Introduction

In China, an extensive afforestation campaign was implemented in the 1980s, aiming to improve the ecological environment and restore the ecosystems heavily degraded due to anthropogenic activities (Ren et al. 2007). Thanks to this campaign, China is home to the largest artificial plantation area worldwide (Piao 2009). In tropical and subtropical regions of southern China, the conventional approach is to plant fast-growing alien tree species (e.g., Eucalyptus sp.) in resource-rich lowland areas (Ren et al. 2007) and highly stress-tolerant coniferous tree species in mountainous areas (Yu et al. 2020). However, these plantations are subject to many adverse ecological problems and are vulnerable to climate change (Pawson et al. 2013). In this light, plantation of native tree species is strongly recommended in recent regional afforestation projects, particularly native timber trees with high economic and ecological values (Zhang et al. 2000; Tang et al. 2019; Guan and Fan 2020). However, the key challenge is the selection of appropriate restoration tree species for diverse degraded areas from species-rich native forests (Wang et al. 2020).

Trait-based approach has been deemed an effective way to select species for restoration projects (Laughlin and Laughlin 2013; Wang et al. 2020). Xylem hydraulic traits are vital to identify ecological performance of tree species (Bittencourt et al. 2020). Hydraulic conductivity represents the water transport capacity of the xylem, and it is linked to the photosynthetic capacity (Brodribb and Feild 2000) and water balance of the whole-plant (He et al. 2020). Xylem cavitation resistance, estimated based on water potential at 50% and 88% loss of hydraulic conductivity using vulnerability curves (P50 and P88, respectively), reflects the threshold of hydraulic failure and relates to the drought tolerance strategies of plants (Choat et al. 2012; Levionnois et al. 2021). According to the “pit area” hypothesis, large vessel pit area increases the water transport efficiency, as well as the risk of embolism diffusion, thus impeding the safety of the xylem and resulting in a tradeoff between hydraulic efficiency and safety (Sperry 2003; Hacke et al. 2006). However, some studies have found that this hydraulic tradeoff is weak (Gleason et al. 2016; Liu et al. 2020), partly due to the fact that hydraulic conductivity and cavitation resistance evolve independently (Sanchez-Martinez et al. 2020). The hydraulic safety margin (HSM), defined as the difference between the minimum leaf water potential and P50 or P88, is widely used to assess the degree of hydraulic risks under drought (Meinzer et al. 2009; Choat et al. 2012; Tan et al. 2020). Many studies have demonstrated that there are significant relationships between hydraulic traits and tree growth (Zhang and Cao 2009; Fan et al. 2012; Liu et al. 2019). In general, fast-growing tree species exhibit lower wood density, larger vessel diameter, and higher hydraulic conductivity than slow-growing species (Hu et al. 2018; Zhao et al. 2021). However, in a recent study, McGregor et al. (2021) identified a close link between growth and drought tolerance traits in temperate broadleaf tree species under drought conditions.

Karst is a unique landform in tropical China, characterized by many exposed limestone and thin soil layer (Geekiyanage et al. 2019). The karst habitats are drought-prone and harbor many endemic species (Wang et al. 2017; Zhang et al. 2021a, b). Owing to long-term anthropogenic disturbances, severe land degradation and water resource depletion have been documented in many karst areas, resulting in substantial rock exposure called rocky desertification (Geekiyanage et al. 2019). Preliminary restoration trials have been conducted in tropical degraded karst areas, and several native economical timber tree species (e.g., Zenia insignis, Dalbergia odorifera, and Erythrophleum fordii) have been proposed as efficient afforestation species (Li et al. 2019a, b; Tang et al. 2019). A number of previous studies have proven that the hydraulic traits of native karst tree species play pivotal roles in explaining their drought adaptation (Zhu et al. 2017; Fu et al. 2019), spatial distributions (Matías et al. 2012; Zhang et al. 2021a, b), and response to extreme climate events (Tan et al. 2020). However, it is unclear whether hydraulic traits could be also used to clarify the performance of timber tree species planted in the dry karst desertification areas.

