Interactive effects of increased temperature and CO2 on the growth of Quercus myrsinaefolia saplings

Authors


Correspondence: Tetsuyuki Usami, Fax: + 81 298 53 6661; e-mail: tetsuyukiusami@mac.com

Abstract

The interactive effects of increased temperature and CO2 enrichment on the growth of 2-year-old saplings of Quercus myrsinaefolia, an evergreen broad-leaved oak, were studied throughout an entire year in the vicinity of their northernmost distribution. Saplings were grown under different conditions in two chambers: (1) a temperature gradient chamber at ambient temperature, 3 and 5 °C warmer conditions with an ambient CO2 concentration, and (2) in a CO2 temperature gradient chamber at 3 °C warmer conditions with 1·5 times the normal CO2 concentration, and 5 °C warmer conditions with doubled CO2 concentration. The 3 and 5 °C warmer conditions enhanced the relative growth rate during almost the entire year, producing 53 and 47% increases in annual biomass production, 27 and 44% enhancement of root growth during shoot dormancy and 3 and 5 week prolongation of the shoot growing period, respectively. However, a daily mean air temperature exceeding 30 °C under the 5 °C warmer condition caused a marked reduction in net assimilation rate (NAR) from July to September. The CO2 enrichment further enhanced the positive effects of warming in spring and the resulting increases in NAR almost completely compensated for the negative effect of warming during summer. From autumn to winter, attenuation of the effects of CO2 was compensated by the increased sink strength produced by the warming. The annual biomass production was more than doubled by the combination of temperature elevation and CO2 enrichment.

Introduction

Hundreds of experiments conducted over the past few decades have shown atmospheric CO2 enrichment to have a beneficial impact on tree growth and development (see reviews by Kimball 1983; Sionit & Kramer 1986; Eamus & Jarvis 1989; Mousseau & Saugier 1992; Ceulemans & Mousseau 1994; Gunderson & Wullschleger 1994; Curtis 1996; Curtis & Wang 1998; Ward & Strain 1999; Norby et al. 1999). Growth enhancement of woody species is thought to result primarily from increased rates of photosynthesis during exposure to elevated levels of CO2. However, many studies have reported that accelerated rates of photosynthesis are not maintained over long time periods, and substantial reductions in photosynthesis may occur some days or weeks after the initial exposure to elevated CO2 (Arp 1991; Bowes 1991; Thomas & Strain 1991; Sage 1994). At the physiological level, this acclimation of photosynthesis is most often related to reduced sink strength and low nutrient availability (Ward & Strain 1999). If sink strength is low, accumulation of assimilates may occur within the source leaves, leading to feedback inhibition of photosynthesis (Farrar & Williams 1991; Stitt 1991). On the other hand, biomass partitioning – particularly the root : shoot ratio – may also alter the temporal aspects of CO2-induced growth, because it determines the efficiency with which substrates are used, and the extent of the investment in productive organs, and therefore the future photosynthetic potential of the entire plant (Thomas & Strain 1991; Norby et al. 1992; Tjoelker, Oleksyn & Reich 1998).

The warming that is predicted to occur with the ongoing global increases in atmospheric CO2 will affect numerous physiological and morphological aspects of plant development, and alter the source–sink relationship and biomass partitioning (Long & Woodward 1988). Models of the biochemistry of C3 photosynthesis predict that acceleration of photosynthesis by CO2 enhancement could be increased with temperature increases (Long 1991; Kirschbaum 1994), because photorespiratory losses will be reduced by elevated CO2 levels. Idso & Idso (1994) showed that plant growth enhancement due to elevated CO2 was highly significant at higher temperatures. Moreover, the increased supply of assimilates caused by a higher atmospheric CO2 concentration could be combined with the increased sink metabolism resulting from the warming to produce larger plants with less feedback inhibition of photosynthesis and a carbohydrate flux that is independent of the pool sizes (Farrar & Williams 1991). Therefore it is essential to observe the effects of temperature and CO2 levels on plant carbon balance, growth and development, and biomass accumulation (Morison & Lawlor 1999). However, most of the available information concerning the interaction of increased temperature and CO2 has been restricted to annual crops; such information about natural forest tree species is still rather limited (Murray et al. 1994; Wang, Kellomäki & Laitien 1995; Beerling & Woodward 1996; Koike et al. 1996; Norby, Wullschleger & Gunderson 1996; Tjoelker et al. 1998).

Although the rise in atmospheric CO2 is a world-wide phenomenon, very few studies have been conducted into the effects of CO2 enrichment on the warm-temperate evergreen broad-leaved tree species (so-called ‘lucidophyll trees’; Kiyota et al. 1992), which comprise the climatic climax formations that cover an extensive area of the east coasts and adjacent islands of Asia, and extend westwards from Yunnan in China to central Nepal in a narrow belt on the southern slopes of the Himalayas (Kira, Ono & Hosokawa 1978). One species of lucidophyll tree, Quercus myrsinaefolia, is widely distributed throughout Laos, Vietnam, China, Taiwan, South Korea and south-western Japan (Horikawa 1972; Institute of Botany, Chinese Academy of Science 1972). This species is characterized by its determinate shoot growth pattern and the highest cold-tolerance among lucidophyll tree species (Sakai 1977); it is considered to be a climax species in the northernmost warm-temperate regions of Japan.

In the present study we investigated the interactive effects of warming and CO2 enrichment by monitoring for one year Q. myrsinaefolia saplings situated at the northernmost boundary of their distribution.

