Received: 9 December 2020 | Accepted: 7 August 2021 DOI: 10.1111/1365-2435.13906 RESEARCH ARTICLE Are leaf, stem and hydraulic traits good predictors of individual tree growth? Teresa Rosas1 | Maurizio Mencuccini1,2 | Carles Batlles1 | Íngrid Regalado1 | Sandra Saura-­Mas1,3 | Frank Sterck4 | Jordi Martínez-­Vilalta1,3 1 CREAF, Bellaterra (Cerdanyola del Vallès), Catalonia, Spain Abstract 2 ICREA, Barcelona, Spain 1. A major foundation of trait-­based ecology is that traits have an impact on individ- 3 Universitat Autònoma de Barcelona, ual performance. However, trait–­growth relationships have not been extensively Bellaterra (Cerdanyola del Vallès), Catalonia, Spain assessed in trees, especially outside tropical ecosystems. In addition, measuring 4 Forest Ecology and Forest Management traits directly related to physiological processes remains difficult and the differ- Group, Wageningen University and Research ences between inter-­and intraspecific relationships are seldom explored. Centre, Wageningen, The Netherlands 2. Here, we use individual-­level data on a set of hydraulic, leaf and stem traits to as- Correspondence Jordi Martínez-­V ilalta sess their ability to predict basal area increment (BAI) and growth efficiency (BAI Email: jordi.martinez.vilalta@uab.cat per unit of tree leaf area, GE) among and within species for six dominant tree spe- Funding information cies along a water availability gradient under Mediterranean climate (Catalonia, Ministerio de Ciencia e Innovación, Grant/ NE Spain). Measured traits include: leaf mass per area (LMA), leaf nitrogen con- Award Number: CGL2013-­46808-­R and CGL2017-­89149-­C2; MINECO; ICREA centration (N), leaf C isotopic composition (δ13C), the leaf water potential at turgor Academia loss (P tlp), stem wood density (WD) and branch-­level estimates of the Huber value Handling Editor: Tamir Klein (Hv), sapwood-­and leaf-­specific hydraulic conductivity (KS and KL) and resistance to xylem embolism (P50). 3. Trait–­growth associations were generally weak, particularly for BAI and within species. High values of both growth metrics were associated with ‘conservative’ leaf and hydraulic traits. In particular, BAI was negatively associated with KL (and wood density), while GE increased with LMA, allocation to sapwood relative to leaves (Hv) and resistance to xylem embolism (P50). Climate effects on BAI and GE were indirectly mediated by changes in traits, stand structure and tree basal area. Overall, these results suggest that maintaining functionality over extended periods of time may be more important than maximum gas exchange or hydraulic capacity to achieve high radial growth under Mediterranean climates. 4. Our study reveals that widely used ‘functional traits’ may be poor predictors of tree growth variability along environmental gradients. Moreover, trait effects (when present) do not necessarily conform to simple hypotheses based on our un- derstanding of organ-­level processes. An improved understanding of trait coordi- nation along common axes of variation together with a revaluation of the variables This is an open access article under the terms of the Creat​ive Commo​ns Attri​bution-NonCo​mmercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. © 2021 The Authors. Functional Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society Functional Ecology. 2021;35:2435–2447.  wileyonlinelibrary.com/journal/fec | 2435 2436 | Functional Ecology ROSAS et al. that better reflect whole-­tree performance can greatly improve our understanding of trait–­growth relationships. KEYWORDS basal area increment, functional trait, growth efficiency, hydraulics, intraspecific variability, leaf economic spectrum, water availability 1 | I NTRO D U C TI O N measure traits. Plant hydraulic strategies have often been summa- rized by two main traits that characterize the xylem conductive A major challenge in ecology is to understand the link between safety and efficiency. The maximum water transport capacity (con- plant demographic responses and key traits to improve our ability ductivity) of the fully hydrated xylem (usually normalized per unit to predict vegetation dynamics and the impacts of climate change of sapwood area, KS) is commonly used as a measure of efficiency, on ecosystem structure and functionality. In the last two decades, while xylem safety is often expressed as the xylem water potential trait-­based approaches have concentrated on investigating the rules at which 50% of hydraulic conductivity is lost due to embolism (P50). that constrain global phenotypic diversity across species, focusing Albeit it is not strictly a hydraulic trait, the leaf water potential at on organ-­level spectra such as the leaf economics spectrum (Wright turgor loss (P tlp) has also been used to assess physiological drought et al., 2004), the wood economics spectrum (Chave et al., 2009) or tolerance across species (Bartlett et al., 2012), as it is associated below-­ground traits (Weemstra et al., 2016). The assumption that with the regulation of plant water loss through stomata (Brodribb traits have an impact on plant performance and, thus, can provide a et al., 2003; Martin-­StPaul et al., 2017). Similarly, traits related to basis to scale up from organisms to ecosystem function and dynam- plant allocation and hydraulic architecture, such as the sapwood-­to-­ ics (Lavorel & Garnier, 2002) has been an important foundation of leaf area ratio (the Huber value, Hv) have been recognized as key this research area. However, the relationship between ‘functional’ components of plant strategies to adjust to changes in water avail- traits (sensu Violle et al., 2007) and demographic rates is usually as- ability (Martínez-­Vilalta et al., 2009; Mencuccini & Grace, 1995). The sumed without testing, particularly for large trees, and there is sub- product of KS by Hv gives the leaf-­specific hydraulic conductivity stantial evidence that widely used traits do not predict demographic (KL), a measure of the plant's hydraulic sufficiency. rates (Yang et al., 2018). Usually, studies relating individual traits with tree growth use Among the most commonly measured traits, wood density has absolute growth metrics (such as the basal area increment, BAI) or emerged as a consistent predictor of tree growth and mortality relative metrics that account for overall size (such as the relative rates, with lower wood densities generally associated with faster growth rate; e.g. Gibert et al., 2016). Although it has received little growth rates and lower survival. This evidence has been gath- attention in trait-­based studies, growth efficiency (GE), defined here ered mostly in tropical systems (Iida et al., 2014; Kraft et al., 2010; as the ratio of BAI to crown leaf area (Waring, 1983), could be an Poorter et al., 2008), although similar studies exist for temperate informative, complementary metric to characterize growth in this biomes (Gleason et al., 2018; Klooster et al., 2007; Martínez-­Vilalta context. GE is a physiologically meaningful variable that normalizes et al., 2010). However, easy-­to-­measure (‘soft’) traits such as wood overall radial growth (BAI) by the area exchanging carbon and water density or leaf mass per area (LMA, or its inverse, SLA) frequently ex- with the atmosphere. As such, it may be a more precise indicator plain only a modest proportion of the observed variability in demo- of physiological performance than other measures of growth, which graphic rates (Paine et al., 2015), particularly for adult trees (Gibert are strongly affected by allometric scaling (Hérault et al., 2011). et al., 2016; Iida et al., 2014; Wright et al., 2010). We would expect Most studies relating traits with