Impact of extreme pre-monsoon drought on xylogenesis and intra-annual radial increments of two tree species in a tropical montane evergreen broad-leaved forest, southwest China

Study site

The study was conducted in a tropical montane evergreen broad-leaved forest (21.65°N, 101.47°E, 1473.6 m a.s.l.) located in Mengla County, Dai Autonomous Prefecture of Xishuangbanna in Yunnan Province, southwest China. The study site is dominated by evergreen tree species from the families Fagaceae, Euphorbiaceae, Theaceae and Lauraceae (Zhu et al. 2006). According to climate parameters derived from Xishuangbanna Station for Tropical Rain Forest Ecosystem Studies in Xishuangbanna Tropical Botanical Garden (21.92°N, 101.27°E, 580 m a.s.l.), which is about 36 km away from the study site, mean annual air temperature was 21.9 °C and mean annual precipitation was 1484 mm during the period 1963 to 2019. Influenced by the tropical monsoon climate, around 84% of annual rainfall occurs during the rainy summer monsoon season (May to October), and only about 16% occurs during the dry season (November to April) (see Fig. S1 available as Supplementary data at Tree Physiology Online).

Monitoring of xylem formation and stem radius increments

In this study, we selected five healthy trees of the species B. alnoides and S. wallichii with an average diameter at breast height (DBH) of 32.97 ± 2.43 cm and 37.15 ± 6.80 cm, respectively. Microcores with a diameter of 2 mm and a length of 15 mm contained the intact cambium were collected at breast height using a Trephor corer (Rossi et al. 2006) in biweekly intervals from January 2020 to April 2022 to investigate the dynamics of intra-annual xylem formation. These samples were extracted spirally after removing the outer bark with a distance of approximately 5 cm among each sampling position to avoid the possible influence of local wound reactions. After collection, the microcores were placed into micro-centrifuge tubes with 50% ethanol solution and stored at 4 °C to prevent cell deterioration. Microcores were then immersed in glycerol–ethanol–water solution (mixed ratio: 10:7:3) for 2 weeks to make the samples soften. Then, these samples were dehydrated in successive ethanol solutions (70%, 80%, 85%, 95%, 95% and 100%), cleared in D-limonene and finally embedded in paraffin blocks. Transverse sections with a thickness of 12 μm were cut using a rotary microtome (RM2245, Leica, Germany), dewaxed with D-limonene, stained with safranin red and astra blue, and mounted on glass slides with Eukitt mounting medium (Sigma, Germany). Finally, these sections were analyzed under polarized light using a microscope equipped with digital camera (DM2500, Leica, Germany). After staining, cambial cells and enlarging cells were stained in blue, wall-thickening cells were stained partly blue and partly red, as well as mature and lignified xylem cells were stained in red. Cambial cells were regularly arranged in close proximity, the diameter of enlarging cells nearly doubled that of cambial cells and wall thickening cells exhibited brightness under polarized light (Fig. 1). Ring boundaries of B. alnoides are characterized by two to three rows of tangential parenchyma cells (Fig. 1a), while ring boundaries in S. wallichii are indistinct. Consequently, it was unfeasible to determine the annual ring width of S. wallichii. We measured the thickness of the cambial zone, the enlarging zone and the wall-thickening zone in both study species, as well as the thickness of the mature zone and the previous year’s ring of B. alnoides in three radial rows per picture using ImageJ software (https://imagej.nih.gov/ij/).

Open multimedia modal

Four individuals of each species among the sampling microcore trees were equipped with high-resolution band dendrometers (DRL26C, EMS Brno, Czech Republic) at breast height. Stem circumference variations were automatically recorded at 10-min intervals at a resolution of 2 μm with the in-built datalogger from January 2020 to December 2021. In order to minimize the impact of shrinking and swelling within the bark, the outermost bark layers were carefully removed before installation of the dendrometer.

Measurements of climatic factors

We installed a microclimate station to record climate variables at 10-min intervals at the study site. However, many missing data occurred because of power failure of datalogger in 2020 (see Fig. S2 available as Supplementary data at Tree Physiology Online). Meanwhile, we obtained daily air temperature, precipitation, and relative humidity (RH) from the meteorological station in Xishuangbanna Station for Tropical Rain Forest Ecosystem Studies, Xishuangbanna Tropical Botanical Garden (XTBG). We found close correlations of climatic factors between the two sites (see Fig. S3 available as Supplementary data at Tree Physiology Online), so we used climate data from XTBG for further analyses. Vapor pressure deficit (VPD) was calculated from air temperature and RH using the RHtoVPD function in R package ‘plantecophys’ (Duursma 2015).

