Leaf and Stem Traits are Linked to Liana Growth Rate in a Subtropical Cloud Forest

: There is accumulating evidence that the abundance and biomass of lianas are increasing with global climate change in the Neotropics. However, our knowledge of growth–trait relationships among lianas is surprisingly rare. Here, we monitored the relative growth rate of 2860 individuals from seven deciduous and four evergreen liana species in a 20 ha subtropical cloud forest dynamics plot at high elevation (2472–2628 m a.s.l.) in southwest China. We linked the relative growth rate of lianas with nine leaf traits associated with leaf morphology, nutrient concentrations, and water hydraulic capacity as indicated by leaf vein density, and ﬁve stem wood traits related to stem water transport capacity and wood density. Our results showed that deciduous lianas have higher relative growth rates than their evergreen counterparts. Across all lianas studied, the relative growth rate was positively correlated with the leaf area and speciﬁc leaf area, but negatively correlated with leaf dry matter content. The relative growth rate of lianas was strongly correlated with nitrogen concentration after excluding the legume liana species. The relative growth rate was decoupled from leaf phosphorus and potassium concentrations, leaf vein density, and stem vessel traits across all lianas investigated. For four evergreen lianas, there were positive associations of the relative growth rate with the leaf thickness and diameter of the largest vessels. This study is the ﬁrst to illustrate the relationships of liana growth with leaf and stem traits in the high-elevation subtropical cloud forest. More studies from diverse forest ecosystems are needed to comprehensively understand the mechanism underlying liana growth patterns. positive correlations of relative growth rate with leaf thickness and mean biggest vessel diameter exist only in evergreen lianas. These results provide a fundamental understanding


Introduction
Plant traits are morphological, physiological, and reproductive properties that potentially determine plant fitness via direct or indirect effects on individual growth, reproduction, and survival [1]. Over the last two decades, trait-based studies on plant performance (e.g., growth) have received particular attention. To date, it is still an open question as to whether plant traits are strongly linked with their growth rate [2][3][4][5]. The majority of previous studies about plant trait-growth patterns have focused on trees [2][3][4][6][7][8], with fewer reports concerning lianas [9,10]. It is, therefore, of importance to unveil how the growth rate of lianas is driven by their interspecific variation in plant traits. correlated with resource conservation traits, such as leaf dry matter content, leaf thickness, and vessel and wood densities.

Study Site
In 2014, the present study was carried out in a 20 ha forest dynamics plot (500 × 400 m, consisting of 500 20 × 20 m quadrats) in a national nature reserve in Ailao Mountains, Yunnan Province, SW China (101.03 • -101.03 • E, 24.53 • -24.54 • N; 2472-2628 m a.s.l.) [44]. This permanent plot was constructed based on the standard protocols proposed by the Center for Tropical Forest Science (CTFS). The mean annual precipitation is ca. 1778 mm, 86% of which is concentrated in the rainy season (from May to October, 2002-2011) [44]. The mean annual temperature is 11.3 • C, with the mean monthly temperature being 5.7 • C in January and 15.6 • C in July. The type of soil is yellow-brown, which contains 129.1 g kg −1 organic matter, 5.2 g kg −1 N, 0.6 g kg −1 P, 9.46 g kg −1 K at 0-20 cm depth. The available N, P, and K concentrations are 45.66, 11.10, and 185.39 mg kg −1 , respectively; the pH value is 4.24. The 20 ha plot is dominated by evergreen broadleaved tree species [44].

