Beijing experiences a typical warm temperate semi-humid continental monsoon climate, characterized by hot, rainy summers and cold, dry winters with wind and minimal snowfall. Extreme minimum winter temperatures generally range from -14°C to -20°C. This experiment was conducted in a P. vivax f. aureocaulis forest at the China National Botanical Garden (North Garden) in Beijing. Three plots measuring 10 × 10 m² were established for the study, which was conducted from December 2021 to March 2022. The illumination (lux), air temperature (°C), and relative humidity (%) were continuously monitored and recorded using a microclimate meter (Pocket Weather Tracker 4000). The vapor pressure deficit (VPD) was calculated based on the relative humidity (RH) and air temperature (Ta). The average plant height was between 8 and 10 m, and the diameter at breast height was between 7 and 8 cm.

New leaves from 20 bamboo plants, each measuring 8–10 m in height and having a ground diameter of 4–6 cm, were selected for photosynthesis determination. Three plants were sampled each month to measure photosynthetic parameters. Ten-millimeter segments were collected from the 5th, 10th, 15th, and 20th internodes of the culm samples and immersed in FAA (formaldehyde: acetic acid: ethanol = 5%:5%:90%) fixation solution. Half of the samples were analyzed using scanning electron microscopy (SEM), whereas the other half were assessed for NSCs one week later. The 10th culm segment was selected, and a 10 cm culm, 3 cm primary twig, and 1 cm petiole were collected. After sampling, both ends of each sample were immediately sealed with hot-melt glue, placed in an icebox, and transported to the computed tomography (CT) room for scanning. Ten to 15 culm and rhizome segments from the 15th to 20th internodes were selected for conductivity loss rate determination, with each segment measuring 10–13 cm in length.

After sampling, both ends of each sample were immediately sealed with hot-melt glue, placed in an icebox, and sent to a CT room for scanning. The Skyscan 1172 micro-CT scanner used in this study (Bruker Corporation, Kontich, Belgium) has a significantly lower inlet dose rate (<1 mGy s^-1) than the previously reported dose rate for micro-CT scans at the beamline of synchrotron radiation facilities (47 mGy s^-1) (Petruzzellis et al., 2018).

Based on preliminary experiments, the scanning parameters were adjusted to minimize the ionizing radiation dose while maintaining an adequate contrast and resolution. The final scanning parameters were as follows: source voltage, 49 keV; source current, 200 μA; exposure time, 220 ms; rotation step, 0°–180° at 0.4° intervals; image pixel size, 3.92 μm, and total scanning time, 15 minutes. The length of each branch was measured at 5.2 mm, and 499 two-dimensional projection images, each with a resolution of 2000 × 1332 pixels, were obtained.

After each micro-CT scan, 499 X-ray projections were obtained and reconstructed using the NRecon software (Bruker Corporation, Kontich, Belgium) to create three-dimensional (3D) images. For each branch segment, the spatial structure of the xylem was aligned with a series of 3D images obtained from repeated scans using the “co-registration” function in Data Viewer software (Bruker Corporation, Kontich, Belgium). A synchronized 3D image with a longitudinal extent of 1 mm (comprising 257 cross-sectional images) was selected, positioned 2 mm below the short line mark in each organ segment, and converted to eight bits using CT An software. Images were processed using ImageJ software (Bruker Corporation, Kontich, Belgium). The tubes and vascular bundles were isolated with water, and a distinction was made between the tubes and those filled with water from the embolized vascular bundles.

The embolization ratio was calculated using the following formula:

Ten to 15 culms from the 15th to 20th segments, including internodes, nodes, and five rhizome segments from the three bamboo samples, were washed to remove the soil and subsequently cut in deionized water. Samples were collected before dawn to assess water conductivity. The samples were then placed in a 0.1% potassium chloride solution, wrapped in a black plastic bag, and promptly transported to the laboratory to measure the actual water hydraulic conductivity. The hydraulic conductivity of the culm and rhizome segments was measured under a specified pressure.

The design of the bamboo culm water-conductivity device is illustrated in Figure 1 . Multiple bamboo culms, each containing nodes, were connected using hoses and pipe clamps were used to fix the hoses to the bamboo culms, thereby preventing water leakage at the connection points. A potassium chloride solution was poured into a water storage container to a height of 1.3 to 1.7 meters, thereby establishing a pressure of 1.3 to 1.7 kilopascals within the container. By opening the flow control valve, the potassium chloride solution flowed through the outlet pipe and was directed toward the end of the bamboo culm. At this juncture, the water potential at the end of the bamboo culm initiated the movement of water through the culm.