Materials and methods

Study site and experimental facilities

The experiments were carried out at the Environmental Research Center of the University of Tsukuba (36°05′ N, 140°05′ E, elevation 27 m), Ibaraki, Japan. The annual mean temperature, annual precipitation and potential evapotranspiration at this location were 13·4 °C, 1255 mm, and 748 mm, respectively (during 1981–94). This climate is humid subtropical (Caf type) or humid megathermal (B3A’sb′2 type) in accordance with Köppen’s or Thornthwaite’s climate classification systems, respectively. This site represents the northernmost border of warm-temperate evergreen broad-leaved forests in Asia.

To simulate the globally warmed world, we built a temperature gradient chamber (TGC) and a CO2 TGC (CTGC). These chambers had a characteristic slender shape (2·5 m high, 3 m wide and 30 m long), and were covered with a 0·15-mm-thick transparent agricultural polyvinyl chloride sheet. The relative photosynthetic photon flux density (PPFD) in the chambers was about 80% of full sunlight. Following an idea of Mihara (1971), exhaust ventilators were installed in one side of the chambers to produce a temperature gradient within the chambers. The ventilators continuously exhausted the air from inside each chamber, and the same amount of fresh air was sucked from outside through the opposite side of each chamber. The exhaust rates of the ventilators were controlled by a personal computer so that when the solar radiation was higher than 100 W m−2, the following temperature gradient was maintained: + 3 and + 5 °C at the entrance relative to 15 and 25 m from the entrance, respectively. In the plots these are referred to as T3 and T5, respectively; the temperature at the entrance is referred to as the control. When the solar radiation was lower than 100 W m−2 (including during the night), heated air was supplied to the chambers through long ventilation pipes positioned from the entrance to the exit, in order to maintain the same gradient over a full growing season.

In addition to the above temperature gradient, the CTGC had a CO2 gradient from the ambient concentration at the entrance to 1·5- and 2·0-times this level at 15 and 25 m from the entrance, respectively. In the plots these are referred to as CT3 and CT5, respectively. Pure gaseous CO2 (from liquid CO2) was metered using an electronic mass-flow controller and injected into an intake of a blower installed adjacent to the exit wall. The CO2-enriched air was supplied into the CTGC through long ventilation pipes set from the entrance to the exit so as to produce the above-mentioned gradient in CO2 concentration. For the control and CT5 plots, CO2 concentrations were continuously monitored using CO2 gas analysers (LI-6250 and LI-6252; LI-COR Inc., Lincoln, NE, USA), and the mass-flow controller was automatically controlled by the computer to maintain the desired gradient. Dry- and wet-bulb temperatures were also monitored using ventilation psychrometers positioned 1·3 m above the ground and placed every 5 m from the entrance to the exit inside both chambers. PPFDs inside and outside the chamber were also monitored using quantum sensors (IKS-25; Koitokogyo Co., Yokohama, Japan). The output signals from the sensors were recorded every 10 s using a data logger (Green-kit 100; ESD Co., Tokyo, Japan), and 5 min averages were stored on the computer’s hard disk. The detailed performance of the TGC and the CTGC has been described by Lee et al. (2000) and Lee, Usami & Oikawa (2001), respectively.

Figure 1a shows the seasonal courses of mean air temperatures measured over 10 d in the TGC and the CTGC. It can be seen that the temperature was controlled satisfactorily throughout the experimental period, except for a minor disruption from late March to May 1997. For the control plot, mean ambient temperatures measured over 10 d changed seasonally, ranging from about 0 °C in winter to 25 °C in summer. Therefore, for the T5 and CT5 plots the temperatures did not drop below 5 °C in winter and reached 30 °C in summer. Throughout the entire period, the averages were 12·8 °C (control), 15·5 °C (T3), and 17·3 °C (T5) in the TGC, and 15·5 °C (CT3) and 17·4 °C (CT5) in the CTGC. The average vapour-pressure deficits were 0·28 kPa (control), 0·52 kPa (T3) and 0·67 kPa (T5) in the TGC, and 0·55 kPa (CT3) and 0·75 kPa (CT5) in the CTGC throughout the entire period, which indicates that a greater deficit occurred in the elevated CO2 treatments. This may be due to stomatal closure in response to elevated CO2 concentrations of the plant material in this experiment and/or of the many other herbaceous plants cultivated simultaneously for other experiments that were also present in the chambers.

Figure 1.

Micro-environments in a temperature gradient chamber (TGC) and a CO2-temperature gradient chamber (CTGC) built at the Environmental Research Center of the University of Tsukuba, Ibaraki, Japan. (a) Seasonal courses of mean air temperatures measured over intervals of 10 d at the entrance (control, s), 15 m after (T3, n) and 25 m after (T5, □) the entrance in the TGC and 15 m after (CT3 ▴) and 25 m after (CT5, ▪) the entrance in the CTGC; and that of photosynthetic photon flux density (PPFD) (▾). (b) Seasonal courses of 10 d mean air CO2 concentrations at the control and CT5 plots. (c) Diurnal variations in 5 min mean air CO2 concentrations.

The CO2 concentration for the CT5 plot was controlled satisfactorily throughout the experimental period, except for some disruption in early May and late September 1997 when the CO2 supply was interrupted accidentally (Fig. 1b). Annual average CO2 concentrations were 400 and 768 µmol mol−1, and daytime concentrations were 380 and 735 µmol mol−1 for the control and CT5 plots, respectively. The CO2 concentration for the CT3 plot was periodically confirmed to be about 1·45-times higher than that at the control plot. The diurnal variation in CO2 concentrations ranged from about 50–200 µmol mol−1 (Fig. 1c).