Relative Growth Rate of Lianas
During October 2015 to January 2016, all rooted individuals of 11 liana species with diameters ≥ 1 cm (Table 2) were tagged, measured, and identified at the species level according to a standardized census method [45][46][47]. Based on this method, we measured liana diameters 130 cm from the last rooting point at the ground surface. When lianas branched below 130 cm (but above 40 cm from the soil), we measured the liana diameter 20 cm below the branching point. We re-measured the diameters of all liana stems at the same point marked in the first census time during May to July 2018. We monitored liana species with individuals ≥ 50 to reduce the measurement error due to small samples. In total, 2860 liana individuals from 11 species of eight families (seven deciduous and four evergreen liana species) were monitored in the present study (Table 2).
Liana growth is strongly correlated with the diameter increment [48][49][50]. We, therefore, used the increment rate of the diameter to estimate the relative growth rate of lianas. The relative stem diameter growth rate (RGR) was calculated as (log[D 2 ]-log[D 1 ])/time (yr) [2], where D 1 and D 2 represent the diameters of lianas at the first (2015-2016) and second (2018) census times, respectively. We excluded lianas with diameter growth rates that were negative and anomalous (≥4 cm), assuming such growth data were caused by measurement error. Plant size needs to be considered when comparing the interspecific growth rate if plant growth is size-dependent and if size distribution patterns significantly differ among species [2]. We evaluated this potential influence for our growth data by comparing liana absolute growth rates standardized to 2.5 cm diameter with an average RGR for the 2860 individuals, as Poorter et al. (2008) suggested [2]. The logarithm of the absolute growth rates of lianas was fitted to the logarithm of the initial diameter for the liana species studied. We then used this equation to estimate the absolute growth rate of lianas at 2.5 cm diameter. We found that the estimated absolute growth rate and the average RGR observed for all liana species were strongly correlated ( Figure 1), indicating that the size effect on the liana growth rate in the present study can be neglected in the subsequent analysis [2].  Liana growth is strongly correlated with the diameter increment [48][49][50]. We, therefore, used the increment rate of the diameter to estimate the relative growth rate of lianas. The relative stem diameter growth rate (RGR) was calculated as (log[D2]-log[D1])/time (yr) [2], where D1 and D2 represent the diameters of lianas at the first (2015-2016) and second (2018) census times, respectively. We excluded lianas with diameter growth rates that were negative and anomalous (≥4 cm), assuming such growth data were caused by measurement error.
Plant size needs to be considered when comparing the interspecific growth rate if plant growth is size-dependent and if size distribution patterns significantly differ among species [2]. We evaluated this potential influence for our growth data by comparing liana absolute growth rates standardized to 2.5 cm diameter with an average RGR for the 2860 individuals, as Poorter et al. (2008) suggested [2]. The logarithm of the absolute growth rates of lianas was fitted to the logarithm of the initial diameter for the liana species studied. We then used this equation to estimate the absolute growth rate of lianas at 2.5 cm diameter. We found that the estimated absolute growth rate and the average RGR observed for all liana species were strongly correlated ( Figure 1), indicating that the size effect on the liana growth rate in the present study can be neglected in the subsequent analysis [2].

Leaf and Stem Traits
We measured 14 key leaf and stem traits of 11 liana species in the rainy season (July to September) in 2016. Briefly, we collected 3-6 individuals with diameters of 2-6 cm per liana species and 3-5 stems per individual with intact, mature, healthy, and sun-exposed leaves. We used a HP Scanjet G3110 scanner and ImageJ software (https://imagej.en.softonic.com/) to measure the areas (LA) of simple leaves or leaflets of compound leaves. Leaflets were considered to be functionally equivalent to simple leaves [51]. We measured leaf fresh weights and then leaf samples were ovendried at 70 °C for at least 48 h. The specific leaf area (SLA) was calculated as the leaf area divided by