demographic rates have been ‘soft’ traits to be key indicators of overall function and demographic conducted at the species level, using trait means. This is particularly rates if they were associated with plant resistance to several stress the case for those relating hydraulic traits and plant performance factors (the relationship between wood density and mortality likely (Anderegg et al., 2016; Eller et al., 2018; Medeiros et al., 2019; falls in this category; Chave et al., 2009) or if a strong coordination Poorter et al., 2010; Russo et al., 2010). However, an increasing exists between different traits giving rise to one or a few functional number of studies show the importance of taking into account in- axes at the whole-­plant level, both within and across species. Such traspecific trait variation (Siefert et al., 2015; Violle et al., 2012). The coordination, reflecting a whole-­plant spectrum from conservative capacity of populations to adjust their traits along environmental to acquisitive plants, has been hypothesized (Reich, 2014), but re- gradients results from different processes, including local adapta- mains largely untested. tion and phenotypic plasticity, and a better understanding of these There is growing consensus that hydraulic traits are highly rel- processes is key to forecast species performance under environmen- evant for assessing drought vulnerability and demographic rates tal change (Valladares et al., 2014). However, very few studies to under water-­limiting conditions (Brodribb, 2017; Choat et al., 2018). date have explored trait–­demography relationships at the individual However, they are still much less studied than other, easier to level (but see Liu et al., 2016; Poorter et al., 2018). ROSAS et al. Functional Ecology | 2437 In this study, we use individual tree data to explore the ability All samples and data were collected from May to December 2015. of traits to predict tree growth along a regional water availability To minimize phenological variation in traits within species, species gradient in Catalonia (NE Spain). This gradient encompasses 90 plots were sampled sequentially (P. halepensis was sampled from mid-­May dominated by six tree species, and hence we address the relation- to end June; Q. humilis, end June and July; F. sylvatica, August; P. ship between traits and growth at the intra-­and interspecific levels. sylvestris, September to mid-­October; Q. ilex, mid-­October to mid-­ The main question we ask is: are leaf, stem and hydraulic traits good November; P. nigra, mid-­November to mid-­December). From each predictors of tree radial growth, either in absolute terms or once tree, two branches were sampled from the exposed part of the can- growth is normalized by leaf area? We hypothesize that the explan- opy in the top half of the crown. Branches were transported to the atory power of hydraulic traits will be higher than that of ‘soft’ traits laboratory inside plastic bags under cool and dark conditions and such as wood density or LMA (Brodribb, 2017). Within this general measurements were taken within 24 hr. See Rosas et al. (2019) for objective, we also address the following, more specific questions: additional details on the sampling scheme. are trait–­growth associations consistent within and across species? To what extent are these associations affected by the environment (climate and stand structure) and tree size? 2.2 | Individual trait data For each of the target trees, nine traits were measured (Table S1): 2 | M ATE R I A L S A N D M E TH O DS leaf mass per area (LMA) and nitrogen concentration (N) as proxies for the leaf economics spectrum (Wright et al., 2004); leaf C isotopic 2.1 | Study area and sampling design composition (δ13C) as a measure of water-­use efficiency (Farquhar et al., 1989); stem wood density (WD) as a central trait defining the The study area included all the forested territory of Catalonia (NE wood economics spectrum (Chave et al., 2009); the Huber value at Spain), which encompasses 1.2 million ha, around 38% of its total the branch level (Hv), defined as the ratio of cross-­sectional sap- land area. Catalonia is very diverse both topographically and cli- wood area to subtended (projected) leaf area; the sapwood-­specific matically: mean annual temperature ranges from 3 to 18°C and an- (KS) and leaf-­specific hydraulic conductivity (KL) of branch segments nual rainfall varies from 400 to >1,500 mm (Climatic Digital Atlas as a measure of xylem transport efficiency and sufficiency, respec- of Catalonia; www.openg​is.uab.cat/acdc). The experimental design tively; the water potential causing 50% loss of hydraulic conductivity and general sampling scheme has been previously outlined in Rosas (P50) in branch segments as a proxy for xylem safety; and the leaf et al. (2019). Briefly, we selected six of the most dominant tree spe- water potential at turgor loss (P tlp) as a measure of stomatal sensitiv- cies in Catalonia (three Pinaceae and three Fagaceae), accounting for ity (Brodribb et al., 2003). Standard protocols (Pérez-­Harguindeguy ~75% of the total forest area (Gracia et al., 2004): Pinus sylvestris et al., 2013) were followed for all trait measurements. Leaf meas- L., Pinus nigra J.F.Arnold., Pinus halepensis Mill., Fagus sylvatica L., urements were conducted on previous-­year needles (conifers) and Quercus humilis Mill. and Quercus ilex L. These species have differ- current-­year leaves (broadleaves) to ensure leaves were fully ex- ent distributions in the study area (Figure S1), largely reflecting their panded. See Rosas et al. (2019) for a complete description of the tolerance to drought stress, which increases in the order: F. sylvat- methods used to measure individual traits. ica < P. sylvestris < P. nigra < Q. humilis < Q. ilex < P. halepensis. Mixed forests are also common, particularly those combining P. sylvestris, P. nigra and Q. humilis; and Q. ilex and P. halepensis. 2.3 | Growth data For each species, 15 plots from the Spanish forest inventory (IFN) were resampled in which the target species was dominant The data on individual tree growth were obtained from the third and (accounting for >50% of the total basal area), maximizing the water fourth Spanish National Forest Inventories (IFN3 and IFN4), the lat- availability gradient occupied by each species in the study region. ter conducted over the same time period as our sampling. Spanish Water availability was quantified as the ratio of precipitation to National Forest Inventory plots are circular with variable radius (5–­ potential evapotranspiration (P/PET) for the spring–­summer pe- 25 m) depending on the diameter of the measured trees. The time riod (see below). Five plots per species were sampled for each of interval between inventories varied between 13.9 and 15.4 years three species-­specific P/PET ranges, following a stratified random depending on the plot. We calculated individual tree BAI as the design (dry, corresponding to P/PET < 33th percentile; wet, for P/ difference between final and initial basal area (over-­bark), divided PET > 66th percentile; and mild for the rest; Figure S1). Plots with by the plot-­specific time interval between surveys. Because not all the two highest stoniness levels and those that had been managed trees where traits were measured had been measured at the IFN3 during the last ~15 years according to IFN surveys were discarded. (due to the variable plot radius) tree growth data were missing for 98 Within each plot, five non-­suppressed canopy trees of the tar- of 450 measured trees. get species with a diameter at breast height (DBH) > 12.5 cm were Total tree leaf area (projected) was estimated from the diameter randomly selected, all