Functional traits and their plasticity predict tropical trees regeneration niche even among species with intermediate light requirements

Authors


Correspondence author. E-mail: laurans@cirad.fr

Summary

  1. Niche differentiation is a key issue in the current debate on community assembly mechanisms. In highly diverse moist tropical forests, tree species sensitivity to canopy openness is thought to be a major axis in niche differentiation. In the past, the syndrome of traits driving the demographic trade-off involved in the niche-based theory of coexistence has always been established among species situated at the two extremities of the shade-tolerance gradient, even though most tropical tree species have intermediate light requirements. In addition, trait plasticity has seldom been linked to tropical tree species distribution along environmental gradients.
  2. This article examines covariations between leaf traits, whole-plant traits and niche parameters among 14 tree species with intermediate light requirements in French Guiana and across a range of canopy openness. Each functional trait measured under field conditions was characterized by a median value and a degree of plasticity expressed under contrasting light regimes. Niche differentiation was characterized in terms of spatial light gradient. We first examined covariations between functional traits then explored to what degree the median value and plasticity in functional traits could predict light niche characteristics at the sapling stage and the ontogenetic change in light availability estimated by adult stature.
  3. Leaf mass per area (LMA) was positively correlated with leaf life span (LLS); species with higher LMA and higher LLS displayed lower diameter growth rates (GRs) and lower responsiveness to canopy gap at both whole-plant and population levels. This proved that the relationships previously established over a broader range of species held true within the narrow range of the light requirements covered.
  4. Height GR plasticity accounted for 49% of the variation in light niche optimum. LMA plasticity, unlike LLS plasticity, was significantly correlated with light niche breadth and adult stature.
  5. Synthesis. This study demonstrates the relevance of considering the phenotypic plasticity in functional traits in community ecology, particularly for quantifying breadth of species distribution over environmental gradients. Our findings did not support Hubbell's hypothesis of functional equivalence and suggest that even a rather subtle variation in forest canopy disturbance promotes the coexistence of tropical tree species.

Introduction

Niche differentiation is a key issue in the current debate on community assembly mechanisms (Leibold & McPeek 2006). On the basis of field experience on the island of Barro Colorado, Hubbell (2005) argued that most tree species are not niche differentiated and developed the ‘core hypothesis’ of functional equivalence that contrasts with the deterministic niche-based view of community assemblage (Silvertown 2004; Kraft, Valencia & Ackerly 2008). The niche theory assumes that species respond differently to ecological heterogeneities, and these differences are usually the result of trade-offs in the abilities of species to interact with various features of their environment (Kneitel & Chase 2004; Leibold & McPeek 2006). In moist tropical forests, the response of tree species to canopy openness is thought to be a major source of niche differentiation (Poorter & Arets 2003; Vincent et al. 2011). Canopy disturbance regimes can therefore have a marked impact on community diversity and composition (Molino & Sabatier 2001). A demographic trade-off between high-light growth rate (GR) versus low-light survival rate has been identified in many studies (Davies 2001; Wright et al. 2003; Baraloto, Goldberg & Bonal 2005; Poorter et al. 2008) as a potential driver of tropical trees' coexistence and has been associated with a syndrome of functional traits (Poorter & Bongers 2006; Sterck et al. 2011). This is challenged by the following argument: demographic trade-off occurs among species situated at the two extremities of the differentiation axis (Hubbell 2005), namely pioneer species and truly shade-tolerant species. But, such species are scarce in tropical trees community (Clark & Clark 1992; Wright et al. 2003). Consequently, the niche theory has so far failed to explain how most tropical tree species coexist. In support of this and to the best of our knowledge, all studies investigating trade-offs among tropical tree species have included pioneer and truly shade-tolerant species. One exception is the study conducted by Bloor & Grubb (2003) that corroborated the previous argument in that the authors found no relationship between survival in low-light conditions and relative GR in high-light conditions. However, as this study considered 15 shade-tolerant tree species at the seedling stage and grown under controlled conditions, doubt persists as to whether niche theory assumptions hold for the majority of tropical tree species. The first objective of the study described herein was to address this question by testing the validity of the trait covariations observed between pioneer and truly shade-tolerant species in a subset of tropical tree species from which all pioneer and truly shade-tolerant species had been excluded. This group of species will hereinafter be referred to as ‘species with intermediate light requirements’. The significant contribution of leaf traits to plant performance has previously been assessed by a modelling approach which showed that specific leaf area, photosynthetic capacity and leaf survival rate may jointly explain 50% of the growth–survival trade-off (Sterck, Poorter & Schieving 2006). Shade-tolerant species tend to have lower photosynthetic capacities, higher wood densities, higher leaf mass per area (LMA) and longer leaf life spans (LLS) than pioneer species (Reich et al. 2003). LMA and LLS are believed to have a direct effect on plant carbon budget and drive interspecific variations in GRs and plant survival by controlling leaf investment returns (Westoby et al. 2002; Vincent 2006). Thus, functional traits have been widely used to predict species performance and demographic characteristics (Poorter et al. 2008; Wright et al. 2010; Herault et al. 2011). To date, functional traits have only seldom been linked to tropical tree species distributions along environmental gradients (Poorter & Bongers 2006; Sterck et al. 2011), although some recent studies advocate their usefulness for quantifying a species’ niche (Mc Gill et al. 2006; Violle & Jiang 2009). In the study described herein, we tested whether the slow–fast syndrome of traits described earlier occurs among species with intermediate light requirements only and whether it is linked to species distribution along a light gradient resulting from canopy openness.