The other end of the culm was then placed in a water-receiving container. The actual water conductivity of the bamboo culm was determined by measuring the water intake in the container.

where K _x is the xylem hydraulic conductivity coefficient (kg·m⁻¹·s⁻¹·MPa⁻¹), A _x is the cross-sectional area (m²), Δh is the water potential pressure (Mpa⁻¹), Q is the mass of aqueous solution flowing out per second (kg·s^-1) and L is the stalk length (m) of the sample. High pressure ranges from 0.13 to 0.15 Mpa whereas low pressure ranges from 0.03 to 0.05 Mpa.

The highest and lowest water conductivities of all samples were measured using a self-guided water conductivity device. The samples were then placed in 10 ml centrifuge tubes and centrifuged (Luxiangyi High-Speed Centrifuge, Shanghai, China) for 2 minutes at a rotational speed of 300 RPM. The water conductivity ( Figure 1 ) and water potential (PSΨPRO Water Potential Measurement System, Hansha Scientific Instrument, England) of the samples were determined by centrifugation. The samples were centrifuged 15–20 times to obtain the water conductivity gradient and the corresponding water potential.

PLC values at a corresponding tension were calculated as 100 × (K_max −K_h)/K_max. The Fitplc package in R was used to analyze the effect of plant leaf water potential on the PLC. The Weibull model was used to fit all data, and bootstrapping was conducted 50 times to estimate the uncertainty of the parameters. Vulnerability curves (VCs) generated by these methods were used to compare the percentage loss of conductivity of bamboo under varying water potentials.

The calculation formula is as follows:

where b and c are parameters and T is the xylem tension (equal to negative xylem water potential).

The VCs of P. vivax f. aureocaulis were fitted using the dual-Weibull equation.

where W1 and W2 represent the two Weibull curves, β denotes the maximum PLC for W1, whereas b₁ and c₁ are constants for W1, and b₂ and c₂ are constants for W2. The xylem pressure corresponding to 50% PLC (P50), 12% PLC (P12), and 88% PLC (P88) was calculated.

The extraction and determination of NSCs was performed according to the method described by Zhong et al. (2013). Approximately 0.0100 g of crushed dry sample was weighed and placed in a test tube. Subsequently, 3 mL of distilled water was added and the mixture was incubated in a water bath at 80°C for 30 minutes. After cooling, the samples were centrifuged at 10,000 g for 10 minutes. The resulting supernatant was used to determine the soluble sugars. A 200 µL aliquot of the supernatant was mixed with 2 mL of 0.4% anthrone-concentrated sulfuric acid. The mixture was thoroughly shaken, placed in a boiling water bath for 5 minutes, cooled to room temperature, and the absorbance was measured at 640 nm. A standard curve was constructed using sucrose.

For photosynthetic measurements, nine healthy leaves were selected from each bamboo plant, with one leaf taken from each cardinal direction at chest height. The net photosynthetic rate (P _n) and transpiration rate (T _r) were measured at three-time intervals, 9:00–10:00 AM, 12:00–1:00 PM, and 3:00–4:00 PM, representing the early, middle, and late parts of the day, respectively. Measurements were conducted using a photosynthesis determination system (Li-COR 6400, USA). The light intensity, CO₂ concentration, and temperature were maintained at 800 mol/m²/s, 400 μmol/mol, and 25°C, respectively.

Culm samples measuring 5, 10, 15, and 20 mm were collected and immersed in the FAA fixation solution. The area of the vascular bundles, the thickness of the parenchyma cells, pitted areas of the conduits and parenchyma cells, and the diameter of the pitted membranes were observed using SEM. The shooting face of each sample was affixed to a conductive adhesive and impurities on the sample surface were removed using an earbud. The samples were then placed in the equipment for gold sputtering, which lasted 90 seconds. Once the gold sputtering was completed, the sample stage was removed and placed in the SEM equipment for vacuum extraction. After vacuum extraction, imaging was performed using a Hitachi SU8010 instrument at 15 kV. Multiple images are presented in Figures 2A–H .

