ mol–1) R2: 0.89δ (K2) R2: 0.85IA (respiratory capacity, dimensionless) anova; R2: 0.65Eo (kJ mol–1) GLM; multiple R2: 0.96
FPFPFPFPFP
  • The dimensionless parameter IA is a measure for the respiratory capacity and was used as a continuous predictor variable to further explain the variation of Eo in a general linear model [(GLM) far right column of Table 3].

  • *

    P < 0.05;

  • **

    P < 0.05;

  • ***

    P < 0.05.

  • RREF, respiration rate at a low reference temperature; Eo, overall activation energy; δ, defines the temperature-dependent change in Eo starting from the reference temperature; n.s., not significant.

IA        49***
Season (S)118***74***40***23***59***
Canopy (C)58***59***44***12***59***
Age (A)279***402***192***94***159***
Treatment (T)10**5.1*4.8*1.2n.s.3.4n.s.
S × C38***63***41***1.1n.s.76***
S × A86***35***26***11.3***48***
C × A28***5.0**3.9**0.1n.s.6.9**
S × T11***1.7n.s.2.2n.s.3.5*0.1n.s.
C × T3.1*1.3n.s.2.1n.s.0.2n.s.2.0n.s.
A × T3.7*0.5n.s.0.1n.s.1.1n.s.2.6n.s.
S × C × A30***51***39***0.6n.s.64***
S × C × T2.5*1.5n.s.1.6n.s.0.3n.s.1.9n.s.
S × A × T6.2*0.3n.s.0.9n.s.0.3n.s.1.1n.s.
C × A × T1.8n.s.0.2n.s.0.4n.s.0.5n.s.0.1n.s.
S × C × A × T2.6*0.4n.s.0.8n.s.0.4n.s.0.6n.s.
Figure 4.

Factors of change (F) of basal respiration rates (RREF) and overall activation energy (Eo) in relation to a reference. The averages of RREF and Eo of the old needles from the lower canopy of the control trees in November served as the reference (REF). The deviation from the reference was calculated according to x(RREF;Eo) = REF + F × REF, where x is any given average of RREF and Eo shown in Table 3. Thus, inline imagefor x > REF, and F < 0 for x < REF. Each group of bars shows the factors from the lower to the upper canopy. Grey/black, control treatment; yellow/red, fertilized treatment; open patterns, F(Eo); filled patterns F(RREF).

Canopy position exerted less control on RREF (Tables 2 & 3). There was a significant interaction with season (Table 3). For example, in November, RREF increased threefold from the lower to the middle canopy, and twofold again from the middle to the upper canopy. In the lower and middle sections of the canopy, RREF of the young needles was 250% greater than that of the old needles, and there was a fourfold difference in the upper canopy. Effects of fertilizer on RREF were small compared with their effects on needle N-concentrations. Overall, fertilizer application increased RREF by 20%, but differences between treatments were mainly restricted to spring. The significant interaction between season and fertilizer treatment was further modified by needle age (Table 3). Differences between fertilizer-treated and control needles were clear in November but non-significant in other seasons (compare Fig. 4).

Significant interaction terms (Table 3) hint at considerable flexibility in the activation state of the respiratory apparatus. For example, we observed a strong correlation between RREF and the N-concentration of young needles during spring in November, but not at other times of the year (not shown, but compare Table 2 with 1).

Effects on overall activation energy (Eo) for RO2

Changes in RREF(15 °C) generally led to changes in Eo(15 °C), but in the opposite direction (Fig. 4). We observed a significant negative correlation between RREF and Eo. This correlation was non-linear, and Eo(15 °C) decreased logarithmically with increasing respiration rates at the reference temperature (Fig. 5). It follows that maximum rates of oxygen reduction at 15 °C were between 17 and 20 nmol g−1 s−1, when Eo(15 °C) was at its minimum (compare Fig. 5).

Figure 5.

Correlation between basal respiration rates (RREF) and overall activation energy for oxygen reduction [Eo(REF)]. Note that maximum respiration rates are achieved at 17–20 nmol g−1 s−1, assuming that the activation energy for the reaction remains positive. The non-linear regression was calculated with n = 144. Black symbols, November; Red symbols, February; Blue symbols, August.