Data analysis

Different xylem formation stages can be accurately represented by generalized additive models (GAMs) at an intra-annual scale (Cuny et al. 2013). Hence, this study describes intra-annual changes of different xylogenesis phases for two studied species as day of year (DOY) by fitting with GAMs in R package ‘mgcv’ (Wood 2011). Because of variable annual ring width along the stem circumference, the thickness of cambial zone, enlarging zone, wall-thickening zone and mature zone of B. alnoides were standardized by the previous year’s ring width (Rossi et al. 2003). In addition, we defined the onset and cessation of xylogenesis as the start of the cell enlarging and the ending of the wall thickening, respectively. The growth duration was the interval between the onset and cessation of xylogenesis. Differences in xylem growth parameters (onset day, cessation day, duration and annual xylem growth) between two species and two years were compared by a t-test.

To calculate daily radial growth, we transformed the dynamics of stem circumference monitored by high-resolution dendrometer into variations of the stem radius. The stem radial variation data were corrected with temperatures monitored by the dendrometer logger using the proc_L1 function in the R package ‘treenetproc’ (Knüsel et al. 2021). The corrected stem radial variations data were further employed with the zero-growth model to extract irreversible radial growth caused by the formation of new xylem and phloem cells, and reversible shrinkage and expansion of the stem because of tree water deficit and resaturation with the R package ‘treenetproc’ (Zweifel et al. 2016^; Knüsel et al. 2021). The start and end dates of radial increments for each tree were determined as the dates when stem radius variations reached 5% and 95% of total annual radial change, respectively (van der Maaten et al. 2018). Growth duration and annual radial increments were the time periods between onset and cessation of radial variations. Differences in radial increments parameters between two years and two species were tested with a t-test.

To reveal the effects of various climatic variables on wood formation and radial increments, we analyzed the relationships between weekly mean climate factors (except for weekly sum of precipitation) prior to the sampling and the thickness of enlarging zone or wall-thickening zone, as well as the correlations between weekly cumulative radial increments and concurrent climatic factors. We compared the impact of species and years on the enlarging zone width, the wall-thickening zone width and weekly radial increments (excluding zero values) by building linear mixed-effect models with individual trees as random effects and years or species as fixed effects (Chen et al. 2022). Corrected Akaike’s Information Criterion (AICc) was calculated with AICc function in R package ‘MuMIn’ to evaluate the models (Kamil 2022). The models were significantly different with AICc increments (ΔAICc) above 2. Since radial increments were significantly different in species but not years (Table 1), for the two species, we built linear mixed-effect models with the enlarging zone thickness, wall-thickening zone thickness or weekly radical increments (excluding zero values) as dependent variables, scaled climatic variables as fixed effects and individual trees nested within two years as random effects, respectively. The model is expressed as follows:

Open multimedia modal

where Y represents the enlarging zone width, wall-thickening zone width or weekly radical increments of each species, X is the climate factor (precipitation, relative humidity, vapor pressure deficit, minimum air temperature, mean air temperature or maximum air temperature), α and β are the intercept and coefficient of fixed effect, and δ and ε are the variation from random effect and residual error term. The dependent variables data was log-transformed to achieve the normal distribution. All models were conducted by the lme function in the R package ‘nlme’ (Pinheiro et al. 2020). In addition, we calculated the proportion of cumulative growth on annual radial increments under different intervals of climatic factors (Tumajer et al. 2022). Subsequent data analysis and plotting were performed in R 4.0.0 (R Core Team 2020).

Climatic conditions

The mean air temperatures were 23 °C and 22.7 °C, as well as annual precipitation was 1206 mm and 1272 mm in 2020 and 2021, respectively (Fig. 2a and b). The climate conditions were warmer and drier than the long-term average (see Fig. S4 available as Supplementary data at Tree Physiology Online). The mean air temperature was higher, but precipitation was less from January to March in 2020 than in 2021 (Fig. 2a and b). The RH was lower, but VPD was higher between January and July in 2020 than 2021 (Fig. 2c and d). The year 2020 showed abnormal arid conditions during the pre-monsoon season for both RH and VPD compared with the long-term mean (see Fig. S4 available as Supplementary data at Tree Physiology Online). Therefore, the year 2020 had a significantly drier and warmer pre-monsoon season compared with 2021 (see Table S1 available as Supplementary data at Tree Physiology Online).