Data are presented as the mean ± standard deviation. Statistical analyses were conducted using SPSS version 26.0 for one-way analysis of variance (ANOVA), least significant difference tests, and multiple comparisons (α = 0.05). SPSS version 26.0 and Excel 2016 were used to create linear models and correlation graphs.

From December 2021 to March 2022, the average VPD, average temperature, and average humidity at the test site during the four-month wintering period were 0.42 ± 0.28 kPa, -0.87 ± 5.50°C, and 35.99 ± 20.66%, respectively. The highest temperature recorded during the test period was 17.9°C on 27 February, whereas the lowest temperature was -13.8°C on 15 February. The average underground soil temperature was approximately 3.8 ± 2.26°C, and the air temperature remained<0°C for 45 days ( Figure 3 ).

From December 2021 to March 2022, significant fluctuations were observed in the culm embolization ratio. The degree of embolism was relatively high during the winter (December to January) and significantly decreased during the spring (February to March). Notably, the ratio of petiole embolisms was consistently lower than that of twigs at each time point (P< 0.05). The maximum embolization ratio recorded in January was 99.44%, whereas the minimum petiole embolization ratio observed in March was 54% ( Figures 4A–C ). Concurrently, leaf accumulation increased gradually over time, accompanied by a corresponding increase in the leaf water content ( Figure 4D ).

There is a direct relationship between xylem embolization and the xylem water potential. When the water potential falls below a certain threshold, the degree of embolization increases rapidly until it reaches its maximum value. A higher threshold corresponds to a greater vulnerability to xylem embolism. The farther to the right of the vertical segment, the higher the water potential threshold; similarly, the higher the distance along the straight segment, the greater the maximum value of embolization and the increased vulnerability to xylem embolism for the corresponding tree species. As illustrated in Figure 5 , PLC in culms increased rapidly in January and February, reaching 50% PLC at -0.3 MPa and -0.4 MPa, respectively. This indicated that the rate of water conductivity loss was more pronounced during these months. In addition, the PLC of culms at -1.6 MPa and -1.4 MPa in December and March also reached 50%. In this study, the vulnerability of culms to embolism was the highest in February.

The NSCs were lower in January across the various segments. The NSCs in segments 10th and 15th were significantly higher in March than in other months, reaching a maximum value of 160.27 mg/g. NSCs in the underground rhizomes at distance of 15 and 30 cm from the culm were significantly higher in March than in December. In contrast, NSC content in the roots increased at a distance of 50 cm from culm, but this difference was not statistically significant ( Figures 6A, E ).

The range of soluble sugar content in the culms and rhizomes varied from 1.24 to 25.7 mg/g. The soluble sugar content in the rhizomes closer to the culms changed significantly from December to March during the overwintering period, whereas no significant change was observed in the rhizomes farther away. ( Figures 6B, F ).

Starch serves as the primary source of NSCs in P. vivax f. aureocaulis. The starch content of underground rhizomes exhibited a significant upward trend from winter to spring. Although there was no significant difference in starch content among the different months in the 5th segment of the culm, the starch content in the 10th and 15th segments of the culms increased significantly in March ( Figures 6C, G ). In March, the starch content in the 10th segment reached 169.22 mg/g.

Both T _r and P _n exhibited significant diurnal and seasonal variation. Notably, midday T _r was negative in January ( Figure 6D ). P _n levels were higher in December and March, and lower in January and February. In both December and March, there were eight days when the midday temperature exceeded 10°C, coinciding with relatively high P _n levels. Conversely, the P _n values remained negative throughout February ( Figure 6H ).

P _n significantly influenced the changes in NSCs, soluble sugars, and starch in bamboo culms ( Figures 7A–C ). The number of chloroplasts in the bamboo leaves decreased in December. In January, the proportion of spongy tissue increased, whereas the proportions of palisade tissue and spindle cells decreased significantly, indicating decreased gas exchange. In February, the number of spindle cells increased, palisade cells became more distinct from the spongy tissue, and the volume of spongy tissue decreased. Multiple instances of vascular bundle gas ejection were observed in March ( Figures 7D–G ). During winter, the degree of stomatal opening in both the lower and upper parts of the canopy was significantly greater in March than in December, January, and February. In addition, the middle part of the canopy exhibited a significantly higher degree of stomatal opening in January and February than in December and March ( Table 1 ).