Plant material and culture

More than 500 2-year-old saplings were obtained from a forestry nursery in December 1996; the seeds that these saplings had been produced from were collected from a forest near the Environmental Research Center and were germinated in 0·7 dm3 polyvinyl chloride pots under ambient atmospheric conditions. At the end of January 1997, stem height (H) and stem diameter 2 cm above the soil surface (D) were measured for all saplings, and were 18 ± 2 cm and 4·3 ± 0·6 mm, respectively (mean ± 1 SD). Sixteen saplings whose sizes showed the same mean and variation were harvested to measure the initial biomass and leaf area (the first harvest). Another 100 saplings with the same mean and variation of plant size were then assigned to each of the following five experimental plots: control, T3, T5, CT3 and CT5. Each group was transferred into the TGC or the CTGC and cultivated for 14 months from February 1997 to March 1998. The CO2 enrichments for the CT3 and CT5 plots were started in April 1997.

During the dormant period from February to March 1997, saplings were transplanted into 5 dm3 cylindrical polyvinyl chloride pots (diameter 15 cm, depth 30 cm) surrounded by a 5-mm-thick insulating material. These pots contained a 1 : 1 (v/v) mixture of light-coloured Andosols and leaf mould with 2·5 g fertilizer grains (15 : 10 : 12, N : P : K) as well as 2·5 g dolomite grains. Each sapling was initially fertilized with a slow-release liquid fertilizer every 10 d from February to April 1997 (29 mg N, 58 mg P and 29 mg K), and then with fertilizer granules containing 0·75 g N, 0·47 g P and 0·50 g K every month from May to November 1997; thereafter, half the previous amount of granules was applied once in December 1997 and then again in February 1998. The saplings were irrigated almost daily with an amount sufficient to avoid water deficit, and repositioned weekly to prevent shading as well as to randomize any effect of position between all the pots in each plot.

Plant size and growth analysis

At the beginning of the experiment, the 16 saplings that corresponded to the 43rd to 58th in order of the plant size index (equal to D2H) were selected out of the 100 saplings for each treatment. These saplings were non-destructively measured for D and H at intervals of approximately 10 d throughout the experimental period, and harvested at the end of the experiment (the last harvest). The seasonal variation of D was fitted to a ninth-order polynomial curve. The date at which thickening growth was deemed to have started was defined as that when D-value showed a 0·2 mm increase over that measured during the initial dormant season, and the date for the termination of thickening growth was that when the value of D reached a point 0·2 mm less than the constant values obtained during the last dormant season.

Destructive harvests of 16 saplings in each plot were taken in June and October 1997, and in January 1998 (second, third and fourth harvests, respectively). At each of these harvests, we confirmed that mean and variations of the D2H values obtained from these destructively sampled saplings were similar to those of the initial group followed with the non-destructive measurements (t-test and F-test, respectively; see Appendix 1). All leaves were removed from each flush on the destructively harvested saplings for counting the number and measuring the leaf areas using an area meter (AAM-7; Hayashi Denko Co., Tokyo, Japan). The roots were separated from the shoots after uprooting and washing away the soil. All these samples were dried for 90 min at 100 °C, followed by heating at 70 °C to constant weights. Leaf mass per unit area (LMA) was calculated as leaf dry weight per unit leaf area.

The data obtained from these five harvests were used to establish the regression equations to relate the non-destructive plant size index (i.e. D2H) to whole-plant biomass, shoot biomass and leaf area for each treatment. The residual errors between the observed data and the estimated parameters were inspected to ensure that the estimates were statistically unbiased across the range of plant size and harvest date (Student’s t-test). Whole-plant biomass expressed on a dry weight basis (W, g) was estimated as a function of D2H (cm3):

inline image (1)

[F1,46 > 973, P < 0·01, 0·98 ≥ R2 ≥ 0·95 (anova); for estimations from February to October 1997]; [F1,78 > 481, P < 0·01, 0·95 ≥ R2 ≥ 0·81 (anova); for estimations from November 1997 to March 1998], where a and b are the constants for each treatment (see Appendix 2a). For the estimates from November 1997 to March 1998, we assumed that only the value of b increased in proportion to the number of days, because there was substantial root growth during the shoot dormancy period as described below (see also Appendix 2a–c ).

Shoot biomass was estimated by the same equation throughout the experimental period [F1,78 > 2520, P < 0·01, 0·98 ≥R2≥ 0·97 (anova); see Appendix 2b]. Leaf area (L, cm2) was also estimated as a function of D2H:

inline image (2)

[F3,76 > 360, P < 0·01, 0·96 ≥ R2 ≥ 0·93 (anova); see Appendix 2d], where c, d, e and f are the constants for each treatment. Table 1 lists the values of a, b, c, d, e and f for all plots.

Table 1.  Parameters for Eqns 1 and 2 regressing whole-plant biomass (I), shoot biomass (II) and leaf area (III) as functions of the non-destructive plant size index (D2H)
 ControlT3T5CT3CT5
  1. x For the estimates of whole-plant biomass from November 1997 to March 1998, it was assumed that b-values increased up to those shown as bx in proportion to the number of days (see also Appendix 2a).