Intraspecific trait variability and phenotypic plasticity (Bradshaw 1965) have seldom been used to analyse functional traits, although an increasing number of studies report that they are, in many cases, quantitatively not negligible when compared to interspecific variability (Albert et al. 2010; Messier, McGill & Lechowicz 2010; Violle et al. 2012) and that they affect key ecological processes (Grime, Crick & Rincon 1986; Lecerf & Chauvet 2008; Violle & Jiang 2009; Berg & Ellers 2010; Jung et al. 2010). Phenotypic plasticity is an important means by which individual plants can cope with environmental heterogeneity. Several studies have evaluated the relationship between phenotypic plasticity and niche breadth in annual species (Sultan et al. 1998; Sultan 2001) and have provided evidence that generalist species show higher levels of phenotypic plasticity than specialist species. Very little is known about the role of phenotypic plasticity in environmental responsiveness among tropical tree species. We hypothesize that phenotypic plasticity in response to light plays an important role in shaping the niche of tropical trees. This is supported by the fact that most species have to cope with highly variable light conditions across (i) the understorey-gap horizontal gradient at the juvenile stage and (ii) the understorey-canopy vertical gradient along ontogenetic stages. To the best of our knowledge, only one study (Popma, Bongers & Werger 1992) has investigated the link between plasticity and light niche breadth in tropical trees: the authors showed among 68 Mexican tropical tree species that leaf traits variations in sun versus shade environments were less significant for gap-specialist and understorey-specialist species than for gap-dependent species that occur over a wider range of light environments. Similarly, very few studies (Cai, Rijkers & Bongers 2005; Martinez-Garza & Howe 2005; Rozendaal, Hurtado & Poorter 2006) have examined the relationship between phenotypic plasticity and the range of light environments experienced along ontogenetic stages that are usually estimated through adult stature. Cai, Rijkers & Bongers (2005) and to a lesser extent Rozendaal, Hurtado & Poorter (2006) reported that leaf traits plasticity was greater in tall species that experience major, predictable changes in irradiance throughout their development than in smaller species that always remain in understorey or gap light conditions. By contrast, Martinez-Garza & Howe (2005) failed to find any such relationship among eight non-pioneer species.

The study described herein aimed to determine how leaf traits, whole-plant traits and niche parameters varied among 14 tropical tree species of intermediate light requirements across a range of canopy openness values. We focused on the sapling stage of development as differences in shade tolerance early in tree ontogeny are known to be a strong determinant of forest dynamics. LMA and LLS were chosen as key descriptors of the species resource investment pattern. Whole-plant functional traits were derived from diameter and height GRs under contrasting light environments. Each functional trait was characterized by a median value and a degree of plasticity in response to light level. First, we aimed to test whether covariations between LMA and LLS and between leaf traits and whole-plant traits held true for species known to have intermediate light requirements. Second, we investigated the relationship between functional traits and niche differentiation characterized in terms of spatial light gradient. We addressed the following questions: (i) to what degree can leaf or whole-plant traits (median value and degree of plasticity) predict population response to canopy disturbance (light niche optimum)? (ii) is leaf or whole-plant trait plasticity related to light niche breadth and/or adult stature?

Materials and methods

This study made use of two independent data sets collected at the same experimental site in French Guiana. Values for functional traits were measured over the 2007–09 period, and niche parameters were derived from data previously published (Vincent et al. 2011) and reanalysed in this study.

Study Site and Field Measurements

The study was conducted in a lowland tropical rain forest at the Paracou experimental site (5° 18′ N, 52° 55′ W) in French Guiana. Rainfall averaged 2875 mm year−1 over the 1986–2005 period with a 3-month dry season (<100 mm month−1) from mid-August to mid-November. The 14 non-pioneer co-occurring species studied are common forest species in French Guiana (Table 1) and account for 27% of the total tree population (>10 cm diameter at breast height) at the Paracou experimental site.