within 25 m of the centre of the plot. In total, of all primary branches, branch-­level ratios between leaf biomass we sampled 75 individuals (15 plots, 5 trees per plot) per species. and diameter (two to eight branches per tree) and tree-­level LMA, 2438 | Functional Ecology ROSAS et al. as explained in Rosas et al. (2019). In order to make values compa- axes, which resulted in a total of four variables. As before, centred rable across species, seasonal maximum leaf area was estimated, values were calculated as the difference between individual PCA taking into account species phenology and the time of sampling. score values and the average value of the corresponding species. Individual GE was calculated as the ratio between individual BAI The resulting four variables were used as explanatory factors in (per year) and total (projected) tree leaf area. In a supplementary two linear mixed effects models to evaluate how trait coordination analysis, GE was also estimated per unit leaf surface area, assuming determined BAI and GE. To confirm that the previous PCA axes, that, for a given projected leaf area, total surface area is 50% higher which were driven by trait variability both within and among spe- in pines than in broadleaves (Johnson, 1984). Note, however, that cies, also reflected the main axes of variation at the intraspecific GE comparisons between broadleaf and needleleaf species may be level, a second PCA was performed directly on centred trait values problematic. at the species level (Figure S2). Finally, to evaluate the direct and indirect effects of climate (P/ PET), stand structure (initial plot basal area at IFN3), tree size (initial 2.4 | Climatic data tree basal area at IFN3) and trait covariation (PCA axes) on growth rates, two different piecewise structural equation models (piece- To estimate P/PET as a measure of water availability for each wise SEM) were fitted, one for BAI and one for GE. In each case we study plot, climate data were obtained from the Climatic Digital started with the ‘saturated’ model including all possible directional Atlas of Catalonia, a collection of digital maps at approximately effects of the two trait (PCA) axes, P/PET, stand structure and tree 200 × 200 m resolution including average annual radiation, mean, size on growth rate, as well as directional effects of P/PET on stand maximum and minimum annual temperature and annual precipita- structure, tree size and traits (PCA axes), plus all possible covaria- tion for the period 1951–­2010. PET values were calculated accord- tions among them. In addition, alternative models were fit in which ing to the Hargreaves–­Samani method and used to estimate P/PET tree height was used as a measure of tree size instead of individual for the spring–­summer period (April–­September) for each sampled basal area. Piecewise SEM allows piecing multiple individual linear plot. mixed models together into a single causal network, taking into ac- count the hierarchical structure of the data (Lefcheck, 2016). Thus, we included plot nested in species as random effects on the inter- 2.5 | Statistical analyses cept in all SEM sub-­models. The overall SEM fit was evaluated using Shipley's test of d-­separation (Shipley, 2013) and Akaike information All variables were checked for normality and natural-­log transformed criterion (AIC). Models were simplified stepwise by removing the whenever required. First, Pearson correlation coefficients were least significant path until a minimal adequate model with the low- used to quantify the association between traits and growth rates est AIC was obtained. Models within two AIC units were considered (BAI and GE). Second, to separate the intraspecific from the inter- equivalent in terms of fit and the simplest one was selected. Marginal specific component of trait–­growth relationships, we fitted two lin- and conditional R 2 values were computed following Nakagawa and ear mixed effects models for each trait, with either BAI or GE as the Schielzeth (2013). SEM models were fitted with the r package piece- response variable. In these models, two variables were included as wiseSEM (Lefcheck, 2016). All statistical analyses were carried out non-­interacting explanatory factors: the mean trait value at the spe- using R statistical software v.3.3.2. cies level and the species-­centred trait value for each tree. The latter was calculated as the difference between the trait value for a given tree and the average value of the corresponding species. Including 3 | R E S U LT S both variables allows isolating the relative importance of among-­ versus within-­species effects on growth (cf. Rosas et al., 2019). We 3.1 | Trait–­growth associations included plot nested in species as random effects on the intercept of each model. Preliminary analyses showed that including a random Our sampling scheme covered a large gradient of water availability species effect on the slope did not improve model fit. The residuals both within and across species, with spring–­summer P/PET rang- of all models showed no obvious pattern and were approximately ing from 0.29 (a Pinus halepensis plot) to 1.33 (a Fagus sylvatica normally distributed. Linear mixed effects models were fit using the plot; Figure 1e). Individual growth rates (Figure 1a,b) and tree sizes lme4 r package (Bates et al., 2015). (Figure 1d) also varied substantially, as well as stand characteristics, To summarize studied traits into overall axes of variation, a with plot basal areas ranging from 6.2 m2/ha (a Quercus humilis plot) principal component analysis (PCA) was performed on the nine to 49.9 m2/ha (a Fagus sylvatica plot; Figure 1c). traits considered in the study. The first two axes of the PCA, which Individual-­level correlations between trait values and GE were together explained 73% of the variability in the trait data, were stronger than trait–­BAI associations, the highest value correspond- retained. To separate the interspecific from the intraspecific com- ing to the LMA-­GE relationship (r = 0.62; Figure S3). In our models ponent, we also computed the mean of each PCA score at the spe- of trait–­growth relationships, a substantial percentage of the vari- cies level and the species-­centred score value for each of the two ance was explained by differences among species and plots (high ROSAS et al. Functional Ecology | 2439 F I G U R E 1 Boxplot of (a) basal area increment (BAI), (b) growth efficiency (basal area increment per unit of total tree leaf area, GE), (c) plot basal area, (d) tree basal area and (e) precipitation to potential evapotranspiration ratio (P/ PET) as a function of species and family (Pinaceae vs. Fagaceae). The limits of boxes indicate the first and third quartiles, and the horizontal line within each box corresponds to the median. The upper whisker extends to the highest value within 1.5 × IQR (interquartile range) of the third quartile. The lower whisker extends to the lowest value within 1.5 × IQR of the first quartile. Abbreviations: Fs, Fagus sylvatica; Ph, Pinus halepensis; Pn, Pinus nigra; Ps, Pinus sylvestris; Qh, Quercus humilis; Qi, Quercus ilex difference between conditional and marginal R 2 in our mixed mod- 3.2 | Associations between growth and composite els), especially for BAI (Tables S2 and S3). KL and WD were the only trait metrics traits that showed a significant relationship with BAI across species, whereby higher BAI was associated with lower WD and KL values Trait data showed two orthogonal axes of variation that explained 49% (Figure 2; Table S2). Within species, only δ13C showed a significant and 24% of the total variance respectively. The first axis was inter- (positive) effect on BAI (Figure 2; Table