of the T‒R response curve Q10 decreased from December to June, but the elevation R5 was almost identical between these two examples Figure 2.5 . These results suggest that Type I acclimation contributed largely to the seasonal adjustment of leaf R in hinoki cypress.
However, extremely high R20 values, which deviated from the regression lines in Figure 2.9b, were observed for the upper and middle layers in April. The T‒R response curve for April was located considerably higher on the y‐axis R rate than the curve for December over the whole T range Figure 2.5 , although Q10 was lower in April than in December Figure 2.6 . Thus, the high R rates in April were caused by a remarkable increase in base R i.e., R5 . Considering the mechanism of Type II acclimation, this period‐specific phenomena may reflect an increase in overall respiratory capacity. The most likely explanation is that April R rates were strongly affected by shoot growth, which would entail high metabolic demands for biosynthesis associated with localized cell division/expansions processes
Amthor 2000 .
However, the shoot elongation rate alone could not explain the extremely high R rates in April Figure 2.8b because the monthly elongation rates were much higher in June and July Figure 2.7c . Shoot growth began at the end of April, suggesting that the onset of shoot growth must demand more energy, probably for processes such as nutrient and/or carbohydrate re‐allocation Egger et al. 1996, Schaberg et al. 2000, Kitao et al. 2004, Wyka et al. 2016 . However, the effects of the onset of shoot growth and/or leaf development on leaf R remain unclear. This issue will require further study.
2.4.2 Thermal acclimation capacity in tree canopies
There are several methods of evaluating the degree of thermal acclimation of R Atkin et al. 2005, Slot and Kitajima 2015 . For example, the set temperature method quantifies the degree of acclimation as AcclimSetTemp RCold at TSet/RWarm at TSet, where RCold and RWarm indicate the rates of R in cold‐grown and warm‐grown plants, respectively Loveys et al. 2003 . If this method is applied to monthly averaged values of R20 shown in Table 2.1 in this case, TSet 20 °C by setting R20 in the coldest month December; Tamb 4.5 °C as RCold and R20 in the warmest month July; Tamb 26.5 °C as RWarm, then AcclimSetTemp would be 2.30. This value is higher than other AcclimSetTemp values reported for evergreen trees in temperate climates according to a recent meta‐analysis Slot and Kitajima 2015 1.54 0.41
SD and 1.03 0.14 SD from laboratory and field measurements, respectively . However, in the present study, the estimated value of AcclimSetTemp differed according to the chosen combination of measurement months. For example, the combination of December and September the second warmest month; Tamb 22.3 °C results in 1.63 as the AcclimSetTemp. Other methods for quantifying the degree of acclimation e.g., LTR10, AcclimLTR10 and AcclimHomeo, see Atkin et al. 2005 are useful for experimental data, but involve potentially similar problems when applied to seasonal data.
For this reason, it is proposed that the slope of the linear regression between R at a reference T and ambient air T can represent overall seasonal acclimation capacity in addition to the linear decrease in Q10 with increasing ambient air T. This study found that the regression slopes did not differ among the three canopy layers Figure 2.9b . This finding suggests that the degree of thermal acclimation of leaf R did not vary vertically within the hinoki cypress canopy. The slope values in this study 0.0227, 0.0228 and 0.0151 for the upper, middle, and lower layers, respectively are comparable with values reported for P. radiata 0.0175 and P. deltoides 0.0205 Ow et al. 2010 . Moreover, the slope of the regression between Q1020 and ambient air T 0.0167 in this study is identical to that of P. deltoides 0.0167 and similar to that of P. radiata 0.0115 Ow et al. 2010 . These similarities suggest the usefulness of regression slopes for comparing acclimation capacity, although the validity of this method should be further verified. Nevertheless, this study presents the novel finding that thermal acclimation capacity in leaf R may not vary with canopy position.
2.4.3 Vertical variation in temperature response of leaf respiration
In the present study, it was found that there were no vertical differences in Q10 in a hinoki cypress canopy throughout the year Figure 2.6 . This agrees with previous studies demonstrating that Q10 or E0 was constant in the tree canopies of Q. rubra in the northeastern United States Xu and Griffin 2006 , E. globulus in Australia O'Grady et al. 2008, 2010 , and 18 deciduous tree species in the southern Appalachian mountains Bolstad et al. 1999 . If Q10 can be assumed to be constant within a tree canopy, this is helpful for simplifying the representation of leaf R in carbon cycle models. However, there is evidence showing that Q10 varied according to canopy height Griffin et al. 2002, Turnbull et al.
2003 . Further studies are needed to further understanding of vertical trends in Q10 within tree canopies for various forest types.
While Q10 remains poorly understood, the fact that leaf R rate decreases with decreasing height in the canopy has been well studied, including in hinoki cypress Oohata et al. 1971, Hagihara and Hozumi 1977, Ohkubo et al. 2009 . Moreover, numerous studies have demonstrated that within‐canopy variation in leaf R during a given season is associated with LMA, leaf N and leaf carbohydrates e.g., Ryan 1995, Reich et al. 1998a, 1998b, Griffin et al. 2001, Turnbull et al. 2001, Tissue et al. 2002, Turnbull et al. 2003 . Results of the current study agree well with these studies, though leaf carbohydrates were not measured. In the present study, Narea was a better predictor of R20 than LMA in terms of not only vertical but also seasonal variations because there was no seasonal difference in the regression between R20 and Narea Table 2.3, Figure 2.8d . However, it is generally difficult to separate the effects of these variables because Narea is product of LMA and Nmass. Furthermore, these leaf structural and chemical traits often co‐vary within canopies and their vertical patterns are determined primarily by within‐canopy light gradients see Araki et al. 2015 in this case, also see review by Niinemets et al. 2015 .
These results further indicate that light gradients, which can be measured relatively easily, could be a good predictor of vertical patterns of leaf R rates, as shown by early studies Kira et al. 1969, Hagihara and Hozumi 1977, Yoda 1978 . Indeed, in the present study, R20 correlated well with RPPFD Figure 2.8a . Furthermore, the combination of RPPFD and ambient air T was able to predict vertical and seasonal variations in R20 well Table 2.4, Figure 2.10 .
Leaf age may also affect vertical variations in R20. Katsuno‐Miyaura et al. 1996 showed for Japanese cedar Cryptomeria japonica that leaf R was higher in current‐year needles than in older ones. Similar effects of needle age on leaf R have been reported for Pinus sylvestris Zha et al. 2002 and P. banksiana Tjoelker et al. 2009 . Miyamoto et al. 2013 reported that leaf longevity in hinoki cypress ranges from approximately 4 to 6 years, based on stand‐level estimates. The present study was not able to consider leaf age. However, it was observed that current‐year leaves were dominant in sample shoots from the upper canopy and older leaves were dominant in lower shoots, as indicated by cumulative shoot elongation rates Figure 2.7c . Thus, differences in leaf age structure in the sample shoots might be partly responsible for the vertical gradient of R20. In contrast, the lack of vertical difference in Q10 in this study suggests that leaf age might not influence the Q10 of leaf R.