Open multimedia modal

Intra-annual xylem formation

The dynamics of wood formation in B. alnoides and S. wallichii varied dramatically between 2020 and 2021 (Fig. 3). In 2020, cell enlargement in B. alnoides started in April (DOY: 117 ± 14) and ceased in August (DOY: 238 ± 31), whereas it started in January (DOY: 17 ± 13) and stopped in September (DOY: 246 ± 22) in 2021 (Fig. 3c). The first wall thickening cells appeared in May (DOY: 135 ± 34) and January (DOY: 27 ± 9) in B. alnoides, whereas cell wall thickening ceased in September (DOY: 259 ± 26) and October (DOY: 274 ± 34) in 2020 and 2021, respectively (Fig. 3e). The enlarging zone in S. wallichii appeared in May (DOY: 146 ± 23) and stopped in November (DOY: 325 ± 33) in 2020, while it occurred in April (DOY: 117 ± 12) and ceased in December (DOY: 364 ± 8) in 2021 (Fig. 3d). However, the wall-thickening zone in S. wallichii ceased in early February (DOY: 34 ± 20) and later February (DOY: 50 ± 18) in 2021 and 2022, respectively (Fig. 3f).

Open multimedia modal

The xylogenesis of both B. alnoides (P < 0.01) and S. wallichii (P = 0.04) started significantly later in 2020 than in 2021 (Fig. 4a). Furthermore, both onset (P₂₀₂₀ = 0.04; P₂₀₂₁ < 0.01) and cessation (P₂₀₂₀ < 0.01; P₂₀₂₁ < 0.01) of xylem formation were significantly earlier for B. alnoides than S. wallichii during the two years (Fig. 4a and b). The xylem growth duration of B. alnoides was significantly longer in 2021 than in 2020 (P < 0.01) (Fig. 4c). Betula alnoides grew significantly shorter than S. wallichii in 2020 (P < 0.01), while there was no significant difference in 2021 (P = 0.13) (Fig. 4c). In addition, annual xylem growth for B. alnoides was significantly lower in 2020 than in 2021 (P = 0.02) (Fig. 4d).

Open multimedia modal

Intra-annual radial increments

Radial increments of B. alnoides monitored by dendrometers mainly occurred from April (DOY: 115 ± 4) to October (DOY: 283 ± 48) in 2020 and from January (DOY: 17 ± 13) to October (DOY: 274 ± 34) in 2021 (Fig. 5a and e). A distinct tree water deficit of B. alnoides was observed between March and April in 2020, but not in 2021 (Fig. 5c). For S. wallichii, radial increments appeared during May (DOY: 138 ± 13) to late September (DOY: 268 ± 8) in 2020 and during April (DOY: 110 ± 22) to the end of October (DOY: 295 ± 3) in 2021 (Fig. 5b and f). Schima wallichii showed a strong tree water deficit from March to April in both years; however, being much more pronounced in 2020 than 2021 (Fig. 5d).

Open multimedia modal

The onset of radial increments of B. alnoides was significantly later in 2020 than in 2021 (P < 0.01), while radial increments started significantly earlier in B. alnoides than S. wallichii during 2020 (P = 0.03) and 2021 (P < 0.01) (Fig. 6a). Growth cessation was significantly later (P < 0.01) and growth duration was significantly longer (P = 0.01) for S. wallichii in 2021 than in 2020 (Fig. 6b and c). The daily maximum growth rate (P = 0.01) and annual radial increments (P = 0.049) of B. alnoides were significantly lower in 2020 than in 2021 (Fig. 6d and e). Further, B. alnoides in 2020 showed significantly higher daily maximum growth rates compared with S. wallichii in 2021 (P = 0.01) (Fig. 6d).

Open multimedia modal

Effects of climate variables on xylogenesis and radial increments

Minimum and mean air temperature had a positive effect on the wall-thickening zone width of both species and the enlarging zone width of S. wallichii (Fig. 7). In addition, the thickness of enlarging zone and wall-thickening zone for S. wallichii was positively related to relative humidity and negatively related to VPD (Fig. 7).

Open multimedia modal

Weekly radial increments of both B. alnoides and S. wallichii were significantly and positively correlated with precipitation and relative humidity, but negatively correlated with VPD and maximum air temperature (Fig. 8). The mean air temperatures showed significant negative effects on the radial increments of B. alnoides (Fig. 8a). However, the radial increments of S. wallichii were positively correlated with minimum and mean air temperatures (Fig. 8b).