S2). preted in terms of conservative leaf resource use and drought resist- Species with higher GE were characterized by high LMA and ance strategies, since high LMA, low N, high resistance to embolism low N, high water-­use efficiency (less negative δ13C) and low (more negative P50) and high water-­use efficiency (less negative δ13C) vulnerability to xylem embolism (more negative P50 ; Figure 3; were all associated with positive loadings (Figure 4). The second axis Table S3). Species GE also showed a positive relationship with Hv was associated with high values of KL (and to a lower extent KS, wood and a negative one with K S that resulted in a non-­significant re- density and Ptlp) and, thus, it was interpreted as a proxy for hydraulic lationship with K L across species (Figure 3; Table S3). However, sufficiency (Figure 4). Associations between these trait axes, repre- when trait–­G E relationships were assessed within species, signif- senting trait covariation, and growth metrics were generally in line icant (positive) relationships were only found for K L , Hv and LMA with the individual trait–­growth relationships reported in the previous (Figure 3; Table S3). For the latter two variables, the intraspecific section. When we explored the effect of trait PCA axes on BAI, we slopes were shallower than the corresponding interspecific slopes. found a negative relationship with the hydraulic sufficiency axis both All results remained similar if GE was estimated per unit leaf sur- within and among species (Table S5). An association between BAI and face area (Table S4). conservative leaf resource use and drought resistant strategies (PCA 2440 | Functional Ecology ROSAS et al. F I G U R E 2 Relationship between basal area increment (BAI) and studied traits at the individual tree level. The black regression lines give the overall among-­species relationships (n = 6 species), and the coloured lines the corresponding within-­species relationships (n = 53–­65 trees per species), when significant (p < 0.05). Variables were natural-­log transformed whenever required to satisfy normality assumptions. See Table S1 for definition of variable abbreviations axis 1) was found only at the intraspecific level (Table S5). For GE, 3.3 | Effects of trait axes, environment and tree size more conservative leaf resource use and drought resistance strategies on plant growth (PCA axis 1), as well as lower hydraulic sufficiency (PCA axis 2) were associated with higher GE, but only across species (Table S5). Models Structural equation models showed that P/PET was positively asso- accounting for trait coordination (PCA axes) explained a higher pro- ciated with initial tree and plot basal areas. At the same time, P/PET portion of growth variance than individual trait models (14% and 52% had a negative effect on both PCA axes, whereby higher values were of BAI and GE respectively; Table S5). related with traits associated with more acquisitive leaf resource use ROSAS et al. Functional Ecology | 2441 F I G U R E 3 Relationship between growth efficiency (basal area increment per unit of total tree leaf area, GE) and studied traits at the individual tree level. The black regression lines give the overall among-­species relationships (n = 6 species), and the coloured lines the corresponding within-­species relationships (n = 53–­65 trees per species), when significant (p < 0.05). Variables were natural-­log transformed whenever required to satisfy normality assumptions. See Table S1 for definition of variable abbreviations strategies, lower drought tolerance and lower hydraulic sufficiency was negatively associated with hydraulic sufficiency (PCA axis 2; (Figure 5). Tree basal area had a strong positive effect on BAI, while Figure 5a), while a positive effect on GE was found for traits related a weaker and opposite effect was found for GE (Figure 5). Plot basal to conservative leaf resource use and drought resistance strategies area only showed a significant (negative) relationship with BAI. (PCA axis 1; Figure 5b). Overall, the model accounted for 58% of the Relationships between trait PCA axes and growth were relatively total variability on BAI and 64% on GE, but the variance explained weak when accounting for climate, stand structure and tree size. BAI by the fixed factors was relatively low (40% for BAI and 10% for GE). 2442 | Functional Ecology ROSAS et al. F I G U R E 4 Principal component analysis (PCA) summarizing trait variability across individual sampled trees. The first two PCA axes with the percentage of explained variance (in brackets) are shown. Variables were natural-­log transformed whenever required to satisfy the normality assumptions. See Table S1 for definition of variable abbreviations (a) (b) P/PET P/PET 0.58 0.27 0.58 0.27 Plot –0.14 –0.23 Tree Plot –0.14 –0.23 Tree basal basal basal basal area area area area PCA PCA PCA PCA 1 2 1 2 –0.23 0.58 –0.16 –0.15 0.23 BAI GE R2 marginal = 0.40 R2 marginal = 0.10 R2 condional = 0.58 R2 condional = 0.64 F I G U R E 5 Piecewise structural equation models relating climate (in terms of precipitation over potential evapotranspiration ratio, P/ PET), forest structure (in terms of initial plot basal area), tree size (in terms of initial tree basal area) and traits (using the first and second PCA components: PCA1 and PCA2 respectively; Figure 4). Panel (a) shows the results for basal area increment (BAI) and panel (b) for growth efficiency (basal area increment per unit of total tree leaf area, GE). Arrows indicate significant links between variables. Solid and dashed lines indicate positive and negative relationships respectively. Standardized path coefficients, as well as the marginal and conditional R 2 values, are shown Results remained largely consistent when tree height was used 4 | D I S CU S S I O N as a measure of tree size instead of individual basal area (Figure S4). In the corresponding structural models, the main difference was that 4.1 | ‘Functional’ traits are weak predictors of tree height was unrelated to P/PET and explained a larger proportion individual radial growth of BAI variance than tree basal area (explained variance remained similar for GE). As a result, a direct effect of P/PET on BAI was de- Only one of the measured traits was predictive of BAI within species tected, which turned the link between hydraulic sufficiency (PCA (δ13C) and two across species (WD, KL), and even in these cases re- axis 2) and BAI non-­significant (p = 0.29). Interestingly, tree height lationships were weak. This is noteworthy considering that all meas- was unrelated to the PCA axes describing trait variability (p ≥ 0.1). ured traits, and particularly those describing the leaf economics ROSAS et al. Functional Ecology | 2443 spectrum (LMA, N), are considered key functional traits and are usu- direction of the effect was consistent, slopes were always steeper ally assumed to be good proxies for growth and performance. As across species. Overall, our results show that trait effects on growth we hypothesized, the predictive power of traits increased for GE, rates were largely driven by differences across species means, likely although relationships were still weak. Even if GE is affected by the reflecting a higher trait variability among species and their turn- distribution of canopy leaf area and by light conditions (Gersonde over along the water availability gradient (Figure 1e; cf. Poorter & O'Hara, 2005; Waring, 1983), it is more closely linked to physi- et al., 2018). ological performance in terms of carbon and water economies than However, a few caveats are in order. First, the fact that we only other measures of growth. It is clear, however, that different growth sampled six