I
a 0·685 0·668 0·656 0·710 0·700
b (Feb.–Oct. 1997) 1·033 1·042 1·051 0·979 1·014
bx (Mar. 1998) 1·469 1·489 1·545 1·335 1·508
II
a 0·841 0·823 0·827 0·828 0·842
b 0·103 0·127 0·129 0·150 0·128
III
c−0·0628−0·0333−0·0360−0·0379−0·0391
d 0·5088 0·2694 0·2686 0·3060 0·3201
e−0·4324 0·1094 0·2102 0·0836 0·0355
f 4·9691 4·6450 4·5398 4·6211 4·6435

The time progressions of biomass and leaf area were calculated for each sapling by applying these equations to the data from the periodic non-destructive measurements. Then we used functional growth analysis (Radford 1967; Hunt 1982) to calculate the relative growth rate (RGR), net assimilation rate (NAR), and leaf area ratio (LAR) at intervals of 10 d:

inline image (3)

where t is time (day) and LWR stands for the leaf weight ratio.

Statistical analysis

Data were analysed by one-factor anova when homogeneity of group variance was confirmed by Bartlett’s test. Post hoc multiple comparisons were carried out using Fisher’s protected least significant difference (PLSD) when the anova showed treatment effects to be significant. When non-homogeneity was detected, data were analysed by the Kruskal–Wallis test, and comparisons between selected pairs out of the five treatments were made using Student’s t-test or the Mann–Whitney U test.

Results

Phenology

In the control plot, terminal bud unfolding and thickening growth began in early May (Fig. 2a,b) when the daily mean air temperature reached approximately 16 °C (cf. Fig. 1a). Both the warming and the CO2 enrichments significantly advanced bud burst [Spearman’s correlation coefficient by rank test (Z < −0·4, P < 0·01)] and the start of thickening growth (Table 2). Thus, the saplings in the warmer plots (T3 and T5) began thickening growth 7–11 d earlier, respectively, in late April. On the other hand, the termination of thickening growth in autumn was significantly delayed (by 17–24 d, respectively) in the warmer plots in comparison with that observed in the control plot in early November. Therefore, the 3 and 5 °C warmer conditions resulted in shoot growing periods that were prolonged by 3 and 5 weeks, respectively, compared with that found in the control plot (Table 2). The CO2 enrichments in the CT3 and CT5 plots prolonged the shoot growing period by 2 weeks compared with the results shown by the warmer plots without CO2 enrichment (T3 and T5 plots).

Figure 2.

Seasonal courses of (a) height, (b) stem diameter, and (c) flush frequency for Q. myrsinaefolia saplings grown in the TGC (control, T3 and T5 plots) and the CTGC (CT3 and CT5 plots). Parentheses in (a) enclose the dates when 90% of saplings showed terminal bud unfolding in spring, and those in (b) enclose the dates of the beginning (indicated with ↓) and the end (with ↑) of thickening growth (See also Table 2).

Table 2.  Growth parameters of Q. myrsinaefolia saplings after 14-month cultivation in the five plots under the different environmental conditions in the TGC and the CTGC
 ControlT3CT3T5
CT5
Statistic
  1. Values are means (±SD) of 16 saplings. Means followed by the same letter within a group do not show any significant difference at the 5% level of probability (Fisher’s PLSD). The values F and H in the statistic column correspond to the one-factor anova and Kruskal–Wallis test, respectively. *The end of thickening growth observed in the control plot is significantly earlier than those in other plots [Mann–Whitney U test (U > 235, P < 0·01)]. †The number of leaves in the CT5 plot was significantly larger than that in the T5 plot [Mann–Whitney U test (U = 56, P < 0·01)]. §The frequency of flushesfound in the control plot was significantly lower than those in other plots [Mann–Whitney U test (U > 173, P < 0·05)]. Biomass is expressed on a dry-weight basis.

Height (cm) 60 ± 15a 86 ± 18bc 79 ± 20bF4,75 = 8·5
 100 ± 29c 94 ± 20bcP < 0·01
Stem diameter (mm)11·3 ± 1·8a 13·6 ± 1·6b 13·4 ± 1·2bF4,75 = 16·6
 14·5 ± 1·6bc 15·3 ± 0·8cP < 0·01
 Start of thickening (Julian day)123 ± 12a 116 ± 10ab 112 ± 8bF4,75 = 10·2
 103 ± 15c 102 ± 9cP < 0·01
 End of thickening (Julian day)307 ± 15* 324 ± 6 331 ± 6H = 47, P < 0·01
 327 ± 8 335 ± 6 
 Period of thickening (d)185 ± 17a 209 ± 13b 220 ± 10cF4,75 = 26·7
 223 ± 17cd 234 ± 14dP < 0·01
Biomass (g)87·7 ± 37·8a134·2 ± 36·7b128·8 ± 32·0bF4,75 = 25·8
184·5 ± 45·1c210·7 ± 38·6cP < 0·01
 Leaf (g) 21 ± 9a 30 ± 7b 30 ± 7bF4,75 = 23·8
 38 ± 9c 48 ± 9dP < 0·01
 Stem (g) 27 ± 12a 49 ± 15b 48 ± 13bF4,75 = 26·1
 64 ± 17c 78 ± 17dP < 0·01
 Root (g) 40 ± 19a  56 ± 17b  50 ± 16abF4,75 = 16·2
  82 ± 25c  85 ± 20cP < 0·01
Leaf area (100 cm2)18·5 ± 6·7a 23·9 ± 5·9b 22·4 ± 4·3abF4,75 = 9·3
 26·3 ± 6·0b 30·6 ± 6·5cP < 0·01
 Number of leaves 208 ± 85 255 ± 81  221 ± 81H = 13·1, P = 0·01
 243 ± 89 325 ± 103 
 Frequency of flushes 3·8 ± 0·4 § 4·3 ± 0·6 4·3 ± 0·5H = 11·8, P = 0·02
 4·2 ± 0·6 4·4 ± 0·5 

The saplings in the control plot flushed three times from May to August 1997, and 80% of them presented a fourth flush in September (Fig. 2c). In the warmer plots, the saplings already showed the fourth flush in August owing to their advanced bud bursts in spring, and about 20–40% of them gave an additional fifth flush by October. The final height observed in the warmer plots in March 1998 was significantly higher (by 32–67%) than that in the control (Table 2).

