This study was performed in a Moso bamboo stand at the Kasuya Research Forest of Kyushu University, Japan (33°37 ‘N, 130°31 ’E). The stand density was 10,600 culms ha^–1. According to meteorological measurements 15 km from the site, mean annual air temperature between 1995 and 2005 was approximately 16^°C, and the mean annual precipitation was 1790 mm (Kume et al., 2010). In 2013, a study area of 10 m by 10 m was selected within this bamboo stand. All culms within the plot into five age classes: current, 1, 2, 3-year-old, and above 4-year-old by yearly labeling of new culms between 2013 and 2015. Current year culms correspond to the culms sprouting in April 2015 and are referred to as “current year culms” in the present study. Incidentally, 1-year-old culms did not sprout in this plot in 2015 (Figure 1). Culm circumference in the plot ranged between 15 to 50 cm, with the highest frequency between 35 and 40 cm. Mean DBH in this plot was 11.6 cm (Katayama et al., 2018). In June 2015, three culms each from four age groups (current, 2, 3, and above 4-year-old) were randomly selected from the plot for respiration measurements. However, due to equipment failure, data from one culm each from the 3-year-old and above 4-year-old group could not be acquired, and therefore these two age groups have data from two culms. Respiration measurements on the ten culms were conducted once on each culm before harvesting. The culm circumference of the ten measured culms ranged between 26–40 cm, with an average circumference of 36.0 ± 4.3 cm.

Respiration measurements were conducted based on the methods outlined by Katayama et al. (2014), using the closed system with an IRGA detector (GMP343, Vaisala, Helsinki, Finland). Prior to measurements, collars made of polyvinyl chloride were sealed to the culm surface with silicon sealant and non-drying clay, approximately 1.2 m high on the culm. During measurements, the IRGA detector was attached to the cylinder on the culm surface on the culm surface to measure the CO₂ efflux of each culm. Air inside the chamber was mixed with a small fan. CO₂ concentration inside the chamber was measured every 5 s and recorded by a monitor (M170, Vaisala, Finland). Air temperature and stem surface temperature were measured with a thermometer (TR- 52i; T&D Co. Ltd., Matsumoto, Japan). All respiration measurements were conducted before dawn to reduce the effects of sap flow and bark photosynthesis (Teskey et al., 2008).

Culm respiration rate (R_culm, μmol m^–2 s^–1) was calculated using Equation (1) (Katayama et al., 2014).

where V is the chamber volume (0.27 L), q is the air temperature inside the chamber (°C), A is the projected area of the chamber (0.3849 × 10^–3 m²) and dc is the increment of CO₂ concentration during the incubation time (dt). Equation (1) assumes that efflux rate is measured under standard barometric pressure. All measurements at each point were repeated two or three times and the means were used.

After the culms were harvested, culm height, culm circumference and culm thickness were measured. One sample from breast height of each culm was used to measure wet volume for culm wood density (g cm^–3) calculation. These samples were then dried for a week in an oven (20-802 P-1, IKEMOTO, Tokyo, Japan) set at 50^°C until sample weights stabilized. Weight after 1 week was determined as the dry weight, which was used to calculate culm wood density. Small portions of the dried samples of each culm were ground with a mill. Nitrogen content (%) in each sample was determined using a CN analyzer (MT-600, Yanaco, Japan) to determine the nitrogen concentration. Nitrogen content per projected chamber area (g cm^–2) was calculated by multiplying nitrogen content (%), culm wood density (g cm^–3) and culm wall width (cm).

Samples 1 cm × 10 cm × culm thickness was obtained from one internode of each harvested culm at approximately DBH position, where the respiration measurement chamber was attached, avoiding areas with nodes. Embedding and sectioning of sample culm sections (approximately 2 mm × 10 mm × culm thickness) were performed following the protocols of Barbosa et al. (2010) and Ferreira et al. (2014). Following the methods outlined by Stockfors and Linder (1998), the sections were washed in distilled water, and stained with 0.05% Coomassie Brilliant Blue solution, which reacts with proteins. The sections were then consecutively stained with Iodine solution. The stained transverse sections were observed with a light microscope (ECLIPSE E600, Nikon, Tokyo, Japan). Cell wall thickness and starch content of ground parenchyma cells, which account for about 50% of bamboo internodes and have larger lumen diameter than fibers (Alvin and Murphy, 1988; Liese, 1998), were measured using ImageJ ver. 1.50b (National Institutes of Health, United States) on cross sectional images taken by a digital camera (NEX5N, Sony, Tokyo, Japan). For detailed examination of cell wall thickness, each section was divided into three from the epidermis side to lacuna side when as inner, middle, and outer region. Thirty parenchyma cells from each region were randomly selected for cell wall thickness and starch content measurement.

We did not find any relationships of culm respiration with morphological traits for Moso bamboo. For trees, morphological factors such as DBH and sapwood volume determine stem respiration rate (Ryan, 1990; Lavigne et al., 1996; Cavaleri et al., 2006; Katayama et al., 2016). Therefore, it is presumable that larger bamboo culms may have higher culm respiration. However, our results did not agree with this assumption, and morphological properties such as culm circumference, culm thickness, and culm height did not show any relation to the respiration rate of all culms (Table 2). Tree increases the amount of living cell in their sapwood for several years although it has variations between species (Bamber, 1987). On the other hand, moso bamboo has only the xylem formed in the current year. The no relationships between culm respiration and morphological traits can be attributed to the growth property of moso bamboo as an annual herbaceous plant.