species greatly contrasts with the large number of spe- metrics characterize different dimensions of whole-­plant growth cies employed in global assessments of interspecific trait variation (e.g. Gibert et al., 2016; Kleinschmidt et al., 2020), thus a key ques- (Díaz et al., 2016; Wright et al., 2004). Although the study species tion remains to determine which of them is the best proxy for fitness capture a substantial part of the functional variability in the study in a given ecological context. Regardless of this, our study adds to region (Vilà-­C abrera et al., 2015), this low sample size may limit our the evidence showing that ‘functional’ traits, including ‘mechanis- capacity to detect relationships across species. Second, although tic’, hydraulic traits (sensu Brodribb, 2017), are not necessarily good sample size was substantial (450 individuals in total, 75 per species) predictors of demographic rates, at least for mature trees (Gibert and covered a relatively wide environmental gradient, it did not et al., 2016; van der Sande et al., 2015; Yang et al., 2018). cover the whole distribution ranges of the study species. In addition, Interestingly, the trait–­growth relationships we did find were we sampled a relatively low number of individuals per plot (n = 5), frequently counterintuitive. For instance, we observed a negative and selected healthy individuals with sun-­exposed branches, which relationship between hydraulic sufficiency (KL) and BAI across spe- is likely to have resulted in an underestimation of total intraspecific cies, in addition to the expected negative effect of WD on BAI. The trait variability. This may be particularly so for species showing high latter has been interpreted to result from the lower sapwood hy- intra-­plot variability, such as Fagus sylvatica (e.g. Aranda et al., 2017). draulic efficiency of denser wood (Chave et al., 2009). However, we In addition, our relatively long, sequential sampling aimed at min- observed a positive relationship between wood density and hydrau- imizing phenological variation within species implies that some lic efficiency, which likely reflects the particular mixture of species of the variability across species could correspond to temporal ef- in our study and the fact that angiosperm wood tends to be both fects, particularly for relatively labile traits such as δ13C. Finally, we denser and more conductive than that of gymnosperms (Sperry should consider that previous assessments of trait–­growth relation- et al., 2006; see also Figure 4). In addition, no WD-­BAI relationship ships at the plot level found globally consistent patterns (Kunstler was detected within species, in agreement with Fajardo (2016) and et al., 2016). Taken together, these results suggest that traits may be Poorter et al. (2018). good proxies for growth and competitive interactions in the context Trait associations with GE were even more unexpected, as this of community assembly, but not necessarily along environmental growth metric increased with leaf traits related to conservative re- gradients, where multiple, coordinated trait shifts occur as a result source use strategies across species (high LMA and low N). Again, of ecotypic variation and species replacements (Rosas et al., 2019). these relationships have to be interpreted in the context of the spe- cific set of species studied here, but also considering that photo- synthetic capacity per unit leaf area (the relevant comparison here) 4.2 | Trait effects cannot be considered in isolation is weakly related with LMA both across (Wright et al., 2004) and within species (Poorter et al., 2009). In our study, LMA and N per One important implication of our results is that relationships be- unit area (the latter calculated by combining mass-­based N and the tween individual traits and growth cannot be understood without corresponding, tree-­level LMA) were positively related (p < 0.001), considering trait covariation (Figure 4), which makes it difficult to and the effect of mean species N on GE turned from significantly interpret individual trait effects and determine whether they are di- negative when N was expressed per unit mass (Figure 3) to margin- rect or indirect. It is possible, for instance, that resistance to xylem ally positive when N was expressed per unit area (p = 0.09; data not embolism allows species to maintain functionality for longer dur- shown). Previous studies have found that LMA can be positively re- ing extended periods of summer drought, which are common in the lated to growth, even without normalizing it by leaf area, under wa- study area, or permits rapid recovery of gas exchange (and growth) ter-­or nutrient-­limited conditions in the tropics (Poorter et al., 2018; after the first autumn rains (Skelton et al., 2017). These effects could van der Sande et al., 2018). indirectly explain the association between conservative leaves and As hypothesized, trait–­growth relationships were not necessarily high growth rates, given that resistance to embolism was associated consistent within and across species, echoing the fact that trait co- with conservative leaf traits. Our results also highlight the impor- ordination along environmental gradients may change at different tance of allocation when considering trait–­growth relationships organizational levels (Anderegg et al., 2018; Fajardo & Siefert, 2018; (Yang et al., 2018). In particular, we found a compensation between Vilà-­C abrera et al., 2015). Of the 10 relationships that were signif- Hv and K s (cf. Mencuccini et al., 2019), such that species showing icant for either BAI or GE, only two were significant at both lev- higher GE rates had lower K s but also higher Hv, resulting in a non-­ els (LMA and Hv effects on GE). Although in these two cases the significant interspecific relationship with KL . 2444 | Functional Ecology ROSAS et al. Our results emphasize the need to account for ecological con- (Figure 5). Our finding that climate effects on growth were largely text and trait coordination when assessing trait–­growth or trait–­ indirect, mediated by changes in stand structure, tree size and plant demography relationships. The composite trait metrics obtained traits is consistent with earlier reports across species in temper- here (Figure 4) are largely consistent with the main axes of func- ate and Mediterranean systems (Martínez-­V ilalta et al., 2010) and tional variation reported for vascular plants globally, in which the highlights the importance of the ecological context in interpret- first axis corresponds to plant size and wood density and the sec- ing climate-­growth relationships (cf. Chu et al., 2016; Michaletz ond to the leaf economics spectrum (Díaz et al., 2016). The fact et al., 2014; Yang et al., 2018). that in our case the importance of these two axes is reversed is to be expected given the limited variability in plant size in our study, and agrees with previous regional assessments (Fyllas et al., 2020). 