Leaf and Stem Traits
We measured 14 key leaf and stem traits of 11 liana species in the rainy season (July to September) in 2016. Briefly, we collected 3-6 individuals with diameters of 2-6 cm per liana species and 3-5 stems per individual with intact, mature, healthy, and sun-exposed leaves. We used a HP Scanjet G3110 scanner and ImageJ software (https://imagej.en.softonic.com/) to measure the areas (LA) of simple leaves or leaflets of compound leaves. Leaflets were considered to be functionally equivalent to simple leaves [51]. We measured leaf fresh weights and then leaf samples were oven-dried at 70 • C for at least 48 h. The specific leaf area (SLA) was calculated as the leaf area divided by leaf dry mass, while the leaf dry matter content (LDMC) was calculated as leaf dry mass per unit leaf fresh weight. Cross-section images of each of five to seven leaves or leaflets were taken with a Binocular Biological microscope (Leica DM2500, Wetzlar, Germany) and leaf thickness (LT) values were measured using ImageJ. An additional three to five leaf segments were immersed in a 5% NaOH solution until leaf veins became clear. Images were then taken using the Binocular Biological microscope and the total vein lengths in whole images were determined using ImageJ. Leaf vein density (D vein ), an indication of leaf hydraulic capacity, was calculated as the total vein length divided by leaf area. Leaf thickness, total vein length, and vessel diameter were measured using the measurement tool in ImageJ software (https://imagej.en.softonic.com/).
Leaf samples of each of three to five individuals per liana species were oven dried at 70 • C for 48 h and ground to pass through a 60-mesh sieve. We measured the leaf N concentration with a Dumas-type combustion C-N elemental analyzer (Vario MAX CN, Elementar Analysensysteme GmbH, Hanau, Germany). We determined leaf P and K concentrations with an inductively coupled plasma atomic emission spectrometer (iCAP 7400, Thermo Fisher Scientific, Bremen, Germany). The N/P ratio was calculated as a proxy of nutrient limitation [52].
Stem wood traits were measured on a 5-cm-long stem segment with a diameter of ca. 1 cm. The volume of fresh wood with the pith and bark removed was first measured using the water displacement method. Wood samples were then oven-dried at 70 • C for 72 h for dry weight. Wood density (WD) was calculated as the wood dry mass per unit fresh volume. Another stem segment of ca. 2 cm in length was used to determine the vessel diameter (D V ) and density (VD). At least ten images were taken at 100-1000× magnification per individual using a microscope (Smartzoom 5, Carl Zeiss, Germany) and then D v was determined for all vessels with ImageJ. Due to the elliptical shape of most vessels, D v was calculated as [53]: where a and b indicate the radii of vessel major and minor axes, respectively. Vessel density was defined as the vessel number per area. For each species, we also selected 6-10 images of each of 3-6 individuals and measured the diameters of the ten biggest vessels (D max ), which are strongly correlated with stem hydraulic conductivity [54]. Stem theoretical hydraulic conductivity (K t ) was determined as: where π is a constant of 3.14, ρ is the water density (997.05 kg m −3 ) and η is the water viscosity (0.89 × 10 −9 MPa s −1 ) at 25 • C, A is the image area, and i = 1 to n vessels in the image [55].

Data Analyses
All variables were log 10 -transformed to improve the normality of distribution before analysis. The differences in RGR and other traits between seven deciduous and four evergreen liana species were compared using independent sample t-tests. Pearson's correlation and principal component analysis (PCA) were used to evaluate trait associations. PCA was performed based on log 10 -transformed data and varimax rotation was chosen to simplify the axes. All analyses were conducted in SPSS (version 22.0; SPSS, Inc., Chicago, IL, USA).