Open multimedia modal

Compared with S. wallichii, the radial increment of B. alnoides occurred within a wider range of temperatures (13 to 28 °C) and precipitation (0 to 120 mm) (Fig. 9a and d). However, the radial increments of both species were highest at 24 to 28 °C or VPD below 0.8 kPa (Fig. 9a and b). The two species almost ceased their radial increments when VPD reached a level higher than 1.2 kPa or RH was below 70% (Fig. 9b and c). In addition, the radial increments of S. wallichii nearly stopped during the period when there was no precipitation (Fig. 9d).

Open multimedia modal

Comparison of microcoring and dendrometer methods

There were non-significant differences in the dates of growth onset between the microcoring and dendrometer methods for both species and both years. However, the dates of growth cessation and growth duration of S. wallichii determined by microcoring method were approximately 100 days later than those of dendrometer measurements in 2020 and 2021 (Table 2). The annual radial increments of B. alnoides determined by dendrometer method were around 0.5 mm and 1 mm higher than the xylem growth detected by microcoring method in 2020 and 2021, respectively (Table 2).

Open multimedia modal

Intra-annual xylem formation and radial increments

We found that both xylogenesis and radial increments of the two studied species in 2020 started three months (B. alnoides) and one month (S. wallichii) later compared with 2021 (Figs 3 and 5). Delayed growth onset may result from higher temperature, lower RH and higher VPD during the pre-monsoon season in 2020 (Fig. 2). Increased VPD could potentially reduce vegetation growth with global warming (Yuan et al. 2019). The high temperature anomaly during the pre-monsoon season in 2020 (see Fig. S4, Table S1 available as Supplementary data at Tree Physiology Online) may also intensify the effect of drought through altering plant carbon metabolism (Wang and Wang 2023). Similar to our results, the xylogenesis onset of Pinus ponderosa located in the Mojave Desert Mountain delayed two months in a year with hyper-arid spring compared with the year with wetter spring conditions (Ziaco et al. 2018). Moisture conditions in the pre-monsoon season are an important driver for the start of cambial activity. Rahman et al. (2022) observed that frequent watering during the pre-monsoon season reactivated cambial cell division of Samanea saman in subtropical Bangladesh. Delayed xylem formation in response to lower pre-monsoon moisture availability could be a strategy against hydraulic failure under dry conditions (Ren et al. 2015).

Both wood formation and radial increments of B. alnoides were significantly lower in 2020 than in 2021 (Figs 4 and 6). Annual radial increments of S. wallichii were also lower in 2020 than 2021, but statistically non-significant (Fig. 6). These differences between the two species might be related to their different intra-annual growth dynamics. Betula alnoides started growing three months earlier than S. wallichii in 2021. Meanwhile, we observed that B. alnoides began budburst and flushed new leaves between December and January, but S. wallichii experienced peak leaf flushing after leaf fall in March. In 2020, B. alnoides experienced extreme drought in the early-growing season, however the growth initiation of S. wallichii occurred after the drought event. Hence, the timing of drought events may determine the extent to which radial growth is constrained. Stem radial growth sharply reduced during the early-growing season drought in Eastern North America (D'Orangeville et al. 2018), while annual radial growth reduction was not observed under the European heatwave during the late-growing season in 2018 (Salómon et al. 2022). In addition, deciduous tree species have lower xylem cavitation resistance than evergreen tree species (Fu et al. 2012), which may result in greater sensitivity to water stress for the deciduous species B. alnoides.

Effects of climatic factors on xylem formation and radial increments

We observed that the thickness of enlarging zone and wall-thickening zone for S. wallichii showed a positive response to RH, but showed a negative response to VPD (Fig. 7). Meanwhile, precipitation and RH were positively correlated with radial increments, while VPD negatively impacted the radial increments of two species (Fig. 8), indicating that stem radial increments were primarily moisture-limited in the tropical montane evergreen broad-leaved forest. Both species almost ceased their growth when no precipitation, RH below 70% or a VPD over 1.2 kPa (Fig. 9). Conversely, tree water deficits for both species occurred during the dry season (Fig. 5), especially during the warmer and drier period in 2020. Similarly, the radial growth of Pinus kesiya var. langbianensis coincided with precipitation, and tree water deficit occurred in drought in a subtropical forest in Yunnan at mid-elevations (Fan et al. 2019). In a drier site, most tree species showed stronger stem shrinkage in tropical forest (Mendivelso et al. 2016). Zweifel et al. (2021) found that stem radial growth primarily depends on water conditions and secondarily on carbon allocation. That is caused by the fact that turgor-driven cell enlarging is mostly affected by water availability, whereas the process of cell-wall thickening may rely on the availability of non-structural carbohydrates (De Micco et al. 2019). Water availability is the most important factor explaining the production of xylem cells in black spruce saplings in a greenhouse (Deslauriers et al. 2016). Correspondingly, the width of the cambial zone was positively related to monthly precipitation in evergreen moist rain forests of Ivory Coast and Thailand (Dié et al. 2012^; Pumijumnong et al. 2021).