Although the temperatures were low in December and January, the bundles and cells remained intact ( Figures 2A, B, E, F ). In February, the bundles and cells exhibited signs of fragmentation, accompanied by the presence of invading bodies ( Figures 2C, G ). By March, the bundles and cells returned to their normal state ( Figures 2D, H ). In addition, a significant number of starch granules appeared in the parenchyma cells during March ( Figure 2I , Table 2 ). Throughout the experiment, the area of the vascular bundles decreased as the number of segments increased, with the area of the aboveground vascular bundles being greater than that of the underground vascular bundles. The average, maximum, and minimum sizes of the vascular bundles in February were smaller than those observed during other periods, indicating that during the winter overwintering phase, vascular bundles contracted to adapt to low winter temperatures ( Figure 2I ).

As shown in Table 1 , the diameter of the short axis of the pit in the duct (D_dps), the diameter of the long axis of the pit in the duct (D_dpl), the number of pit holes in the parenchyma cell (N_pp), and parenchyma cell pit chamber area (A_pcp) were lower in January and February than in December and March, respectively. There were no significant differences in the vessel density (VD) or the number of pit holes in the duct (N_dp) during the overwintering period. The maximum vascular bundle area (V_max) in March was significantly greater than that in the other months ( Table 2 ).

Embolism increased significantly during winter due to persistent embolism caused by freeze-thaw cycles (Nardini and Salleo, 2000). In this study, the average temperature was -0.87°C, with the lowest recorded temperature of -10.6°C occurring in December 2021, specifically dropping to -10.6°C on 25 December. This indicated a rapid decrease to a higher level during this period. The average temperatures in January and February 2022 were -1.02°C and -2.06°C, with minimum temperatures of -9.6°C and -13.8°C, respectively. These data indicate a persistently low-temperature environment. However, the average temperature in March remained low at -1.02°C, with the lowest temperature dropping to -13.8°C on 15 March, which adversely affected embolism repair. Although the presence of soluble substances in the plant reduced the freezing point ( Figures 6B, F ), the minimum temperature was significantly lower than the freezing point ( Figure 3B ), preventing the restoration of water potential at night and hindering embolism repair.

In this study, the embolization ratio was in the following order: twig > culm > petiole. This variation was attributed to the high sensitivity of bamboo xylem ducts to cavitation induced by abiotic stress. Even under optimal soil moisture conditions and normal photosynthetic gas exchange, embolization can occur in xylem ducts, leading to water transport failure (Cochard et al., 1994). Interestingly, bamboo thrives under conditions of severe culm embolism, suggesting low water requirements during winter months. However, an increase in the number of dead leaves was noted as winter progressed, coinciding with an increase in leaf biomass ( Figure 4D ). This observation aligns with previous studies, indicating that plants can minimize the risk of embolism by sacrificing the smallest and most peripheral expendable parts, thereby protecting organs with high carbohydrate accumulation (Zimmermann, 1983).

To adapt to low temperatures, plants actively accumulate carbohydrates, reduce osmotic potential and freezing points, enhance the osmotic regulation capacity of their cells, prevent freezing induced by low temperatures and mitigate xylem embolism caused by freezing and thawing. In addition, they decrease the annual frequency and duration of freeze-thaw cycles (Charrier et al., 2015; Lintunen et al., 2017).

In December, NSCs and starch contents were lower in various parts of the rhizome and different segments of the culm. The soluble sugar content was higher in rhizomes located farther from the culm, and no significant changes were observed in the different segments of the culm ( Figures 6A–C, E–G ). This observation suggests that the freezing point can be lowered by maintaining non-structural carbon in living cells and xylem vascular bundles, thereby reducing the risk of freezing (Charrier et al., 2015). Although the chlorophyll content and photosynthetic capacity of the leaves decreased ( Figure 7D ), the anatomical structure of the cells remained intact, and starch granules were still observed ( Figures 2A, E ).

NSCs, starch, and soluble sugar contents were lower in various internodes in January ( Figures 6E–G ). However, the photosynthetic rate was higher when temperatures were relatively elevated ( Figure 6H ), indicating that increased photosynthetic rates supplied the resources necessary for the production of antifreeze substances, thereby enhancing cold resistance (Hamilton et al., 2016). Concurrently, the proportion of spongy tissue increased, improving aeration capacity ( Figure 7E ). Loosely arranged and irregularly shaped spongy tissue cells can enhance light scattering, extend the transmission path of light within the mesophyll, and maximize the absorption of limited available irradiance. Photosynthetic products are transported from the leaves to other organs for energy metabolism, osmotic regulation, and synthesis of defense compounds (Hartmann and Trumbore, 2016).