Results
Deciduous lianas had significantly higher RGR values than evergreen lianas (p < 0.01; Figure 2). Across liana species, RGR was positively significantly correlated with leaf area (Figure 3a). However, the positive association of RGR with leaf area only existed in deciduous lianas (Figure 3a). RGR was positively related to SLA (Figure 3b), but negatively related to LDMC (Figure 3c) across all lianas. Surprisingly, RGR was positively associated with leaf thickness in evergreen lianas, but negatively associated with leaf thickness in deciduous lianas (Figure 3d).  Table 2 for species codes. Data were means ± SE.
There were no significant correlations of RGR with P, K concentrations, and N/P ratios ( Figure  4) across deciduous, evergreen, or pooled lianas. When the legume liana (Callerya dielsiana (Harms) P. K. Loc ex Z. Wei & Pedley) was excluded, RGR was positively correlated with N concentration across the remaining ten liana species (Figure 4a). As for the associations of RGR with leaf vein and stem xylem properties ( Figure 5), we only found that RGR was positively related to diameter for the biggest vessels in evergreen lianas (Figure 5a).
The results of principal component analysis based on RGR and 14 traits of 11 liana species showed that the first and second components accounted for 34.3% and 27.6% of the total variance, respectively ( Figure 6). The first axis was positively correlated with traits representative of RGR and stem xylem vessel parameters (e.g., Dmax, DV, Kt). At the opposite end were species with high LDMC, vessel density, and vein density (Figure 6a). The second axis correlated positively with N and P concentrations and negatively with the N/P ratio. Deciduous lianas overlapped with evergreen lianas to a large extent in the multivariate trait space (Figure 6b).  Table 2 for species codes. Data were means ± SE. There were no significant correlations of RGR with P, K concentrations, and N/P ratios ( Figure 4) across deciduous, evergreen, or pooled lianas. When the legume liana (Callerya dielsiana (Harms) P. K. Loc ex Z. Wei & Pedley) was excluded, RGR was positively correlated with N concentration across the remaining ten liana species (Figure 4a). As for the associations of RGR with leaf vein and stem xylem properties ( Figure 5), we only found that RGR was positively related to diameter for the biggest vessels in evergreen lianas (Figure 5a).
The results of principal component analysis based on RGR and 14 traits of 11 liana species showed that the first and second components accounted for 34.3% and 27.6% of the total variance, respectively ( Figure 6). The first axis was positively correlated with traits representative of RGR and stem xylem vessel parameters (e.g., D max , D V , K t ). At the opposite end were species with high LDMC, vessel density, and vein density (Figure 6a). The second axis correlated positively with N and P concentrations and negatively with the N/P ratio. Deciduous lianas overlapped with evergreen lianas to a large extent in the multivariate trait space (Figure 6b).       Table 2 for species codes and text for trait abbreviations.

Discussion
Our results showed that deciduous lianas grew more quickly than evergreen lianas (Figure 2), supporting our first hypothesis. The reason may lie in the fact that deciduous lianas have higher SLA values than evergreen lianas (227.81 ± 20.18 vs. 147.48 ± 17.62 cm 2 g −1 ; also see Figure 3b), consistent with similar patterns in trees [56][57][58][59], presumably resulting in the conclusion that relatively faster growth is a general feature of deciduous species. Consistent with our second hypothesis, we found that lianas with larger leaf areas had higher growth rates across all lianas studied (Figure 3a). Throughout a given year, the proportion of sunny days in the Ailao Mountains is low, and it is extremely low in the rainy season (Table 1). Increased leaf area will allow lianas to harvest more light for photosynthesis [8,11,22,60], which is possibly an important strategy for lianas to adapt to cloud forest conditions. Interestingly, the positive relationship of RGR with leaf area is existent only in deciduous species (Figure 3a). This is highly important for deciduous lianas because they need to take advantage of the rainy season to grow when their leaves are present.  Table 2 for species codes and text for trait abbreviations.