Needle age was the most important effect on Eo (Table 2, Fig. 4). On average, Eo was 103 kJ mol−1 in old needles and 64 kJ mol−1 in young needles. Season and canopy position had a lesser effect, but were strongly interactive (Table 3). This effect was further modified by needle age (significant season × canopy × age effect). For the young needles in November, average Eo was 120 kJ mol−1 in the lower, 31 kJ mol−1 in the middle and 18 kJ mol−1 in the upper canopies. In February, average Eo of the young needles was 43 kJ mol−1 in the lower, 70 kJ mol−1 in the middle and 98 kJ mol−1 in the upper canopies. In August, Eo was almost constant throughout the canopy (also compare Fig. 4). Fertilizer application had a minor effect and reduced activation energies by 5% on average.

The parameter δ as a means to describe the dynamic response of RO2 to changes in temperature: a comparison with the Q10 value

The parameters Eo and δ are tightly linked (compare Fig. 6), and changes in Eo with the measurement temperature are therefore entirely predictable. From the viewpoint of a low reference temperature, it is therefore possible to predict the temperature response of RO2 with the knowledge of only two parameters: RREF, which is the respiration rate at the reference temperature, and Eo(REF), which is the overall activation energy at the reference temperature. The precision of RO2 determination will dictate how many measurements are necessary to accurately derive the aforementioned parameters. Employing calorimetric methods, four to five measurements at different temperatures are sufficient for this purpose. For comparative studies, it is necessary to relate measurements to the same reference temperature. This can readily be achieved for any reference temperature. The respiration rate at any temperature is given by Eqn 4 (compare Eqn 3 in the Materials and Methods),

Figure 6.

Correlation between the overall activation energy for oxygen reduction (Eo) and the parameter δ, which describes the dynamic adjustment of Eo to temperature, starting from the reference temperature 288 K (15 °C). The linear correlation explains 94% of the observed variation. Note that δ becomes zero at ca. 45 kJ mol−1. At this activation energy, the overall oxygen reduction is entirely Arrhenius-like, and the temperature sensitivity of respiration does not change across a broad range of measurement temperatures. Every data point depicts the mean ± SD of eight independent replicates. The linear regression was calculated with n = 144. Black symbols, spring; red symbols, summer; blue symbols, winter; closed symbols, old needles; open symbols, young needles; circles, lower canopy; squares, middle canopy; triangles, upper canopy. δ, defines the temperature-dependent change in Eo starting from the reference temperature.

image(4)

where inline image.

The overall activation energy at any temperature (TB) other than the reference temperature can be calculated from the first derivative of Eqn 4,

image(5)

Consider the example in the Materials and Methods, with ln RREF = −0.189 at 15 °C (288 K), Eo(15 °C) = 132 kJ mol−1 and δ(15 °C) = −23.8 103 K2. For TB = 293 K (20 °C), we obtain RO2(20 °C) = 2.0 nmol g−1 s−1 and Eo(20 °C) = 108 kJ mol−1. For TB = 283 K (10 °C), we obtain RO2(10 °C) = 0.29 nmol g−1 s−1 and Eo(10 °C) = 156 kJ mol−1. In this way, the temperature response of respiration can be defined unambiguously, even if respiration measurements are made at different temperatures.

By contrast, the description of the respiratory temperature response by the Q10 value is less accurate, as the Q10 is, by definition, derived from only two measurements (compare Eqn 2, Materials and Methods). Equally, the dependence of Q10 on two measurement temperatures, frequently confounds comparisons among studies (see Fig. 2 in the Materials and Methods). For example, using the same temperature intervals for its calculation, Q10 declines with increasing measurement temperature (or more precisely, Q10 declines, if the mid-point of the temperature interval used for its calculation moves to higher temperatures), but the magnitude of this decline depends on the temperatures chosen for respiration measurements. In the earlier example, with RREF = 0.828 nmol g−1 s−1, Eo(15 °C) = 132 kJ mol−1 and δ(15 °C) = −23.8 103 K2, Q10 decreases from 12.1 at 15.5 °C to 10.6 at 34.5 °C, if we chose 1 °C intervals for its calculation. Conversely, Q10 decreases from 7.2 at 16 °C to 5.64 at 34 °C for 2 °C intervals, and from 4.7 at 17.5 °C to 2.8 at 32.5 °C for 5 °C intervals. In the first scenario, Q10 decreased by 12%, in the second by 22% and in the third by 32%. The decrease is 45%, if we compare Q10 for 15–25 °C (4.6 at 20 °C) with that for 25–35 °C (2.5 at 30 °C).