There were no significant differences in the levels of NSCs and starch between February and January, and soluble sugar concentrations were significantly increased in the 20th culm ( Figures 6E–G ). Photosynthesis remained negative throughout the day in February ( Figure 6H ), and soluble sugar levels were significantly positively correlated with P _n ( Figure 7C ). This correlation suggests that osmotic substances, such as hexose, are produced through photosynthesis and starch granule hydrolysis under prolonged low temperatures, which increases the concentration of cell fluids. Elevated soluble sugar concentrations play a crucial role in cold hardening, helping plants cope with low-temperature stress (Sicher, 2011; Sitnicka and Orzechowski, 2014).

In March, despite the low temperatures, the NSCs and starch content in various internodes of the culm and different positions of the rhizome significantly increased ( Figures 6A, C, E, G ). The parenchyma cells were filled with starch granules ( Figures 2D, H ), and P _n also showed a significant increase ( Figure 6H ). The soluble sugar content in the 10th section was significantly higher in March than in other months; however, the soluble sugar levels in the 5th, 15th, and 20th sections in March did not differ significantly from those in other months ( Figure 6F ). In addition, leaf structure, photosynthetic conversion capacity, and soluble sugar content were enhanced ( Figure 7G ). In contrast, embolization ratios of the culm, twig, and petiole decreased ( Figure 4B ). Culm embolism vulnerability was lower in the spring of the following year ( Figure 5D ), and high water conductivity facilitated more efficient water transport to the bamboo leaves. This resulted in increased stomatal opening ( Table 1 ) and higher transpiration and photosynthetic rates ( Figures 6D, H , 7G ) (Farquhar and Sharkey, 1982; Franks, 2006). The elevated concentration of soluble sugars causes sugar to transfer from the symplast to the exoplasm, specifically within the vessel, leading to increased culm pressure and the reflow of the embolized vessel (Mayr et al., 2014). Numerous studies have demonstrated a strong correlation between xylem hydraulic efficiency and C assimilation ability across various tree species (Cowan, 1982; Bacelar et al., 2007; Brodribb et al., 2007), a finding confirmed in this study.

The average, maximum, and minimum vascular bundles in February were smaller than those in other months. This observation indicated that the vascular bundles contracted during winter to acclimate to lower temperatures. During this period, the vascular bundles and cells exhibited signs of fragmentation in response to the invading bodies. In December and March, the culm ducts were larger ( Table 2 ), which facilitated starch storage; however, they also had a greater tube wall surface area, making them more susceptible to developing larger pores or cracks in the pitted membrane or having more imperfect cell walls, which could lead to embolization.

The vascular bundle area of bamboo exhibited a significant increase in March ( Table 2 ), and starch granules were observed within parenchyma cells ( Figures 2D, H ). Concurrently, the rate of loss of culm water conductivity and vulnerability decreased ( Figure 5D ). This suggests that a larger vessel diameter ( Table 2 ) enhances vessel porosity, thereby increasing the hydraulic transport capacity (Fichot et al., 2009). This improvement in efficiency facilitates water transport from the roots to the leaves (Yin et al., 2018), which, in turn, enhances the photosynthetic rate and C uptake (Bacelar et al., 2007; Franks, 2006). In addition, vascular bundle walls were thicker in December and March ( Table 2 ). A thicker vascular bundle wall can strengthen the vascular bundle, prevent deformation, and protect against pit-membrane damage (Hacke et al., 2001). This structural enhancement often helps prevent the formation of embolisms (Wheeler et al., 2005; Jansen et al., 2009; Li et al., 2016; Martina et al., 2021). Previous studies have demonstrated that the presence of pitted pores and parenchyma improves hydraulic connectivity between vascular bundles (Salleo et al., 2004; Sano et al., 2011; Barotto et al., 2016). The area of parenchyma cells and conduit pores increased in December and March ( Table 2 ), and there was a significant increase in starch granules ( Figures 2D, H ). This suggests that by enhancing the connections between vascular bundles, the cells surrounding the vascular bundles can function as bridges, facilitating water flow between the vascular bundles and contributing to embolic refilling (Barotto et al., 2016).