In the present study, we investigated the long-term growth rate of plant height and stem diameter at breast height in seven economical timber tree species, which have been used for afforestation in a tropical karst rocky desertification area over three decades. We measured their xylem anatomical traits, hydraulic conductivity, vulnerability to cavitation, and hydraulic safety margin. By examining hydraulic-growth relationships, we aimed to clarify which hydraulic characteristics can be used to predict the growth rate of these timber tree species. Because frequent episodes of extreme droughts threaten normal tree growth in the dry karst environments, plants’ growth is thus more likely driven by their drought tolerance. To this end, we hypothesized that the long-term growth of the tested timber tree species is more closely associated with their hydraulic safety (e.g., vulnerability to cavitation, safety margin) rather than their hydraulic efficiency (e.g., hydraulic conductivity).

Materials and methods

Study site and plant materials

The present study was conducted at the Daqingshan Karst Hill Arboretum (22°07′32″ N, 106°44′34″ E; 250 m a.s.l.) of the Experimental Center of Tropical Forestry (ECTF) in a typical karst region of southwestern Guangxi. The study site is influenced by the monsoon climate, resulting in distinct dry and wet seasons. The mean annual precipitation is ~ 1400 mm, and the mean annual temperature is 21 °C. The dry season extends from October to the following May and receives less than 20% of the annual precipitation. Because of the presence of a karst substrate with low water retention capacity, plants are often exposed to water stress throughout the year (Wan et al. 2016). During the last several decades, the frequency of extreme drought events at the study site has markedly increased due to climate change (Zhou et al. 2014; Zhang et al. 2021a, b).

The study site is located on a typical “peak-valley cluster” karst landform (Fig. 1), and the original forest is heavily degraded. In the 1980s, a reforestation project was undertaken at this site, aiming to selecting suitable tree species for karst restoration. An experimental trial was designed to plant precious tropical timber tree species in low-slope areas with approximately 60% rock exposure (Fig. 1). In the present study, seven timber tree species were selected, including the karst endemic tree species Excentrodendron tonkinense and Diplodiscus trichospermus, and tropical rainforest tree species Erythrophleum fordii, Dalbergia odorifera (a famous rosewood tree species), Zenia insignis, Cornus wilsoniana, and Adenanthera microsperma (Table 1). These tree species represent high economic values due to their highly valuable timber. All trait measurements were conducted between 2019 and 2020, during which an extreme drought event was recorded (Fig. 2). The crowns of several species (e.g. A. microsperma and E. tonkinense) were wilted under the extreme drought (personal observation).

Fig. 1
figure 1

a The typical “peak-valley cluster” landform in tropical karst region and original forests. b An example of karst degradation (rocky desertification). c The 38-year-old plantation of Dalbergia odorifera (a high-value endangered rosewood species) planted in the karst degradation area

Full size image
Table 1 The seven economical timber tree species investigated in this study
Full size table
Fig. 2
figure 2

Monthly precipitation (bars) and atmospheric temperature (average day and night temperature; circles) of 2019 (black) and 2020 (grey). The black arrow indicates a severe drought episode in the summer of 2020

Full size image

Growth rate

In the 1980s, approximately 1-m-tall seedlings of each species for the seven species were planted in the study area. However, the seedling numbers and initial planting density could not be obtained because of the missing of original records. All these timber tree species have successfully established at the site (Fig. 1). In 2020, we established a plot with an area of 400 m2 (20 m × 20 m) in each plantation. To avoid the influence of topography on traits (Zhang et al. 2021a, b), these seven plots had similar altitude (250–280 m) and slope (20°), and thus similar micro-site conditions. In each plot, we surveyed current plant density, and randomly selected 10 individuals to measure their canopy height (H; m) using an altimeter (Vertex IV, Haglof, Langsele, Sweden) and diameter at breast height (DBH; cm). The growth rate of DBH and height (DGR and HGR, respectively) was calculated by dividing the DBH and height values by the duration since the time of plantation (Table 1; Zhang and Cao 2009).