Discussion
Our results showed that deciduous lianas grew more quickly than evergreen lianas (Figure 2), supporting our first hypothesis. The reason may lie in the fact that deciduous lianas have higher SLA values than evergreen lianas (227.81 ± 20.18 vs. 147.48 ± 17.62 cm 2 g −1 ; also see Figure 3b), consistent with similar patterns in trees [56][57][58][59], presumably resulting in the conclusion that relatively faster growth is a general feature of deciduous species. Consistent with our second hypothesis, we found that lianas with larger leaf areas had higher growth rates across all lianas studied (Figure 3a). Throughout a given year, the proportion of sunny days in the Ailao Mountains is low, and it is extremely low in the rainy season (Table 1). Increased leaf area will allow lianas to harvest more light for photosynthesis [8,11,22,60], which is possibly an important strategy for lianas to adapt to cloud forest conditions. Interestingly, the positive relationship of RGR with leaf area is existent only in deciduous species (Figure 3a). This is highly important for deciduous lianas because they need to take advantage of the rainy season to grow when their leaves are present.
Partly inconsistent with our third hypothesis, we did not find that RGR was positively correlated with N, P, K, and the N/P ratio across all lianas studied ( Figure 4). However, RGR was significantly positively related to N concentration (Figure 4a) when a legume liana species, Callerya dielsiana, was excluded. This may suggest that legume lianas employ alternative approaches to fix and use N resources [61]. The N/P ratio has been considered to be an indicator of nutrient limitation, which was divided into three levels: N/P ratio < 14 (N limitation), >16 (P limitation), and 14 ≤ N/P ratio ≤ 16 (limitation of N and P or both not) [52,62,63]. We found that the N/P ratios for eight in 11 lianas were much lower than 14 (Figure 4d), indicating that N is a growth-limiting nutrient for most lianas in subtropical evergreen broadleaved forests in SW China. This does not seem to be true for the legume liana species (Callerya dielsiana), because it had the highest N concentration among all lianas measured (Figure 4a).
We found that the relative growth rates for lianas were related to leaf resource acquisition and conservation traits, and were seldom associated with stem hydraulic properties, partly supporting our fourth hypothesis. We found that there were strong correlations of RGR with SLA and LDMC, with deciduous lianas positioned at high SLA or low LDMC ends (Figure 3b,c). A high SLA or low LDMC means low investment in construction but high investment in the photosynthetic apparatus [22,27,64,65], ultimately translating into rapid growth. The positive SLA-growth relationship was also confirmed in seedlings grown under non-limiting conditions [66,67].
Across 11 lianas studied, RGR was not correlated with wood density (Figure 5b), consistent with a report on lianas in Barro Colorado Island, Panama [9]. However, there was a significantly negative correlation between the relative growth rate and wood density among lianas in a montane forest in SW China [10]. Strong leaf and stem water transport capacities are usually positively linked with high growth rates among trees [27]. However, we did not find any significant associations of leaf vessel and stem vessel properties with RGR across lianas ( Figure 5), with the exception of a significant correlation between RGR and the mean biggest vessel diameter in evergreen lianas (Figure 5e). These results may suggest that water seems an unlikely limitation on liana growth in the Ailao Mountains, as indicated by high precipitation and a high proportion of foggy days during a year (ca. 63%; Table 1). Surprisingly, we found some associations of RGR with leaf morphology and stem hydraulic properties only in evergreen lianas, suggesting that deciduous or evergreen lianas may employ some differentiated strategies to adapt to the cloud forest environment. We found that the RGR of evergreen lianas was positively related to leaf thickness ( Figure 3d) and the mean biggest vessel diameter (Figure 5e). Thicker leaves usually have longer lifespans [68]. In addition, leaf thickness is positively related to leaf hydraulic conductivity [69]. Evergreen lianas with thicker leaves or larger vessels are able to supply enough water and continuously assimilate in the winter time, during which temperate and sunny weather in the cloud forest (Table 1) is still suitable for photosynthesis [41,70]. Therefore, a strong coordination of RGR with leaf thickness and mean biggest vessel diameter in evergreen lianas could be expected in the cloud forest in the Ailao Mountains.

Conclusions
In this study, we attempted to investigate the relationships between liana functional traits and relative growth rate. We found that deciduous lianas possessed higher relative growth rates than evergreen lianas in the subtropical cloud forest. The variation in relative growth rate was mainly driven by leaf area, specific leaf area, leaf dry matter content, instead of leaf nutrient status, leaf hydraulic capacity, and stem hydraulic properties. Interestingly, the relative growth rate for lianas was strongly correlated with N concentration after excluding the legume liana species, suggesting that legume lianas may employ alternative strategies to use N. Deciduous lianas probably perform differently from evergreen lianas, as positive correlations of relative growth rate with leaf thickness and mean biggest vessel diameter exist only in evergreen lianas. These results provide a fundamental understanding of how the liana growth links or decouples from leaf and stem traits in a high-elevation cloud forest. Future investigations must consider liana growth in diverse ecosystems to fully understand the global pattern of liana growth.