Complications arise if measurement intervals at the low and high end of the temperature range are not the same. Using the modified Arrhenius equation in place of the Q10 approach, there is only one answer to the question how increasing measurement temperatures affect respiration rates: in the example given here, Eo decreased from 108 kJ mol−1 at 20 °C to 64 kJ mol−1 at 30 °C – a reduction of some 40%. While it could happen that Q10 underestimated the physiological response to increased temperatures in some cases, it clearly overestimates this response in other cases. For example, at ca. 45 kJ mol−1, the parameter δ becomes zero and the overall activation energy does not change with temperature. Still, Q10 declines under these circumstances (compare the example in the Materials and Methods, Fig. 2). Figure 7 depicts the magnitude by which Eo(15 °C) changes with increasing temperatures. In relative terms, the temperature dependent decline in Eo(15 °C) slows at lesser activation energies, and increases at activation energies <45 kJ mol−1. Note that Fig. 7 tells us little about the expected change in respiration in absolute terms. Models that represent respiration as a dynamic function of temperature should also reflect the partial coupling of low respiration rates to large activation energies at low temperatures.

Figure 7.

Dynamic response of Eo to the measurement temperature. The response of Eo to temperature, starting from the reference temperature (15 °C in the present study) depends on the magnitude of Eo(REF). While, in relative terms, the response of Eo to temperature is linear, its strength and the direction depends upon Eo(REF). There is a large relative decrease at high activation energies, which becomes smaller at lower Eo and eventually turns into an increase at activation energies >45 kJ mol−1. Eo, overall activation energy.

The relation between the capacity and the overall activation energy (Eo) for oxygen reduction

The parameter RREF may be an unreliable indicator for the respiratory capacity (combined enzyme activity of COX and AOX) and hence ‘Type II’ acclimation, as RREF is partly dependent on the overall activation energy for oxygen reduction. If the proportional contribution of AOX and COX to the total respiratory capacity is altered, changes in Eo will result in changes of RREF, even though the respiratory capacity may not be affected.

A more reliable parameter for the respiratory capacity can be obtained if the respiratory response is integrated over a distinct temperature interval (i.e. 15 to 35 °C in the present study). For the sake of simplicity, such a dimensionless parameter can also be calculated from the function describing the temperature dependency of ln RO2,

image(6)

where inline image.

We will thus use the dimensionless parameter IA as a measure for respiratory capacity (the combined catalytic activity for oxygen reduction). In contrast to RREF, the parameter IA is less flexible because Eo and RREF cancel each other under many conditions. IA was, conveniently, only affected by the main factors explored in the current study, and some important interactions (compare Table 3). IA was 0.35 for the old needles and 0.49 for the young needles. Combined activity averaged to 0.49 in November, 0.36 in February and 0.42 in August. IA was particularly high for the young needles in November, when it averaged to 0.6 (significant season × age interaction). The combined activity increased from 0.37 in the lower canopy to 0.43 in the middle canopy, and further to 0.46 in the upper canopy. The fertilizer treatment effect on IA was dependent on the season. The application of fertilizer-N resulted in an increase in IA of 5% in November and 15% in February. However, the IA of fertilizer-treated trees was 8% less than the control in August (by this time, needle-N of fertilized trees had also declined, see above).

The overall activation energy for oxygen reduction (Eo) was negatively correlated with the respiratory capacity (IA). This correlation is linear (compare linear regression in Fig. 8), and explains about 50% of the variation in Eo. Additional influences on Eo are illustrated in Fig. 8. We can analyze the strength of these influences by a general linear model (GLM), with IA as a continuous predictor variable and the factors season, canopy, age and treatment as categorical predictor variables. About 96% of the observed variation of Eo(15 °C) can be explained by the impact of these main factors and their interaction (compare Table 3). The GLM separates the effect of the respiratory capacity from other influences. It is possible to isolate the main effectors, and to study their impact on Eo separately (compare Fig. 9). Any increase in respiratory capacity leads to a reduction in Eo, owing to a greater contribution of AOX. This effect is modified by needle age and season. The fraction of AOX activity (as compared with COX activity) is greater in the young needles (Fig. 9d) and during the growing season in spring (Fig. 9b). By contrast, canopy position has a rather large effect on Eo (after being separated from the effects imposed by IA), and acts in the opposite direction: the Eo of the young needles in February, for example, increased from the lower to the upper canopy, because concomitant increases in IA were weak (and because the canopy effect was enforced by the seasonal effect). In November, however, the increase in IA with canopy height (that was exaggerated by season and age effects) far outweighed the canopy effect, resulting in a significant decrease in Eo higher in the canopy (compare Fig. 8). These often opposing effects of canopy position, season, age and IA explain many of the interactive terms that are significant for Eo (compare Table 3). The interaction of Eo with IA affects RREF and produces many significant interactions, as described above.