Hydraulic conductivity, Huber value, and xylem vulnerability curve

We collected 30–40 long branches (2–3 m) from at least five individuals per species in the early morning using a retractable long-reach pruner (maximum length is 20 m). The cut ends were wrapped with wet tissue papers, and the samples were enclosed in black plastic bags and immediately transported to the laboratory within 20 min. The maximum vessel length (MVL) for the seven species was measured using the air infiltration method (Cohen 2003; Zhang et al. 2021a, b). Five branches were used except for Zenia insignis (three replicates), because the height of this species was too high to harvest enough canopy branches. The distal end of the branch segment was perfused with air at 60 kPa pressure, and a 1-cm-long segment of the proximal end of the branch was cut under water until air bubbles continuously emerged from the cut ends. The length of the remaining branch plus 0.5 cm was considered the MVL. For each species, 10 branch segments with a length of approximately 110% MVL were used to measure the maximum hydraulic conductivity. Each segment was flushed with 20 mmol KCl solution at 0.1 MPa pressure for approximately 30 min to remove air embolism. Then, the maximum hydraulic conductivity (kh; kg m s−1 MPa−1) of the branch segment was measured using a hydraulic apparatus equipped with a digital liquid flow meter (Liqui-Flow L13-AAD-11-K-10S; Bronkhorst, Ruurlo, Netherlands), following the methodology described by Blackman et al. (2019). We calculated kh using the following equation: ki = Jv/(∆P/∆L), where Jv is the water flow rate (kg s−1) and ∆P/∆L is the pressure gradient causing the flow in the segment (MPa m−1). Sapwood specific hydraulic conductivity (ks; kg m−1 s−1 MPa−1) is equal to kh divided by sapwood cross-sectional area (As). Leaf specific hydraulic conductivity (kl; × 10–4 kg m−1 s−1 MPa−1) was calculated as the ratio of kh to the distal leaf area of the branch (Al). Huber value (HV, × 10–4) was calculated as As/Al.

The remaining long branches were used to construct vulnerability curves using the bench drying technique (Wheeler et al. 2013; Vaananen et al. 2020). Briefly, the branches were allowed to dehydrate on a bench to obtain different water potentials. During dehydration, leaf water potential was measured regularly. Upon reaching the desired water potential, the whole branch was wrapped in a plastic bag for 1–2 h to equilibrate. Then, the water potential of three leaves from the distal, middle, and basal of the branch was measured using a pressure chamber (PMS, Corvallis, Oregon, USA). We assumed the equilibrium of water potential of the whole branch when the difference in values was less than 0.2 MPa. The mean value of leaf water potential was equal to the xylem water potential (Ψxylem). Next, a branch segment was cut under water after removing a piece equivalent to the MVL length from the base, and the initial hydraulic conductivity (ki) and maximum hydraulic conductivity after flushing (kmax) were measured as described above. We ensured to follow a release-tension procedure under low xylem water potential, as described by Wheeler et al. (2013). The percentage loss of conductivity (PLC; %) was calculated using the following equation: PLC = 100 × (kmax–ki)/kmax. Based on a series of xylem water potential values and the corresponding PLC, the vulnerability curve for each species was fitted with a “sigmoidal” model using the “fitplc” function of the fitplc package (Duursma and Choat 2017) in R v. 4.0.3 (R Core Team 2020). The xylem water potential at 50% and 88% loss of hydraulic conductivity (P50 and P88, respectively) was calculated to estimate the vulnerability to drought-induced cavitation for each species (Choat et al. 2012).

Wood density and anatomy

For each species, five branch segments (from different individuals) with length of 5 cm and diameter of 0.8–1 cm were used for the measurements of wood density and xylem anatomy. After removing the bark and pith, the wood samples were immersed in water overnight for saturation, and the water saturated wood volume (cm−3) was measured using the water displaced method. After drying in an oven at 70 ℃ for 2 days, dry wood mass (DM; g) was measured, and wood density (WD; g cm−3) was calculated by dividing the DM by wood volume.