5 | CO N C LU S I O N S Previous global studies, however, have not accounted for other po- tentially relevant functional dimensions, such as those described Our results show that traits, including widely used ‘functional traits’ by root and hydraulic traits. Concerning the latter, a study of a re- and even ‘mechanistic’, hydraulic traits (sensu Brodribb, 2017), may forestation trial in Costa Rica reported similar results as ours, with be poor predictors of tree growth variability along environmental the leaf economics spectrum being roughly orthogonal to wood gradients. Although trait–­growth relationships were more predict- density and hydraulic conductivity, although in their case hydrau- able across than within species, their direction was frequently unex- lic conductivity (estimated from anatomy) was negatively related pected and contrary to the notion that the conservative–­acquisitive to wood density (Kleinschmidt et al., 2020). In another study in continuum of traits at the organ level is an indicator of whole-­plant Bolivia, wood density strongly covaried with leaf economics traits performance (cf. Reich, 2014). These results have to be considered and vulnerability to xylem embolism, whereas hydraulic conduc- in the context of the study species, and specifically the contrast- tance was largely orthogonal to that axis (Markesteijn et al., 2011. ing functional properties of Pinaceae and Fagaceae (Vilà-­C abrera Although our results provide evidence for a tight coordination be- et al., 2015), but they also highlight the importance of accounting tween stem hydraulics (at least P50) and leaf-­level traits (cf. Reich, for ecological context and trait coordination when assessing and 2014), further studies are needed to better understand this coor- interpreting the relationships between traits and performance. It dination and its implications for whole-­plant performance at local, is not (only) that ‘soft’ traits are not necessarily good proxies for regional and global scales. more ‘hard’ or ‘mechanistic’ traits, but also that different traits may reflect different functional dimensions that covary in ways we do not yet fully understand. A refined understanding of plant resource 4.3 | Climate effects on growth are mediated by use strategies that explicitly addresses trait coordination at the changes in traits, stand structure and tree size whole-­plant level is needed to improve our ability to predict tree performance (and fitness) from trait measurements under changing We did not find a direct effect of P/PET on growth metrics but environmental conditions. only indirect effects through changes in traits, tree basal area and stand structure. As expected, plots with higher water availability AC K N OW L E D G E M E N T S (higher P/PET) were denser and contained larger trees. Wetter This research was supported by the Spanish Ministry of Economy sites were also associated with more acquisitive leaf and hydraulic and Competitiveness (MINECO) via competitive grants CGL2013-­ strategies (low LMA, less negative P50), as seen in previous stud- 46808-­R and CGL2017-­89149-­C2-­1-­R . T.R. was supported by an ies (e.g. Maherali et al., 2004; Wright et al., 2005). We also found FPI scholarship from MINECO. J.M.-­V. benefited from an ICREA a negative relationship between P/PET and hydraulic sufficiency Academia award. (PCA axis 2), although significant climatic effects on K S , K L or WD (individually) were not detected in a previous study using the same C O N FL I C T O F I N T E R E S T traits (Rosas et al., 2019). Higher plot basal area, presumably as- The authors have no known conflict of interest to declare. sociated with competition intensity, only showed a negative ef- fect on BAI, probably because GE already captures compensatory AU T H O R S ' C O N T R I B U T I O N S changes in tree leaf area as a function of competitive environ- J.M.-­V., M.M., T.R., F.S. and S.S.-­M. conceived the ideas and designed ment (Waring, 1983). Not surprisingly, trees with higher basal area methodology; T.R., C.B. and I.R. collected the data; T.R. analysed showed higher BAI, but the opposite was true when we normalized the data; T.R. and J.M.-­V. led the writing of the manuscript. All au- BAI by total tree leaf area (Mencuccini et al., 2005). Importantly, thors contributed critically to the drafts and gave final approval for when environmental drivers and tree size were simultaneously publication. considered, results were consistent with trait–­growth associations obtained for individual traits: high hydraulic capacity and dense DATA AVA I L A B I L I T Y S TAT E M E N T wood were associated with low BAI, while GE was enhanced by Data available from the Dryad Digital Repository https://doi. conservative leaf resource use and drought tolerance strategies org/10.5061/dryad.vt4b8​g tqr (Martinez-­Vilalta et al., 2020). ROSAS et al. Functional Ecology | 2445 ORCID Farquhar, G. D., Hubick, K. T., Condon, A. G., & Richards, R. A. (1989). Carbon isotope fractionation and plant water-­use efficiency. Teresa Rosas https://orcid.org/0000-0002-8734-9752 In P. Rundel, J. R. Ehleringer, & K. A. Nagy (Eds.), Stable iso- Maurizio Mencuccini https://orcid.org/0000-0003-0840-1477 topes in ecological research (Vol. 68, pp. 21–­4 0). https://doi. Sandra Saura-­Mas https://orcid.org/0000-0001-8539-427X org/10.1007/978-­1-­4612-­3 498-­2_2 Frank Sterck https://orcid.org/0000-0001-7559-6572 Fyllas, N. M., Michelaki, C., Galanidis, A., Evangelou, E., Zaragoza-­ Jordi Martínez-­Vilalta https://orcid.org/0000-0002-2332-7298 Castells, J., Dimitrakopoulos, P. G., Tsadilas, C., Arianoutsou, M., & Lloyd, J. (2020). Functional trait variation among and within species and plant functional types in mountainous mediterranean REFERENCES forests. Frontiers in Plant Science, 11, 212. https://doi.org/10.3389/ Anderegg, L. D. L., Berner, L. T., Badgley, G., Sethi, M. L., Law, B. E., & fpls.2020.00212 HilleRisLambers, J. (2018). Within-­species patterns challenge our Gersonde, R. F., & O'Hara, K. L. (2005). Comparative tree growth ef- understanding of the leaf economics spectrum. Ecology Letters, ficiency in Sierra Nevada mixed-­conifer forests. Forest Ecology 21(5), 734–­744. https://doi.org/10.1111/ele.12945 and Management, 219(1), 95–­108. https://doi.org/10.1016/j. Anderegg, W. R. L., Klein, T., Bartlett, M., Sack, L., Pellegrini, A. F. A., foreco.2005.09.002 Choat, B., & Jansen, S. (2016). Meta-­analysis reveals that hydrau- Gibert, A., Gray, E. F., Westoby, M., Wright, I. J., & Falster, D. S. (2016). On lic traits explain cross-­species patterns of drought-­induced tree the link between functional traits and growth rate: Meta-­analysis mortality across the globe. Proceedings of the National Academy shows effects change with plant size, as predicted. Journal of Ecology, of Sciences of the United States of America, 113(18), 5024–­5029. 104(5), 1488–­1503. https://doi.org/10.1111/1365-­2745.12594 https://doi.org/10.1073/pnas.15256​78113 Gleason, S. M., Stephens, A. E. A., Tozer, W. C., Blackman, C. J., Butler, Aranda, I., Bahamonde, H. A., & Sánchez-­Gómez, D. (2017). Intra-­ D. W., Chang, Y., Cook, A. M., Cooke, J., Laws, C. A., Rosell, J. population variability in the drought response of a beech (Fagus A., Stuart, S. A., & Westoby, M. (2018). Shoot growth of woody sylvatica L.) population in the southwest of Europe. Tree Physiology, trees and shrubs is predicted by maximum plant height and as- 37(7), 938–­949. https://doi.org/10.1093/treep​hys/tpx058 sociated traits. Functional Ecology, 32(2), 247–­259. https://doi. Bartlett, M. K., Scoffoni, C., & Sack, L. (2012). The determinants of leaf org/10.1111/1365-­2435.12972 turgor loss point and prediction of drought tolerance of species and Gracia, C., Burriel, J. A., Ibàñez, J. J., Mata, T., & Vayreda, J. (2004). biomes: A global meta-­analysis. Ecology Letters, 15(5), 393–­4 05. Inventari ecològic i forestal de Catalunya. CREAF. https://doi.org/10.1111/j.1461-­0248.2012.01751.x Hérault, B., Bachelot, B., Poorter, L., Rossi, V., Bongers, F., Chave, Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear J., Paine, C. E. T., Wagner, F., & Baraloto, C. (2011). Functional mixed-­effects models using lme4. Journal of Statistical Software, traits shape ontogenetic growth trajectories of rain forest 67(1), 1–­48. https://doi.org/10.18637/​jss.v067.i01 tree species. Journal of Ecology, 99(6), 1431–­1440. https://doi. Brodribb, T. J. (2017). Progressing