Biomass, leaf area and partitioning

The biomass observed in the control plot started from 6·8 ± 1·8 g and increased, finally reaching 87·7 ± 37·8 g at the end of the experimental period (Fig. 3a). The warming significantly increased the final biomass, by 53% in the T3 plot and 47% in the T5 plot (Table 2). The combination of warming and CO2 enrichment resulted in an extremely significant increases in the final biomass, by 110% in the CT3 plot and 140% in the CT5 plot. The 1·5 times (CT3) and doubled CO2 concentrations (CT5) brought about 37 and 64% increases, respectively, in the final biomass compared with those treated in the comparable warmer plots without CO2 enrichment (T3 and T5, respectively).

Figure 3.

Seasonal courses of (a) whole-plant biomass, (b) shoot biomass, (c) root biomass and (d) leaf area for Q. myrsinaefolia saplings grown in the TGC (control, T3 and T5 plots) and the CTGC (CT3 and CT5 plots). Each value was estimated as a function of non-destructive plant size index (D2H) related to the biomass (expressed on a dry weight basis) or leaf area by using data obtained from the five harvests during the experimental period.

The leaf area per tree observed in the control plot started from 159 ± 39 cm2, finally reaching 1855 ± 678 cm2 at the end of the shoot growth period (Fig. 3d). The final leaf areas observed in the T3, CT3 and CT5 plots significantly increased by 29, 42 and 65%, respectively, but a difference observed between the T5 and control plots (of 21%) was statistically not significant (Table 2). The doubled CO2 concentration in the CT5 plot brought about a significant extension in leaf area, by 37% compared with that observed in the T5 plot. These increases were not due to larger leaves but to a greater number of leaves per tree. Almost all leaves that had been produced before the start of the experiment were shed during the experimental period.

Table 3 lists the LMA and root : shoot ratio obtained at the five harvests. In the control plot, the LMA decreased by 12% during the shoot growing period and then recovered to approximately the initial value during shoot dormancy. Both the warming and the CO2 enrichments enhanced the increase in the LMA during shoot dormancy. In March 1998, the LMA observed in the T3, T5, CT3 and CT5 plots was significantly larger (by 9, 19, 28 and 38%, respectively) than that observed in the control The 1·5 times (CT3) and doubled (CT5) CO2 concentrations significantly increased the LMA (by 18 and 15%) compared with those obtained in the corresponding warmer plots without CO2 enrichment (T3 and T5, respectively).

Table 3.  Leaf mass per unit area and root : shoot of Q. myrsinaefolia saplings observed for each harvest during 14-month cultivation in the five plots under the different environmental conditions in the TGC and the CTGC
 ControlT3
CT3
T5
CT5
Statistic
  1. Values are means (±SD) of 16 saplings. Means followed by the same letter within a group do not show any significant difference at the 5% level of probability (Fisher’s PLSD). The values F and H in statistic column correspond to one-factor anova and Kruskal–Wallis test, respectively. xCalculation was based on the estimated root biomass data shown in Fig. 3c.

Leaf mass per unit area (LMA, mg cm2)
 February 199711·9 ± 0·9   
 June10·5 ± 1·910·8 ± 1·610·1 ± 1·6Not significant
11·3 ± 1·211·1 ± 1·8(One-factor anova)
 October10·5 ± 0·9a11·1 ± 1·2a10·6 ± 1·3aF4,75 = 13·7
12·0 ± 1·0b13·1 ± 1·3cP < 0·01
 January 199811·6 ± 1·0a12·2 ± 0·9a11·8 ± 0·9aF4,75 = 29·3
13·4 ± 1·0b14·9 ± 1·3cP < 0·01
 March11·4 ± 1·3a12·4 ± 1·2b13·6 ± 1·1cF4,75 = 23·7
14·6 ± 1·3d15·7 ± 2·0eP < 0·01
Root : shoot
 February 19971·22 ± 0·34   
 June0·82 ± 0·220·81 ± 0·340·54 ± 0·10H = 39, P < 0·01
0·53 ± 0·110·58 ± 0·09 
 October0·41 ± 0·10a0·33 ± 0·09bd0·25 ± 0·06cF4,75 = 7·2
0·34 ± 0·11b0·28 ± 0·07cdP < 0·01
 January 19980·60 ± 0·110·53 ± 0·100·51 ± 0·10Not significant
0·57 ± 0·110·52 ± 0·09(One-factor anova)
 March0·81 ± 0·19a0·71 ± 0·19ab0·64 ± 0·14bF4,75 = 3·9
0·81 ± 0·19ab0·68 ± 0·14bP < 0·01
Root biomass ratiox3·06 ± 0·16a3·89 ± 0·22b4·40 ± 0·21cF4,75 = 129
(March/October) 3·69 ± 0·22d4·39 ± 0·64cP < 0·01