Wood cross-sections (20 μm thickness) were obtained using a sliding microtome (SM2010R, Leica, Nusslock, Germany) and subsequently stained with 0.5% safranine solution for 15 min and 12% Alcian blue solution for 30 s. Five fields were selected from each section, and photographs were obtained under a light microscope (DM2500, Leica, Wetzlar, Germany). The vessels were identified using the polygon selection tool in ImageJ (FIJI v.1.52n; NIH, Bethesda, MD, USA). The vessel density (VD; no. mm−2) was calculated by dividing the number of vessels by the images area at 10 × magnification. Hydraulically weighted vessel diameter (Dh; μm) was calculated using the following equation: Dh = (\(\sum {d^{4} } /N\))1/4, where d is the vessel diameter of the equivalent circle and N is the number of measured vessels. Vessel wall reinforcement was calculated as t divided by d, where t is the double-wall thickness between the neighboring vessel (Hacke et al. 2001).

Leaf minimum water potential and hydraulic safety margin

We measured midday leaf water potential using a PMS pressure chamber at 13:00–15:00 h in the early dry season (Ψnormal) and under an extreme drought event (Fig. 2; Ψextreme). Hydraulic safety margin (HSM, MPa) in the normal dry season was calculated as the difference between Ψnormal and the vulnerability to cavitation (P50 and P88, respectively): HSM50_normal = Ψnormal–P50, HSM88_normal = Ψnormal–P88. Hydraulic safety margin under the extreme drought was calculated as the difference between Ψextreme and the vulnerability to cavitation: HSM50_extreme = Ψextreme–P50, HSM88_extreme = Ψextreme–P88.

Statistical analysis

The Pearson correlations between hydraulic traits and growth rate were analyzed using “cor_person” function of the Hmisc package (Frank and Charles 2021), and the nonlinear regressions were fitted to the “exp3P” model using the “trendline” function of the basicTrendline package (Mei et al. 2020) in R v.4.0.3 (R Core Team 2020).

Results

Among the plantations of seven timber tree species, A. microsperma showed the lowest DGR (0.51 cm year−1), E. fordii showed the lowest HGR (0.50 m year−1), and Z. insignis showed both the highest HGR and DGR (0.93 m year−1 and 1.09 cm year−1, respectively) (Table 1). D. odorifera had the shortest MVL (35.0 cm), and Z. insignis had the longest value (104.0 cm). Hydraulic-related traits varied across species. A. microsperma had the highest values of P50 and P88 (− 1.36 MPa and − 2.21 MPa, respectively), Z. insignis and E. fordii showed the most negative P50 (− 2.98 MPa) and P88 (− 6.15 MPa), respectively. E. tonkinense showed the lowest ks and kl, and D. odorifera showed the highest values. HV ranged from 0.76 × 10–4 in E. fordii to 1.20 × 10–4 in C. wilsoniana, VD ranged from 21 no. mm−2 in Z. insignis to 130 no. mm−2 in C. wilsoniana, WD ranged from 0.43 g cm−3 in D. trichospermus to 0.73 g cm−3 in E. tonkinense (Figure S1 and Table S1).

For five out of the seven timber tree species, hydraulic safety margin decreased significantly under the extreme drought events relative to that in the normal dry season (Fig. 3). During the normal dry season, A. microsperma showed the smallest hydraulic safety margin (HSM50_normal and HSM88_normal were  − 0.93 and − 0.08 MPa, respectively), C. wilsoniana showed the largest HSM50_normal (2.28 MPa), and Z. insignis showed the largest HSM88_normal (4.73 MPa). Under extreme drought, A. microsperma showed the smallest HSM50_extreme and HSM88_extreme (− 1.04 and − 0.19 MPa, respectively), Z. insignis showed the highest HSM50_extreme (0.87 MPa), and E. fordii showed the largest HSM88_extreme (4.02 MPa) (Fig. 3 and Table S1).