from ‘functional’ to mechanistic traits. org/10.1111/j.1365-­2745.2011.01883.x New Phytologist, 215(1), 9–­11. https://doi.org/10.1111/nph.14620 Iida, Y., Poorter, L., Sterck, F., Kassim, A. R., Potts, M. D., Kubo, T., & Brodribb, T. J., Holbrook, N. M., Edwards, E. J., & Gutiérrez, M. Kohyama, T. S. (2014). Linking size-­dependent growth and mortality V. (2003). Relations between stomatal closure, leaf tur- with architectural traits across 145 co-­occurring tropical tree spe- gor and xylem vulnerability in eight tropical dry forest trees. cies. Ecology, 95(2), 353–­363. https://doi.org/10.1890/11-­2173.1 Plant, Cell and Environment, 26(3), 443–­450. https://doi. Johnson, J. D. (1984). A rapid technique for estimating total surface org/10.1046/j.1365-­3 040.2003.00975.x area of pine needles. Forest Science, 30(4), 913–­921. https://doi. Chave, J., Coomes, D., Jansen, S., Lewis, S. L., Swenson, N. G., & org/10.1093/fores​t scie​nce/30.4.913 Zanne, A. E. (2009). Towards a worldwide wood econom- Kleinschmidt, S., Wanek, W., Kreinecker, F., Hackl, D., Jenking, D., ics spectrum. Ecology Letters, 12(4), 351–­366. https://doi. Weissenhofer, A., & Hietz, P. (2020). Successional habitat filtering org/10.1111/j.1461-­0248.2009.01285.x of rainforest trees is explained by potential growth more than by Choat, B., Brodribb, T. J., Brodersen, C. R., Duursma, R. A., López, R., functional traits. Functional Ecology, 34(7), 1438–­1447. https://doi. & Medlyn, B. E. (2018). Triggers of tree mortality under drought. org/10.1111/1365-­2435.13571 Nature, 558(7711), 531–­539. https://doi.org/10.1038/s4158​ Klooster, S.-­H.-­J.-­T., Thomas, E. J. P., & Sterck, F. J. (2007). Explaining 6-­018-­0240-­x interspecific differences in sapling growth and shade tolerance in Chu, C., Bartlett, M., Wang, Y., He, F., Weiner, J., Chave, J., & Sack, L. temperate forests. Journal of Ecology, 95(6), 1250–­1260. https://doi. (2016). Does climate directly influence NPP globally? Global Change org/10.1111/j.1365-­2745.2007.01299.x Biology, 22(1), 12–­24. https://doi.org/10.1111/gcb.13079 Kraft, N. J. B., Metz, M. R., Condit, R. S., & Chave, J. (2010). The rela- Díaz, S., Kattge, J., Cornelissen, J. H. C., Wright, I. J., Lavorel, S., Dray, tionship between wood density and mortality in a global tropical S., & Gorné, L. D. (2016). The global spectrum of plant form and forest data set. New Phytologist, 188(4), 1124–­1136. https://doi. function. Nature, 529(7585), 167–­171. https://doi.org/10.1038/ org/10.1111/j.1469-­8137.2010.03444.x natur​e16489 Kunstler, G., Falster, D., Coomes, D. A., Hui, F., Kooyman, R. M., Laughlin, Eller, C. B., de Barros, F. V., Bittencourt, P. R. L., Rowland, L., Mencuccini, D. C., Poorter, L., Vanderwel, M., Vieilledent, G., Wright, S. J., Aiba, M., & Oliveira, R. S. (2018). Xylem hydraulic safety and construction M., Baraloto, C., Caspersen, J., Cornelissen, J. H. C., Gourlet-­Fleury, costs determine tropical tree growth. Plant Cell and Environment, S., Hanewinkel, M., Herault, B., Kattge, J., Kurokawa, H., … Westoby, 41(3), 548–­562. https://doi.org/10.1111/pce.13106 M. (2016). Plant functional traits have globally consistent effects on Fajardo, A. (2016). Wood density is a poor predictor of competitive competition. Nature, 529, 204–­207. https://doi.org/10.1038/natur​ ability among individuals of the same species. Forest Ecology e16476 and Management, 372(15), 217–­225. https://doi.org/10.1016/j. Lavorel, S., & Garnier, E. (2002). Predicting changes in community com- foreco.2016.04.022 position and ecosystem functioning from plant traits: Revisiting Fajardo, A., & Siefert, A. (2018). Intraspecific trait variation and the the Holy Grail. Functional Ecology, 16, 545–­556. https://doi. leaf economics spectrum across resource gradients and levels of org/10.1046/j.1365-­2435.2002.00664.x organization. Ecology, 99(5), 1024–­1030. https://doi.org/10.1002/ Lefcheck, J. S. (2016). piecewiseSEM: Piecewise struc- ecy.2194 tural equation modelling in r for ecology, evolution, and 2446 | Functional Ecology ROSAS et al. systematics. Methods in Ecology and Evolution, 7, 573–­579. https:// Pérez-­Harguindeguy, N., Díaz, S., Lavorel, S., Poorter, H., Jaureguiberry, doi.org/10.1111/2041-­210X.12512 P., Bret-­Harte, M. S., Cornwell, W. K., Craine, J. M., Gurvich, D. Liu, X., Swenson, N. G., Lin, D., Mi, X., Umaña, M. N., Schmid, B., & Ma, E., Urcelay, C., Veneklaas, E. J., Reich, P. B., Poorter, L., Wright, I. K. (2016). Linking individual-­level functional traits to tree growth J., Ray, P., Enrico, L., Pausas, J. G., de Vos, A. C., Buchmann, N., … in a subtropical forest. Ecology, 97(9), 2396–­2405. https://doi. Cornelissen, J. H. C. (2013). New Handbook for standardized mea- org/10.1002/ecy.1445 surment of plant functional traits worldwide. Australian Journal of Maherali, H., Pockman, W. T., & Jackson, R. B. (2004). Adaptive variation Botany, 23(34), 167–­234. https://doi.org/10.1071/BT12225 in the vulnerability of woody plants to xylem cavitation. Ecology, Poorter, H., Niinemets, Ü., Poorter, L., Wright, I. J., Villar, R., & Niinemets, 85(8), 2184–­2199. https://doi.org/10.1890/02-­0538 U. (2009). Causes and consequences of variation in leaf mass per Markesteijn, L., Poorter, L., Bongers, F., Paz, H., & Sack, L. (2011). Hydraulics area (LMA): A meta-­analysis. New Phytologist, 182(3), 565–­588. and life history of tropical dry forest tree species: Coordination of https://doi.org/10.1111/j.1469-­8137.2009.02830.x species' drought and shade tolerance. New Phytologist, 191(2), 480–­ Poorter, L., Castilho, C. V., Schietti, J., Oliveira, R. S., & Costa, F. R. C. 495. https://doi.org/10.1111/j.1469-­8137.2011.03708.x (2018). Can traits predict individual growth performance? A test Martinez-­V ilalta, J., Rosas, T., Mencuccini, M., Batlles, C., Regalado, Í., in a hyperdiverse tropical forest. New Phytologist, 219(1), 109–­121. Saura-­Mas, S., & Sterck, F.. (2020). Data from: Are leaf, stem and hy- https://doi.org/10.1111/nph.15206 draulic traits good predictors of individual tree growth? (FUN2FUN Poorter, L., McDonald, I., Alarcón, A., Fichtler, E., Licona, J.-­C ., Peña-­ project). Dryad Digital Repository. https://doi.org/10.5061/dryad. Claros, M., Sterck, F., Villegas, Z., & Sass-­Klaassen, U. (2010). vt4b8​g tqr The importance of wood traits and hydraulic conductance Martínez-­V ilalta, J., Cochard, H., Mencuccini, M., Sterck, F., Herrero, A., for the performance and life history strategies of 42 rainfor- Korhonen, J. F. J., Llorens, P., Nikinmaa, E., Nolè, A., Poyatos, R., est tree species. New Phytologist, 185, 481–­492. https://doi. Ripullone, F., Sass-­Klaassen, U., & Zweifel, R. (2009). Hydraulic ad- org/10.1111/j.1469-­8137.2009.03092.x justment of Scots pine across Europe. New Phytologist, 184, 353–­ Poorter, L., Wright, S. J., Paz, H., Ackerly, D. D., Condit, R., Ibarra-­ 364. https://doi.org/10.1111/j.1469-­8137.2009.02954.x Manríquez, G., Harms, K. E., Licona, J. C., Martínez-­Ramos, M., Martínez-­V ilalta, J., Mencuccini, M., Vayreda, J., & Retana, J. Mazer, S. J., Muller-­L andau, H. C., Peña-­Claros, M., Webb, C. O., (2010). Interspecific variation in functional traits, not cli- & Wright, I. J. (2008). Are functional traits good predictors of de- matic differences among species ranges, determines demo- mographic rates? Evidence from five neotropical forests. Ecology, graphic rates across 44 temperate and Mediterranean tree 89(7), 1908–­1920. https://doi.org/10.1890/07-­0207.1 species. Journal of Ecology, 98(6), 1462–­1475. https://doi. Reich, P. B. (2014). The world-­wide ‘fast–­slow’ plant economics spectrum: org/10.1111/j.1365-­2745.2010.01718.x A traits manifesto. Journal of Ecology, 102(2), 275–­3 