The elevated CO2 by itself did not significantly alter the root : shoot ratio at the final harvest in March 1998 compared with those observed in the warmer plots without CO2 enrichment (Table 3), as was the case for growth under high nutrient conditions (Eamus & Jarvis 1989; Mousseau & Saugier 1992). On the other hand, the root : shoot ratio observed in the T5 plot was significantly decreased (by 21%) compared with the control, as was often found under warmer conditions (Farrar 1988). However, the root : shoot ratio in all plots showed marked decreases and increases during the experimental period (Table 3). This is because the shoot growth had priority from May to October, and roots grew mainly during the rest of the year when shoots remained dormant (cf. Fig. 3b,c). The root biomass more than tripled in all plots from October 1997 to March 1998, and the root growth in this period was further enhanced (by 21–44%) in the warmer plots compared with that in the control (Table 3).

Growth analysis

The seasonal course of the RGR was largely regulated by that of the NAR(Fig. 4a–d; see Eqn 3), and nearly paralleled that of the air temperature (cf. Fig. 1a). Thus the RGR and NAR were positively correlated with the air temperature in all plots (R > 0·8, P < 0·01), and they increased with the air temperature when it was lower than 30 °C (Fig. 5a,b). Owing to this temperature dependence of the RGR, the warming significantly enhanced the RGR values obtained in the T3 and T5 plots during most of the year, except in summer (Fig. 4a). The enhancement of the RGR in spring was attributable to increases in both the NAR and the LAR that were induced partly by the advanced bud burst in the warmer plots. From October 1997 to March 1998, the saplings in the T5 plot maintained RGR values more than 30% larger than those exhibited by the control plot. However, when the daily mean air temperature exceeded 30 °C in the T5 plot from mid-June to early September, the NAR was significantly depressed to values that were less than 70% of that in the control (Figs 4c & 5b). Thus the RGR exhibited by the T5 plot remained less than or equal to that obtained in the control during summer (Figs 4a & 5a), in spite of a marked increase in the LAR (Figs 4e & 5c). The annual maximum values of the mean RGR measured over intervals of 10 d were observed for all treatments in July, of which that for the control saplings was 15·9 mg g−1 d−1, whereas a 15% increase and a 5% decrease was shown by the saplings treated in the T3 and T5 plots, respectively, both in comparison with the control value.

Figure 4.

Seasonal courses of (a, b) 10 d mean RGR, (c, d) NAR and (e, f) LAR for Q. myrsinaefolia saplings grown in the control, T3 and T5 plots in the TGC (on the left-hand side column) and in the CT3 and CT5 plots in the CTGC (on the right-hand side column). Note that the values for saplings in the control plot shown in both columns are the same but indicated by the different marks (solid and open) for easier comparison with marks for other plots. Values are mean with SD bars (n = 10). Where no bar is visible, the SD is smaller than the symbol size.

Figure 5.

Temperature dependencies of (a) RGR, (b) NAR and (c) LAR, which were obtained by replotting the values given in Figure 4 against the mean air temperatures during the corresponding period shown in Figure 1a.

In the CT3 and CT5 plots, elevated CO2 concentrations significantly enhanced the RGR and NAR values from spring to early summer compared with those in the warmer plots without CO2 enrichment (Fig. 4a–d). The doubled CO2 concentration in the CT5 plot almost completely compensated for the depression of the NAR observed in the T5 plot in summer (Figs 4d & 5b). Accordingly, the annual maximum value of the RGR in the CT3 and CT5 plots significantly increased (by 44 and 32%, respectively) relative to the control. On the other hand, the effect of CO2 enrichment alone on the RGR was hardly detected in the CT5 plot from October 1997 to March 1998, compared with those obtained in the T5 plot (Fig. 4a,b). This was caused by marked reduction in the LAR in the CT5 plot (Figs 4e,f & 5c), which neutralized the significant increase in the NAR (Figs 4c,d & 5b). The reduction in the LAR was partly caused by the increase in the LMA (cf. Table 3; see Eqn 3). However, the saplings treated in the CT5 plot maintained significantly larger RGR (by 29–74%) than that exhibited by the control saplings from October 1997 to March 1998. The effects of CO2 enrichment on the RGR and the NAR in the CT3 plot were negative (−19%) in comparison with those in the T3 plot in January and February 1998.

Discussion

Effects of warming

Warming apparently enhanced the RGR of Q. myrsinaefolia saplings in the vicinity of the northernmost boundary of their distribution, except in summer. This enhancement brought about a significant increase in the total biomass production by the end of the experiment. These results are consistent with those of other studies that have shown that trees of most temperate and boreal species respond to increased temperature by growing faster when they are well supplied with water (e.g. Cannell, Grace & Booth 1989; Kozlowski, Kramer & Pallardy 1991). Beuker (1994) also reported that Pinus sylvestris and Picea abies in Finland showed marked increases in wood production when they were transplanted from their northernmost limit southwards to where the annual mean effective temperature sum is close to that expected in northern areas as a result of projected global climatic changes.