We found high variation among age in culm respiration for Moso bamboo: respiration rate for current year culms were much higher than that of older culms (Figure 2). This result was different from the previous study of Isagi et al. (1997), who reported that culm respiration decreased rather linearly with culm age. However, in our study, a sudden drop-in respiration rate was observed between current year culms and older culms. The respiration rates of older culms were between four to ten times smaller than that of younger culms (Figure 2). The differences in results may be due to different sample designs and respiration measurement methods. Isagi et al. (1997) used five culms, which have different age between 0.5 and 12.5, and measured their respiration after harvesting.

The respiration rates for current year culms (1.9 ± 0.46 μmol m² s^–1) was surprisingly high for their diameter, leveling the respiration rates of large trees in tropical rain forests [approximately 2.0 μmol m² s^–1, 25^°C, DBH > 60 cm, Katayama et al. (2014)]. On the other hand, respiration rates of older culms (0.17 ± 0.09 μmol m² s^–1) were lower than that of 50-year-old Chamaecyparis obtusa trees growing in Japan, whose respiration rate was measured in July [approximately 0.40 μmol m² s^–1, DBH = 14.9 cm, Araki et al. (2010)]. Thus, the considerable change among culm age in respiration can be one of characteristics of bamboo species.

Culm wood density increased with age, and the largest increase occurred in the first two or three years (Figure 3). Strong relationships between cell wall thickness of ground parenchyma cells and culm wood density (R = 0.89, P < 0.01) suggested that density increase was mainly caused by the increase in the cell wall thickening of culm cells (Table 1 and Figure 4). Although this study only examined the parenchyma cells, previous studies suggested the increase of fiber wall thickness with age (Liese and Weiner, 1996; Huang et al., 2015), which is also a factor likely to have contributed to the density increase. The increase of culm wood density and the cell wall thickness by aging were consistent with earlier studies conducted on other bamboos (Alvin and Murphy, 1988; Wahab et al., 2010), although these studies did not conduct the annual culm sampling from current year. The presence of starch in the tissues of the bamboo has been reported before (Liese, 1998; Lian et al., 2021). In this study, starch was observed in the older culms but not in the one or two-year-old culms. This suggests the younger culms were metabolically active and consume carbohydrate immediately.

A substantial amount of respiration must have been required for the large increase in biomass of culms. In previous studies of tree stem respiration, growth respiration has often been shown in relation with stem diameter increment (Ryan, 1990; Maier, 2001). Whereas trees require growth respiration to grow outward (as stem diameter growth), this study suggests that bamboos require respiration to grow inward (as cell wall thickening) with no change in diameter, which may be defined as a type of growth respiration.

Young culms would also have comparatively higher maintenance respiration than old culms. The thickening of parenchyma cell walls with age (Table 1) causes limitation in cytoplasm area in ground parenchyma cells (Figure 4). A decrease in cytoplasm area with culm age would have resulted in a decrease in respiration, because respiration is conducted in the cytoplasm. Current year culms with a larger area of cytoplasm potentially needed larger maintenance respiration than older culms. In addition, many previous studies on tree stem respiration have shown strong positive relationships between stem respiration and nitrogen content (Lavigne et al., 1996; Maier et al., 1998; Ceschia et al., 2002). The present study revealed a significant relationship between culm respiration rate and nitrogen content (Table 2). Ryan et al. (1996) has pointed out that nitrogen is tightly linked with cellular activity and result in higher maintenance respiration. Higher nitrogen content in current year culms would be related to the high maintenance respiration rates in the current year culms. Larger live tissue volume and higher nitrogen concentration in current year culms potentially contribute to higher growth respiration and maintenance respiration for the rapid increase in the culm biomass.

The presence of starch in only older parenchyma suggests the older culms store excess photosynthate as starch. Previous studies have shown that even 12-year-old culms of Phyllostachys viridiglaucescen can store starch (Liese and Weiner, 1996). Magel et al. (2005) indicated a relocation of carbon toward another growing culms during the growing period in Sasa palmata culms. Younger culms would consume the starch for metabolic processes, and have no starch stored in their cells as suggested by Liese and Weiner (1996).

This study showed the large difference in the respiratory mechanism between bamboo and trees. Factors used to explain variations in tree stem respiration, such as sapwood volume and stem diameter increment, cannot be applied to bamboo culms. In contrast, changes in the culm anatomical structure and physiological properties, such as cell wall thickness, culm wood density, and nitrogen content, which accompany the aging of culms, contribute to the large variation in respiration rate among individual culms. In this study, respiration measurements were conducted only before dawn. However, the green bamboo culms would conduct photosynthesis during daytime, and cause lowering the overall CO₂ efflux from the culm. Diurnal measurements of culm respiration and their correspondence with anatomical and morphological changes of culm will lead to a better understanding of the unique respiratory characteristics of bamboo culms.