Figure 8.

The overall activation energy Eo(REF) as affected by the respiratory capacity (IA) and other factors. The effects of season, canopy position, needle age and fertilizer treatment (the latter effect has not been singled out in Fig. 8) were analysed by analysis of variance, and results are shown in Table 3. Variability of Eo was assessed by a general linear model, with IA as continuous predictor variable and the factors season, canopy position, age and treatment as categorical predictor variables (see Table 3 for results). Every data point depicted in Fig. 8 represents the mean ± SD of eight independent replicates. The linear regression was calculated with n = 144. Black symbols, spring; red symbols, summer; blue symbols, winter; closed symbols, old needles; open symbols, young needles; circles, lower canopy; squares, middle canopy; triangles, upper canopy. Eo, overall activation energy; RO2, oxygen reduction; T, absolute temperature.

Figure 9.

Sensitivity of the overall activation energy for oxygen reduction [Eo(15 °C)] to the different impact factors studied. The values that are shown for the dependent variable are the predicted responses at each level of each factor (continuous for IA, categorical for season, canopy, age and treatment), holding all other factors (including the block factor) constant at their current levels. Red line, predicted value for the average activation energy, if all other factors are held constant. IA, respiration capacity.

DISCUSSION

A common cause for two phenomena: how do alterations in the pathway of oxygen reduction affect sensitivity and dynamic adjustment of respiration to temperature?

In the present study, we demonstrated that the shape of the exponential function (i.e. its slope) that defines the temperature response of respiration is tightly linked to the dynamic adjustment of oxygen reduction to temperature. That is, the overall activation energy of the process [Eo(RO2)] was highly correlated with a factor (δ) that describes the dynamic response of Eo to temperature. The same results were also obtained in a large acclimation study comprising 12 different Eucalyptus species (Kruse and Adams, unpublished results). In our companion study (Kruse et al. 2008), we have shown that the application of specific inhibitors for either COX (cyanide) or AOX (SHAM) alters the overall activation energy for oxygen reduction. Inhibition of the COX pathway significantly reduced Eo(REF), coupled to a concomitant increase in δ. By contrast, the inhibition of the AOX pathway acted in the opposite direction (compare Kruse et al. 2008). How can these results and those of the present study be reconciled?

We might first consider a single, membrane-bound enzyme that follows the Michaelis–Menten kinetics. We assume that the substrate concentration at a low measurement temperature is sufficient to allow for maximum velocity of the catalysed reaction (Vmax). With increasing temperature, Vmax will increase, depending on its temperature sensitivity. However, greater Vmax will ensure that substrate concentrations are not sufficient to saturate the enzyme(s), if there are no other compensating processes. This is to say, at greater temperatures, the half-saturation constant (K1/2), which reflects the affinity of the enzyme for the substrate, becomes more relevant to reaction rate. The K1/2 of many enzymes increases with temperature (Davidson & Janssens 2006). Deviation from strictly exponential Arrhenius kinetics, which can be described by the factor δ in phenomenological terms (see above), is probably the result of different temperature sensitivities for Vmax and K1/2 at the molecular level (Davidson & Janssens 2006; Davidson, Janssens & Luo 2006).

The two enzymes COX and AOX, catalysing mitochondrial oxygen reduction, presumably have different catalytic properties. The two oxidases act in parallel and compete for electrons (Hoefnagel et al. 1995). They both use ubiquinone as substrate, freely diffusing in the inner membrane, and their combined activities will affect the temperature response of overall oxygen reduction (RO2). Equation 7 combines Michaelis–Menten kinetics with Arrhenius kinetics to describe the temperature dependency of overall oxygen reduction:

image(7)

where Vmax;REF and K1/2;REF are the maximum velocity and half-saturation point of the respective enzyme at a low reference temperature, Ea(Vmax) and Ea(K1/2) are the temperature sensitivities of the respective enzymes for Vmax and K1/2, C is the substrate concentration at the reactive site of the enzyme, R is the gas constant and inline image.