On the other hand, warming significantly depressed the NAR in summer when the daily mean air temperature exceeded 30 °C in the T5 plot where the midday leaf temperatures often exceeded 40 °C. Such temperatures apparently exceed an optimum for the growth of Q. myrsinaefolia saplings even in the northernmost boundary of their distribution. However, the negative effects of warming in summer lasted only 1–2 months, and the positive effects during the other 10–11 months overcame these negative effects. The similar value for the annual biomass production obtained in both the T3 and T5 plots was due to the enhancement in the T3 plot being equal to the net enchantment in the T5 plot where the summer depression in productivity was compensated for with enhanced production throughout the rest of the year. Beuker (1994) also showed that an increase in the annual mean temperature sum of the sort projected by climate change scenarios might have a negative effect on the growth of trees in their southernmost habitats. Therefore, in warmer habitats the negative effects of warming in summer may be more serious and counteract the positive effects on the annual biomass production of Q. myrsinaefolia saplings.

Among the benefits of warming to this evergreen oak, initiation of shoot growth was advanced in spring and induction of shoot dormancy was delayed in autumn. The prolongation of the shoot growing period enabled the saplings to present an additional flush of new shoots in the warmer plots. Such predominating temperature effects on phenology can usually be seen in forest trees at different elevations (Kozlowski et al. 1991). On the other hand, note that a marked change in the seasonal course of biomass partitioning was apparent for the root : shoot ratio. As shown in Fig. 3 and Table 3, the root biomass more than tripled during shoot dormancy, and the root growth was further enhanced by warming. Since roots can usually store substantial amounts of carbohydrate (Kozlowski 1992), the continuous root growth during shoot dormancy should provide a large reservoir of carbohydrate for continual carbon assimilation during cold months. Moreover, the enlarged reservoir of carbohydrate in the warmer plots could give a further enhancement to biomass production in the subsequent year, because the stored carbohydrates usually support growth in the following spring (Kimura 1969; Kozlowski 1992).

Kozlowski et al. (1991) demonstrated that the roots of many woody plants in the temperate zone probably have no inherent dormancy, whereas shoot growth processes such as leaf initiation and cambial growth usually require minimum temperatures (about 16 °C in the present study) higher than those required by root growth [between 0 and 5 °C (Kramer & Kozlowski 1979)]. Thus the shoot growing period and the seasonal course of biomass partitioning for Q. myrsinaefolia may also be influenced by the fact that the individual threshold temperatures required for shoot and root growth are different.

Effects of CO2 enrichment

The doubled CO2 concentration resulted in a very significant (i.e. 64%) increase in the total biomass production compared with the corresponding warmer plots (CT5 versus T5), which is comparable with the mean biomass increment observed for many deciduous broad-leaved trees (63%) and larger than that for coniferous trees (38%) (Ceulemans & Mousseau 1994). On the other hand, the effects of warming on the total biomass production in the T3 (53%) and T5 (47%) plots were also comparable with those of the CO2 enrichment for many tree species.

A very important finding was that the doubled CO2 concentration almost completely compensated for the negative effects of excess warming on the RGR and NAR values observed in the T5 plot in summer. Using biochemically based mechanistic models of C3 photosynthesis, Long (1991) and Long, Osborne & Humphries (1996) showed that the effects of instantaneous stimulation of CO2 uptake and the resulting improvement in water-use efficiency (WUE) by elevated CO2 will be larger at higher temperatures. This is brought about by the temperature-dependent increase in the fraction of carbon lost by photorespiration under ambient CO2 concentrations. Thus, the stimulating effect of high CO2 concentration also increases with temperature. Reviewing more than 50 experiments on CO2 enrichment, Idso & Idso (1994) stated that the growth-enhancing effect of atmospheric CO2 enrichment typically increased with increasing air temperature over the entire range of temperatures investigated (5–45 °C). Lewis, Tissue & Strain (1996) and Tissue, Thomas & Strain (1997) also showed that the relative enhancement of photosynthesis in Pinus taeda produced by an elevated CO2 concentration was greater in summer (60–130%) and lower in winter (14–44%). Moreover, the increase in WUE under higher CO2 concentrations enables plants to maintain a higher water potential, making them generally more drought-resistant in comparison with those grown under the ambient CO2 level (Tyree & Alexander 1993). Therefore, such temperature dependence of the CO2 enrichment effects and improved drought resistance should enable Q. myrsinaefolia saplings in the CT3 and CT5 plots to maintain a higher growth rate during summer.

However, the effect of CO2 enrichment on the RGR in the CT3 and CT5 plots was hardly detected from autumn to winter. This could also be attributed to the temperature dependence of the CO2 enrichment effects that may have negative impacts on assimilation rate (Long 1991) and biomass production (Idso et al. 1987) at low air temperatures. Sage & Sharkey (1987) showed that lower air temperatures induced insensitivity of photosynthesis to CO2 at lower partial pressures of CO2. This means that CO2 enrichment at low temperatures will be accompanied by an increase in feedback limitations on photosynthesis. On the other hand, photo-inhibition caused by chilling temperatures may be more serious for evergreen tree species because the activities of anti-oxidative enzymes and the chlorophyll content could decrease under the elevated CO2 concentrations (Roden & Ball 1996; Pritchard et al. 2000). Moreover, the temperature decline from autumn to winter also reduced the RGR and could diminish sink strength, which is a common denominator causing acclimation and starch accumulation in leaves (Poorter 1993; Sage 1994). The substantial increase in the LMA observed in the CT5 plot may indicate a reduction in sink strength and subsequent starch accumulation during winter (cf. Table 3), because carbohydrate accumulation in evergreen leaves usually leads to an increased LMA (e.g. Kozlowski & Clausen 1965; Idso, Kimball & Hendrix 1993).