Fig. 3
figure 3

Hydraulic safety margin (HSM) of the seven precious timber tree species under extreme drought event (grey) and in the normal dry season (black). The HSM is calculated based on xylem water potential at 50% and 88% loss of hydraulic conductivity (HSM50 and HSM88, respectively). Species abbreviations are shown in Table 1.The horizontal grey lines indicate HSM equals to zero. Data are means ± SE. The symbols within the barckets indicate significant level: ns, P > 0.05; *P < 0.05; **P < 0.01

Full size image

Across species, DGR was significantly correlated with P88 and HSM88_extreme, indicating that tree species with higher DGR were more resistant to cavitation and showed larger HSM under extreme drought. Similarly, HGR was positively correlated with P88 and HSM88_extreme except E. fordii many large lateral branches (Fig. 4). However, both HGR and DGR were not significantly correlated with P50, kl, ks, WD, HV, VD, and Dh (Fig. 4 and Table S2). It was found that ks was not significantly correlated with P50 and P88 (Figure S2 and Table S2), indicating a lack of tradeoff between hydraulic efficiency and safety.

Fig. 4
figure 4

Relationship between hydraulic-related traits and long-term growth rate. DGR DBH growth rate, HGR height growth rate; ks, sapwood specific hydraulic conductivity, WD wood density; P88, xylem water potential inducing 88% loss of hydraulic conductivity; HSM88_extreme, hydraulic safety margin under extreme drought condition. Species symbols are shown in Table 1. Data are means ± SE except for P88 and HSM88_extreme. Note that Erythrophleum fordii (■ within the circles) is an outlier and excluded from the correlation analysis in g–h. Significant level: ns, P > 0.05; *P < 0.05; **P < 0.01

Full size image

Discussion

We found that the same rosewood species (i.e., Dalbergia odorifera) was also planted in the nearby non-karst lowland areas, with DGR and HGR (0.80 m year−1 and 1.10 cm year−1, respectively; data provided by ECTF) two times higher than those reported in this study. This is because the harsh conditions in karst area (e.g., low water availability) may slow down the normal growth of the tree species originated from tropical rain forests.

Consistent with our hypothesis, the higher growth rate of timber tree species planted in the karst rocky desertification area was associated with their greater cavitation resistance and larger HSM under extreme drought (Fig. 4). These results are contrary to the previous findings that increased xylem cavitation resistance (high structural reinforcement) is conferred at the expense of reduced plant growth (Poorter et al. 2010; Qi et al. 2020), as evident in tropical (Fan et al. 2012) and temperate deciduous trees (McCulloh et al. 2016) and as reported in a global meta-analysis (Liu et al. 2019). However, similar to our results, a positive relationship between growth rate and cavitation resistance has also been reported in subtropical forest tree species (Villagra et al. 2013) and poplar genotypes (Fichot et al. 2010). A possible explanation for this relationship is that increased cavitation resistance prevents hydraulic dysfunction under drought conditions, thereby allowing normal growth. We found P88 to be a better predictor of long-term tree growth than P50 in the present study (Table S2), probably because the threshold for catastrophic hydraulic failure in angiosperms is much closer to P88 (Choat et al. 2012; Vaananen et al. 2020). In addition, 50% loss of xylem hydraulic conductivity did not greatly impair photosynthetic function, particularly in drought-adapted plant species (Li et al. 2019a, b). Moreover, the growth performance of the studied seven timber tree species was more closely associated with HSM under extreme drought than under normal dry conditions (Fig. 4 and Table S2). Tropical karst woody plants were found to exhibit contrasting hydraulic responses under normal and extreme drought periods (Fig. 3; Tan et al. 2020); therefore, extreme drought episodes represent unique natural conditions to assess species-specific drought damage (Nardini et al. 2013).

In the present study, Erythrophleum fordii showed a higher DGR but a relatively lower HGR, thus representing an outlier in the HSM-HGR relationships (Fig. 4). As opposed to the remaining six timber tree species, Erythrophleum fordii presents many large lateral branches, resulting in wide and low crowns. In a recent study, Ding et al. (2021) investigated the aboveground allometry of woody plants across humid-to-arid habitats in eastern Australia and found that with increasing climatic aridity, some Acacia species were shorter but born more branches and wider canopies. Such increase in canopy area represents a conservative water use strategy of plant species to adapt to drought conditions, because it is beneficial for improving water stress tolerance of species and reducing the risk of xylem embolism (Jacobsen et al. 2007; Dantas and Pausas 2013). Consistently, Erythrophleum fordii showed a relatively greater resistance to embolism in the present study (Figure S1). To obtain high wood volume, artificial pruning is regularly implemented in the non-karst Erythrophleum fordii plantation (Hao 2017). In this study, all the tree species did not receive artificial pruning. However, it is not clear whether such treatment affects hydraulic safety and drought adaptation of this species in the dry karst environment.