In our previous research, we observed that the overall activation energy of RO2 at a low reference temperature [Eo(REF)] was significantly reduced when COX was inhibited by cyanide (Kruse et al. 2008). We thus conclude that Ea(Vmax, COX) > Ea(Vmax, AOX). The contribution of AOX to RO2 is flexible, but COX dominates under many conditions (Buchanan et al. 2000). The observation of negative δ values prevailing under many conditions, suggests that Ea(Vmax, COX) ≤ Ea(K1/2, COX). A good example of the reverse situation is that AOX becomes much more prominent in young growing tissues (i.e. it contributes to more than 50% to overall oxygen reduction). Under these circumstances, Eo is rather low (<45 kJ mol−1) and δ becomes positive. This result suggests that Ea(Vmax, AOX) > Ea(K1/2, AOX). The interpretation of our results in this way can entirely explain, why there is a continuum of observations that tightly link Eo(RO2) to δ; as the contribution of AOX to combined oxygen reduction increases (coupled to a concomitant relative decrease of COX activity), Eo(RO2) declines because of different temperature sensitivities for Vmax, and δ increases because of different temperature sensitivities for K1/2 of AOX and COX.

The link between the respiratory capacity and the path to oxygen reduction: re-considering the importance of AOX for plant metabolism

The capacity for mitochondrial respiration depends on the number of active enzymes involved. The respiration rate at a low measurement temperature is a poor marker for the respiratory capacity and hence ‘Type II’ acclimation, because of differing enzymatic properties of COX and AOX (and RREF must always depend on Eo to some degree, see preceding paragraph). The integrative measure IA introduced in the present study, which combines developmental and environmental impacts on RREF and Eo, is a more reliable way to estimate the respiratory capacity, i.e. the sum of active COX and AOX proteins. This indirect physiological approach to gain insight into the mechanisms underlying ‘Type II’ acclimation is comparatively simple, reliable and can be applied in field studies. It can therefore yield valuable supplementary information to biochemical data about mRNA and protein levels, or in vitro activities.

We have shown that the overall activation energy for RO2 (Eo) is negatively correlated with IA, which indicates that the proportion of AOX activity increases with respiratory capacity. The contribution of the AOX to oxygen reduction can be substantial, but its role in plant metabolism remained unclear for many years, partly because its activity does not contribute to mitochondrial ATP production (Lambers, Chapin & Pons 1998). Hence, AOX activity has sometimes been termed a ‘wasteful’ or ‘futile’ component of plant respiration (Thornley & Cannell 2000). However, different lines of investigation suggest a different interpretation (Day et al. 1995; Siedow & Umbach 1995, 2000). It is now well established that the cytochrome pathway is primarily controlled by the availability of ADP and Pi. Low cellular demand for ATP acts via feedback on O2 uptake, and eventually the tricarboxylic acid (TCA) cycle slows. By contrast, AOX is not prone to regulation by adenylates (Wagner & Krab 1995) and operates, during times of large demand for carbon skeleton intermediates, at a faster rate to maintain a high carbon flux through glycolysis and the TCA cycle. Taken together, AOX plays an important role in the regulation of plant metabolism, in addition to its role as a safety valve to prevent the formation of reactive oxygen species under conditions of environmental stress (Simons & Lambers 1999; Moller 2001; Moore et al. 2002; Raghavendra & Padmasree 2003).