Interactive effects of warming and CO2 enrichment

Although the effect of CO2 enrichment on the RGR was attenuated from autumn to winter, the saplings in the CT5 plot maintained an RGR values that was larger than that exhibited by the control. In this period, the root growth observed in the CT3 and CT5 plots was also enhanced compared with that in the control. These growth enhancements were apparently due to the warmer growth conditions during the cold season. Therefore, the most important finding in the present study is that the interactive effects of warming and CO2 enrichment were synergistic in spring but complementary from summer to winter. That is, the advanced leaf area extension and the increased growth rate produced by the warming were enhanced by the increased NAR as a result of the CO2 enrichment in spring. Then, the elevated CO2 concentration compensated for the depression of NAR observed in the warmer T5 plot in summer. Subsequent warming from autumn to winter compensated for the reduction of CO2 enrichment effect through the enlarged sink strength caused by such factors as the prolonged shoot growing period, an additional flush of new shoots in autumn, and the enhanced root growth during shoot dormancy. Therefore, the interactive effects of warming and CO2 enrichment brought about a larger growth enhancement compared with that obtained by warming alone. Obviously the seasonal change of temperature played a very important role in the interaction. Our experimental facilities did not include plots for CO2-enriched condition without warming, and we can not interpret the effects of the combination of temperature increase and CO2 enrichment to be of synergistic or additional in spring. This is a flaw in the design of our experiment. However, based on the above-mentioned mechanisms of the temperature dependence of the effects of CO2 enrichment, we speculate that CO2 enrichment without warming would be likely to be less beneficial for the growth of Q. myrsinaefolia than that given in the CT3 or CT5 plot, as there could neither be the synergistic (or additional) enhancement in spring nor the complementary increase in sink strength from autumn to winter. These mechanisms of the interaction provide important bases for the mechanistic terrestrial ecosystem models such as the Sim-CYCLE developed by Ito & Oikawa (2000), as little information is available on the interactive effects of warming with CO2 enrichment on growth and gas exchange for evergreen broad-leaved tree species, and what is available is limited to experimental periods restricted to less than 1 year (Murray et al. 1994; Wang et al. 1995; Tjoelker et al. 1998).

There are some predictions that disturbance and decline of biomass production will be greater in ecotones owing to the difference between the shifting of climatic zones and the migration rate of vegetation formations under climate change (e.g. Ohta, Uchijima & Oshima 1993). However, the present study showed that the sapling growth rate was more than doubled in the CT3 and CT5 plots situated at the vicinity of the northernmost distribution of this species. Therefore, the predicted global warming will enhance the growth of Q. myrsinaefolia saplings in natural forests, and accelerate succession and poleward migration into the ecotone lying between warm-temperate evergreen broad-leaved forests and cool-temperate deciduous broad-leaved forests. However, some problems remain for future study regarding the application of the results that we have obtained in saplings treated on an experimental scale to natural forests on the regional scale.

Acknowledgments

We are grateful to the staff of the Environmental Research Center of the University of Tsukuba for their support throughout this work and to Dr Richard Weisburd and Dr Shigeko Nakanishi for their critical reading of this manuscript. This study was financially supported by a grant-in-aid from the Ministry of Education, Science and Culture, Japan to T.O. (08408019).

Received 7 February 2001;received inrevised form 17 June 2001;accepted for publication 17 June 2001

Appendices

Appendix 1

Table A1.  Mean and SE of D2H of samples (n = 16) used for destructive and non-destructive measurements
 ControlT3T5CT3CT5
Jun. 1997
 Non-destructive9·2 ± 0·611·0 ± 0·814·3 ± 1·312·6 ± 0·918·4 ± 1·6
 Destructive9·0 ± 1·09·6 ± 0·813·8 ± 1·313·0 ± 1·315·8 ± 1·7
 t-value−0·16 (P== 0·87)−1·28 (P== 0·21)−0·27 (P== 0·79)0·25 (P== 0·80)−1·15 (P== 0·26)
 F-value 2·56 (P== 0·08) 1·00 (P== 0·99) 1·05 (P== 0·92)2·18 (P== 0·14) 1·32 (P== 0·81)
Oct. 1997
 Non-destructive67·9 ± 7·8111·1 ± 9·1101·0 ± 9·4157·5 ± 17·9151·5 ± 12·8
 Destructive61·3 ± 6·2 91·7 ± 7·0107·7 ± 12·1165·6 ± 14·7165·8 ± 14·9
 t-value−0·69 (P== 0·50)−1·76 (P== 0·09)0·45 (P== 0·66)0·36 (P== 0·72)0·75 (P== 0·45)
 F-value 0·62 (P== 0·38) 0·48 (P== 0·17)1·65 (P== 0·34)0·68 (P== 0·46)1·36 (P== 0·56)
Jan. 1998
 Non-destructive79·7 ± 9·4152·9 ± 15·0145·2 ± 12·3207·5 ± 21·9218·3 ± 18·4
 Destructive79·5 ± 9·0146·4 ± 11·4147·8 ± 15·8238·0 ± 17·9250·2 ± 19·9
 t-value−0·01 (P== 0·99)−0·36 (P== 0·72)−0·14 (P== 0·89)1·11 (P== 0·27)1·22 (P== 0·23)
 F-value 0·92 (P== 0·88) 0·58 (P== 0·31) 1·65 (P== 0·34)0·67 (P== 0·44)1·16 (P== 0·77)

Appendix 2

Ancillary