Furthermore, our results revealed no significant trade-off between hydraulic efficiency and safety across the seven timber species (Figure S2). Similarly, the lack of xylem hydraulic tradeoff has been reported in 17 native tree species in adjacent natural tropical karst forests (Zhang et al. 2021a, b). In the present study, neither ks nor P50 was correlated with WD or any wood anatomical traits (e.g., vessel density, vessel diameter, and vessel wall reinforcement coefficient) (Table S2), and several species (e.g. Zenia insignis) with larger vessel diameter and relatively higher hydraulic conductivity also showed greater cavitation resistance (Figure S2 and Table S1). Similar results have been reported in several Eucalyptus species growing in non-commercial plantations in Argentina; as such, the vessels of these Eucalyptus species were surrounded by vasicentric tracheids, and this structural complex elicites relatively low resistance to water flow and limits the spread of embolism (Barotto et al. 2016). Furthermore, the xylem of several tropical karst tree species bears abundant banded axial parenchyma (e.g. Excentrodendron tonkinense and Diplodiscus trichospermus). In a recent study, Aritsara et al. (2021) found that the vessel-to-parenchyma connectivity and proportion of axial parenchyma fraction significantly affected xylem hydraulic conductivity and cavitation vulnerability, leading to the co-optimization of hydraulic efficiency and safety in 28 woody magnoliids in a tropical rain forest. These results indicate that in addition to the xylem vessel structure, other xylem tissue fractions (e.g., parenchyma cells and vasicentric tracheids) and their functional connections affect hydraulic efficiency and safety.

Contrary to the observations reported in native tropical karst tree species (Fu et al. 2012), we found no significant differences in mean hydraulic traits between evergreen and semi-deciduous timber tree species in the present study (Table S1). Similarly, Markesteijn et al. (2011) found no obvious difference in P50 between evergreen and deciduous tree species in a tropical dry forest. In another study, de Souza et al. (2020) demonstrated that deciduous and evergreen woody species in a seasonally dry neotropical forest exhibited similar isohydric behavior. Together, these reports indicate that evergreen and deciduous trees may likely adopt similar hydraulic strategies to cope with drought conditions, resulting in the convergence of hydraulic traits. In a previous study, Siddiq et al. (2017) investigated the sap flux of plantations of 21 tropical broadleaved timber species in southwestern China and found that evergreen trees consumed more water than deciduous trees during a year. In addition, in the present study, we observed that several semi-deciduous timber tree species (e.g., Zenia insignis) showed relatively high cavitation resistance and great hydraulic safety. Therefore, deciduous timber trees may be more suitable for afforestation while conserving water resources in tropical karst regions.

Conclusions

In the present study, our results showed that the long-term growth rate of seven economical timber tree species planted in a tropical dry karst area was associated with their vulnerability to 88% loss of hydraulic conductivity and hydraulic safety margin under extreme drought conditions. These results are contrary to the conventional view that hydraulic conductivity is a reliable predictor of the tree height and diameter growth rate. Therefore, we suggest that the hydraulic safety-related characteristics can be used as the proxy for selecting tree species for the restoration of tropical rocky desertification areas. As a result of climate change, extreme droughts are becoming more frequent in these tropical regions. Therefore, we recommend the measurements of hydraulic traits combined with the long-term monitoring of water potential and sap flux to better understand the hydraulic risks and water use strategies of afforestation tree species in tropical dry karst environments.

Author contribution statement

SDZ designed the experiment; DLH, ZGL, and WX collected the data; DLH, SDZ and KFC analyzed the data; DLH and ZGL led the writing; KFC revised the initial manuscript. All authors contributed critically to the drafts and gave final approval for publication.