It has been suggested that the AOX might function to balance carbon metabolism and electron transport as the supply or demand for carbon skeletons, reducing power and ATP changes (Vanlerberge & McIntosh 1997). The results of the present study support this view. For example, crown development is optimized according to plant water relations (Kruse & Adams 2008); needle growth starts at the top of the tree in spring when environmental conditions and substrate supply are favourable, and summer growth appears to continue for a longer period in the lower canopy. In the present study, needles had reached their final length in the upper canopy by the end of summer, but were still growing in the lower canopy. In the young needles, this led to a strong positive gradient of the respiratory capacity from the lower to the upper canopy in spring. During this time of the year, respiration rates were strongly correlated with needle nitrogen, and mitochondrial enzymes appeared to be fully activated (also compare with Kruse et al. 2008, and Kruse & Adams 2008). Under these conditions growth and, eventually, needle length can be maximized if N-allocation within the canopy follows time-integrated irradiance (as frequently happens; Hirose & Werger 1987; Niinemets, Kull & Tenhunen 2004; Eichelmann et al. 2005). The greater respiratory capacity of upper canopy needles during growth in spring was coupled to enhanced AOX activity. This pattern was then reversed until the end of summer, when the contribution of AOX to the respiratory capacity was greater in the lower as compared with the higher canopy. The shift from ‘growth’ to ‘maintenance’ respiration is associated with a change in the demand for ATP. It has been noted that maintenance processes require a larger amount of ATP, as compared with the demand for reducing power and carbon skeleton intermediates (Amthor 2000). It has also long been known that ‘maintenance respiration’ is more sensitive to temperature changes than ‘growth respiration’ (de Wit, Brouwer & Penning de Vries 1970; McCree 1974). In the present study, needle age had the largest impact on Eo, an effect also observed by Atkin et al. (2005). Clearly, the alternative path is losing importance as needles age. These developmental changes ensure that rates of respiration by older foliage are more susceptible to control by adenylates. By contrast, respiration rates of growing young needles may only be affected in cases of waning substrate supply.

All of the environmental influences considered in the present study affect, in one way or another, needle development and the partitioning of electron transfer between the AOX and COX path. Canopy position, for example, had an important influence on COX activity. After separation from other effects in the GLM, it becomes clear that the canopy gradient generally acts to increase COX activity from the lower to the upper canopy. This effect probably owes much to the larger requirement of ATP for the maintenance of proteins (Ryan et al. 1996; Vose & Ryan 2002), and for the export of sugars higher in the canopy (compare with Kruse & Adams 2008). Taken together, studying the temperature response of respiration can clearly improve our mechanistic understanding of respiration and should help to develop process-based models.

Implications for carbon balance modelling: outlook

Modelling likely long- and short-term impacts of altered temperature on plant growth requires that temperature effects on plant metabolism not be isolated from other environmental variables. For example, respiratory capacity is likely to be greater at higher temperatures, when water availability is abundant. Under these circumstances, the dynamic response to temperature does not necessarily cause respiration rates to decline at high measurement temperatures. Opposite effects of rising temperatures can be expected under conditions of drought. Put differently, the predicted temperature rise will probably have contrasting effects on plant metabolism in the tropics, temperate climes or the boreal zone. Recently, Wythers et al. (2005) estimated aboveground net primary production could increase by ca. 20–40%, after including two simple, general algorithms into growth models: one algorithm described the temperature response of the respiratory capacity and the other one described the dynamic response of the Q10 to temperature. We believe such simplifications and generalizations may not adequately reflect actual changes in plant metabolism in response to temperature.

This is not necessarily a moot point. On a global scale, carbon fluxes associated with autotrophic respiration (~60 Gt C year−1, Amthor 1997) are about one order of magnitude greater than fluxes resulting from fossil fuel combustion (ca. 5.5 Gt C year−1, Schimel 1995). Carbon balance models, that predict increases or decreases in respiration and plant growth by a few percent points as a result of higher global temperatures, can have significant political and policy impact. We strongly suggest that the use of Q10 approaches in carbon balance models is not up to the task. Q10, as a means to precisely evaluate the temperature sensitivity of respiration, is afflicted by considerable shortcomings, some of which have been described here (see Results). By contrast, we have demonstrated that the temperature response of oxygen reduction can be comprehensively and accurately described by three simple parameters in an exponential function. These parameters appear to have a clear physiological underpinning and may assist the development of a process-based model of respiration. The future challenge lies in applying the rationale of the present study to respiratory CO2 release, in order to arrive at a full physiological description of respiratory processes (compare with Kruse et al. 2008).

ACKNOWLEDGMENTS

We wish to thank Hancock Victorian Plantations Pty Limited for the financial and logistic support and Najib Ahmady for the technical assistance. We thank the Australian Research Council for the financial support (Linkage Grant) and Phil Gerschwitz for assistance in the field.

Ancillary