The radial increments of both species were negatively correlated to maximum air temperature, and B. alnoides was also restricted by mean air temperature (Fig. 8). The negative effect of temperature on radial increments is mainly linked with a higher VPD, which limited stem growth by reducing turgor pressure (Peters et al. 2023). Mendivelso et al. (2016) also revealed that high air temperature probably enhancing evapotranspiration rates and negatively affected the radial increments in two tropical dry forests. However, the wall-thickening zone width of both species was positively correlated with minimum and mean air temperature (Fig. 7). The radial increments as well as the thickness of enlarging zone for S. wallichii were positively related to minimum and mean air temperature (Figs 7 and 8). Similarly, Bräuning et al. (2009) found that stem radial growth of Cedrela montana in a humid mountain rainforest in Ecuador was positively related to temperature from January to April. Radial growth of Quercus faginea was limited by temperature in spring but limited by precipitation in summer in central Portugal (Vieira et al. 2022).

Though the radial increments of B. alnoides and S. wallichii responded differently to temperature, both species showed peak of growth at 24 to 28 °C (Fig. 9). Stem radial growth of three temperate broadleaved tree species peaked at 12 to 18 °C in northeastern Germany (Tumajer et al. 2022). Maximum xylem cell production of beech occurred at around 16 °C in Slovenia (Prislan et al. 2013). Moreover, balsam fir started radial growth above 9 to 10 °C in a cold and humid environment (Oogathoo et al. 2023). The temperatures during radial growth occurrence in these temperate and boreal sites were lower than those of our study.

Comparison of microcoring and dendrometer methods

In our study, xylem formation and radial increments were simultaneously monitored by applying microcoring and dendrometer measurements. The annual radial increments obtained from dendrometer were higher compared with microcoring method (Table 2), which was consistent with the study by Kaewmano et al. (2022). Dendrometer measurements included phloem growth, though water-driven radius fluctuations were minimized with the zero-growth model (Zweifel et al. 2016). Moreover, the annual radial increments recorded by dendrometer derived from stem circumference, whereas ring width from microcoring captured radial growth in only one position (Debel et al. 2024). Consequently, both approaches inevitably show differences in annual radial growth.

Given that the wall thickening period failed to increase the number of xylem cells, growth cessation determined by microcoring data was later than dendrometer method. Meanwhile, stem radial variations were particularly susceptible to water fluctuations when the xylem growth rate was low (Deslauriers et al. 2007^; Stangler et al. 2021), so the growth cessation dates monitored by dendrometer method likely deviated from the microcoring method, especially in the late growing season. Correspondingly, we found that stem radial growth of S. wallichii derived from dendrometer method ceased significantly earlier in 2020 than 2021 (Fig. 6), as the VPD in 2020 was higher than 2021 in September (Fig. 2). Similar to our results, Kaewmano et al. (2022) showed that the date of growth cessation monitored by dendrometers was one month earlier than that of the microcoring method. On the contrary, Debel et al. (2024) found that the growth cessation dates derived from microcoring measurements were earlier than those from dendrometer data, resulting from the fact that stem swelling increased the radial increments during the late-growing season. For coniferous trees, the application of an absolute threshold from dendrometer data, such as concrete growth rate, might be precise and unbiased for growth cessation in the southern Black Forest (Miller et al. 2022). However, it is unknown whether the absolute threshold value is fit for the identification of growth ended date in tropical forests.

Compared with microcoring, dendrometer measurements are suitable for long-term monitoring but are unable to reveal the accurate growth onset and cessation dates. Hence, models combining microcoring data with high-resolution dendrometer data are required to accurately estimate stem growth parameters. Under climate change, the frequency of drought events has gradually increased in Yunnan province during the past decades (Zhao et al. 2023). Pre-monsoon drought events caused significant growth reductions in B. alnoides but not in S. wallichii in 2020. Therefore, we should pay more attention to the earlier growth onset tree species, which are more vulnerable to extreme pre-monsoon drought events.

Acknowledgments

We thank the Xishuangbanna Station for Tropical Rain Forest Ecosystem Studies for providing the underlying climate data used in this study. We thank the Xishuangbanna National Nature Reserve Administration Bureau for providing logistic support during field work of the present study. We also thank Ma Hong for his assistance on field sampling.