01. https://doi. Martin-­StPaul, N., Delzon, S., & Cochard, H. (2017). Plant resistance to org/10.1111/1365-­2745.12211 drought depends on timely stomatal closure. Ecology Letters, 20(11), Rosas, T., Mencuccini, M., Barba, J., Cochard, H., Saura-­Mas, S., & 1437–­1447. https://doi.org/10.1111/ele.12851 Martínez-­V ilalta, J. (2019). Adjustments and coordination of hy- Medeiros, C. D., Scoffoni, C., John, G. P., Bartlett, M. K., Inman-­Narahari, draulic, leaf and stem traits along a water availability gradient. New F., Ostertag, R., Cordell, S., Giardina, C., & Sack, L. (2019). An ex- Phytologist, 223(2), 632–­6 46. https://doi.org/10.1111/nph.15684 tensive suite of functional traits distinguishes Hawaiian wet and Russo, S. E., Jenkins, K. L., Wiser, S. K., Uriarte, M., Duncan, R. P., dry forests and enables prediction of species vital rates. Functional & Coomes, D. A. (2010). Interspecific relationships among Ecology, 33(4), 712–­734. https://doi.org/10.1111/1365-­2435.13229 growth, mortality and xylem traits of woody species from Mencuccini M., & Grace J. (1995). Climate influences the leaf area/sap- New Zealand. Functional Ecology, 24, 253–­262. https://doi. wood area ratio in Scots pine. Tree Physiology, 15(1), 1–­10. https:// org/10.1111/j.1365-­2435.2009.01670.x doi.org/10.1093/treep​hys/15.1.1 Shipley, B. (2013). The AIC model selection method applied to path an- Mencuccini, M., Martínez-­V ilalta, J., Vanderklein, D., Hamid, H. A., alytic models compared using a d-­separation test. Ecology, 94(3), Korakaki, E., Lee, S., & Michiels, B. (2005). Size-­mediated ageing re- 560–­564. https://doi.org/10.1890/12-­0976.1 duces vigour in trees. Ecology Letters, 8(11), 1183–­1190. https://doi. Siefert, A., Violle, C., Chalmandrier, L., Albert, C. H., Taudiere, A., Fajardo, org/10.1111/j.1461-­0248.2005.00819.x A., Aarssen, L. W., Baraloto, C., Carlucci, M. B., Cianciaruso, M. V., Mencuccini, M., Rosas, T., Rowland, L., Choat, B., Cornelissen, H., Jansen, Dantas, V. L., Bello, F., Duarte, L. D. S., Fonseca, C. R., Freschet, G. S., Kramer, K., Lapenis, A., Manzoni, S., Niinemets, Ü., Reich, P. B., T., Gaucherand, S., Gross, N., Hikosaka, K., Jackson, B., … Wardle, Schrodt, F., Soudzilovskaia, N., Wright, I. J., & Martínez-­V ilalta, J. D. A. (2015). A global meta-­analysis of the relative extent of intra- (2019). Leaf economics and plant hydraulics drive leaf:wood area ra- specific trait variation in plant communities. Ecology Letters, 18(12), tios. New Phytologist, 224(4), 1544–­1556. https://doi.org/10.1111/ 1406–­1419. https://doi.org/10.1111/ele.12508 nph.15998 Skelton, R. P., Brodribb, T. J., McAdam, S. A. M., & Mitchell, P. J. (2017). Michaletz, S. T., Cheng, D., Kerkhoff, A. J., & Enquist, B. J. (2014). Gas exchange recovery following natural drought is rapid unless Convergence of terrestrial plant production across global climate limited by loss of leaf hydraulic conductance: Evidence from an ev- gradients. Nature, 512, 39–­43. https://doi.org/10.1038/natur​ ergreen woodland. New Phytologist, 215(4), 1399–­1412. https://doi. e13470 org/10.1111/nph.14652 Nakagawa, S., & Schielzeth, H. (2013). A general and simple method Sperry, J. S., Hacke, U. G., & Pittermann, J. (2006). Size and function in co- for obtaining R 2 from generalized linear mixed-­effects mod- nifer tracheids and angiosperm vessels. American Journal of Botany, els. Methods in Ecology and Evolution, 4(2), 133–­142. https://doi. 93(10), 1490–­1500. https://doi.org/10.3732/ajb.93.10.1490 org/10.1111/j.2041-­210x.2012.00261.x Valladares, F., Matesanz, S., Guilhaumon, F., Araújo, M. B., Balaguer, L., Paine, C. E. T., Amissah, L., Auge, H., Baraloto, C., Baruffol, M., Bourland, Benito-­Garzón, M., Cornwell, W., Gianoli, E., Kleunen, M., Naya, D. N., Bruelheide, H., Daïnou, K., de Gouvenain, R. C., Doucet, J.-­L ., E., Nicotra, A. B., Poorter, H., & Zavala, M. A. (2014). The effects of Doust, S., Fine, P. V. A., Fortunel, C., Haase, J., Holl, K. D., Jactel, phenotypic plasticity and local adaptation on forecasts of species H., Li, X., Kitajima, K., Koricheva, J., … Hector, A. (2015). Globally, range shifts under climate change. Ecology Letters, 17(11), 1351–­ functional traits are weak predictors of juvenile tree growth, and 1364. https://doi.org/10.1111/ele.12348 we do not know why. Journal of Ecology, 103(4), 978–­989. https:// van der Sande, M. T., Arets, E. J. M. M., Peña-­Claros, M., Hoosbeek, M. R., doi.org/10.1111/1365-­2745.12401 Cáceres-­Siani, Y., van der Hout, P., & Poorter, L. (2018). Soil fertility ROSAS et al. Functional Ecology | 2447 and species traits, but not diversity, drive productivity and biomass Wright, I. J., Reich, P. B., Westoby, M., Ackerly, D. D. D., Baruch, Z., stocks in a Guyanese tropical rainforest. Functional Ecology, 32(2), Bongers, F., & Villar, R. (2004). The worldwide leaf economics spec- 461–­474. https://doi.org/10.1111/1365-­2435.12968 trum. Nature, 428(6985), 821–­827. https://doi.org/10.1038/natur​ van der Sande, M. T., Zuidema, P. A., & Sterck, F. (2015). Explaining biomass e02403 growth of tropical canopy trees: The importance of sapwood. Oecologia, Wright, S. J., Kitajima, K., Kraft, N. J. B., Reich, P. B., Wright, I. J., Bunker, 177(4), 1145–­1155. https://doi.org/10.1007/s0044​2-­015-­3220-­y D. E., Condit, R., Dalling, J. W., Davies, S. J., Díaz, S., Engelbrecht, Vilà-­C abrera, A., Martínez-­V ilalta, J., & Retana, J. (2015). Functional B. M. J., Harms, K. E., Hubbell, S. P., Marks, C. O., Ruiz-­Jaen, M. trait variation along environmental gradients in temperate and C., Salvador, C. M., & Zanne, A. E. (2010). Functional traits and Mediterranean trees. Global Ecology and Biogeography, 24(12), the growth –­ Mortality trade-­off in tropical trees. Ecology, 91(12), 1377–­1389. https://doi.org/10.1111/geb.12379 3664–­3674. https://doi.org/10.1890/09-­2335.1 Violle, C., Enquist, B. J., McGill, B. J., Jiang, L., Albert, C. H., Hulshof, Yang, J., Cao, M., & Swenson, N. G. (2018). Why functional traits do not C., Jung, V., & Messier, J. (2012). The return of the variance: predict tree demographic rates. Trends in Ecology & Evolution, 33(5), Intraspecific variability in community ecology. Trends in Ecology & 1–­11. https://doi.org/10.1016/j.tree.2018.03.003 Evolution, 27, 244–­252. https://doi.org/10.1016/j.tree.2011.11.014 Violle, C., Navas, M. L., Vile, D., Kazakou, E., Fortunel, C., Hummel, I., & Garnier, E. (2007). Let the concept of trait be functional! Oikos, 116(5), S U P P O R T I N G I N FO R M AT I O N 882–­892. https://doi.org/10.1111/j.0030-­1299.2007.15559.x Additional supporting information may be found online in the Waring, R. (1983). Estimating forest growth and efficiency in relation to Supporting Information section. canopy leaf area. Advances in Ecological Reserach, 13, 327–­3 49. Weemstra, M., Mommer, L., Visser, E. J. W., van Ruijven, J., Kuyper, T. W., Mohren, G. M. J., & Sterck, F. J. (2016). Towards a multidimensional root trait framework: A tree root review. New Phytologist, 211(4), How to cite this article: Rosas, T., Mencuccini, M., Batlles, C., 1159–­1169. https://doi.org/10.1111/nph.14003 Regalado, Í., Saura-­Mas, S., Sterck, F., & Martínez-­Vilalta, J. Wright, I. J., Reich, P. B., Cornelissen, J. H. C., Falster, D. S., Groom, P. (2021). Are leaf, stem and hydraulic traits good predictors of K., Hikosaka, K., Lee, W., Lusk, C. H., Niinemets, Ü., Oleksyn, J., individual tree growth? Functional Ecology, 35, 2435–­2447. Osada, N., Poorter, H., Warton, D. I., & Westoby, M. (2005). Modulation of leaf economic traits and trait relationships by cli- https://doi.org/10.1111/1365-­2435.13906 mate. Global Ecology and Biogeography, 14, 411–­421. https://doi. org/10.1111/j.1466-­822x.2005.00172.x