Table 1. List of study species with family name [following the Angiosperm Phylogeny Group classification (Angiosperm Phylogeny Group 2009)] abbreviations used (genus initial followed by species initial) in the figures, number (N) of saplings per species and by light index (LI) class and diameter range (min-max). Light niche parameters are indicated: POP-RESP is a measurement of the correlation between abundance and degree of canopy openness and reflects the niche optimum; degree of specialization reflects species sensitivity to canopy openness and indicates species niche breadth; and Hmax indicates adult stature
FamilySpeciesAbbreviation N TotalDiameter range (mm)POP-RESPDegree of specialisation H max
LI = 1LI = 2LI = 3LI = 4
Fabaceae Bocoa prouacensis BP271430445.6–20−0.031.934
Fabaceae Dicorynia guianensis DG2421162636–20.20.001.252
Fabaceae Eperua falcata EF2425204735–18.50.031.744
Fabaceae Eperua grandiflora EG3324181765.2–17.2−0.062.142
Lecythidaceae Gustavia hexapetala GH201740418.3–18.30.041.820
Chrysobalanaceae Licania alba LA3121220745–17.20.001.231
Lecythidaceae Lecythis persistens LP2021121548.7–19.3−0.021.337
Annonaceae Oxandra asbeckii OA2024210655.9–17−0.082.218
Sapotaceae Pradosia cochlearia PC201051365.6–17.2−0.051.649
Vochysiaceae Qualea rosea QR2120140555.5–17.90.031.646
Lauraceae Sextonia rubra SR212540504.8–20.20.001.444
Clusiaceae Symphonia sp. 1 SS2225251736–16.60.002.626
Fabaceae Tachigali melinonii TM1725193645.4–160.083.535
Myristicaceae Virola michelii VM2330212763.1–23.20.072.841

To evaluate species-specific responses to different light regimes, an extensive search throughout the Paracou experimental station was conducted to identify suitable saplings (0.5–3 m tall) in all light regime classes. These saplings were to be located outside seasonally flooded areas, and any obviously resprouted stems were excluded (Table 2). In all, 41–76 saplings per species (total 844) were selected, tagged and mapped. All saplings and their light environments were measured annually from 2007 to 2009 (or from the date of first encounter, after 2007).

Table 2. Median values at low-light (LI = 1) and plasticity index for (g m−2), leaf life span (months), diameter growth rate (mm year−1) and height growth rate (cm year−1). Monotonic increases and decreases in functional trait median values with light level are indicated by sign ‘+’ and ‘−’, respectively. Plasticity index standard deviations (SD, calculated by bootstrapping) are given. The Kruskall–Wallis test was applied to functional trait values across the two light classes corresponding to maximum and minimum median values. Levels of significance in this test are shown with < 0.1, *< 0.05, **< 0.01, ***< 0.001
Species nameLeaf mass per area (g m−2)Leaf lifespan (months)Diameter growth rate (mm year−1)Height growth rate (cm year−1)
Median at low-lightPlasticity indexSD P Median at low-lightPlasticity indexSD P Median at low-lightPlasticity indexSD P Median at low-lightPlasticity indexSD P
Bocoa prouacensis 6973.7 1025434 0.2+0.90.7**4.29.34.6 
Dicorynia guianensis 41+172.1***2985*0.6+1.30.3***6.1+25.54.6***
Eperua falcata 52+122.1***86−5530**0.4+1.10.1***4.1+8.73.0***
Eperua grandiflora 74+122.8***121−5238**0.2+0.80.2***4.6+10.23.2**
Gustavia hexapetala 47+83.5*591515 0.5+1.00.3**3.1+10.03.1**
Licania alba 74+172.5***855036 0.3+1.10.2***0.5+26.66.6***
Lecythis persistens 87+153.4***1105167 0.3+0.70.2***9.5+0.64.3*
Oxandra asbeckii 64+112.8***71−3444*0.3+0.80.1***2.0+6.74.4*
Pradosia cochlearia 59+246.5**51−2814*0.4+0.10.3 1.6+7.31.5**
Qualea rosea 51+151.9***37−154**1.0+2.40.7***7.8+26.210.1***
Sextonia rubra 51+183.7*43145**0.3+1.10.4*2.0+19.85.7**
Symphonia sp. 1 64+113.7***56−256***0.9+0.90.2***7.920.04.0***
Tachigali melinonii 50+91.8***45−176***0.4+1.10.4***3.7+33.110.6***
Virola michelii 45+143.6***3258 0.7+2.20.6***2.6+40.39.1***

Light Measurement

The light environment of each sapling was evaluated during each census by two observers using a light regime visual estimate based on the structure of the vegetation above and around the sapling. We used a scoring system similar to Clark & Clark (1992) adapted to suit the forest structure at Paracou where 1 = no direct light, dense understorey; 2 = light understorey (some lateral light due to close by gap or thin upper canopy layer); 3 = significant direct illumination associated with position either on the edge of a large gap or well inside a small gap; 4 = abundant vertical illumination (large gap centre, track side). The mean of the two observers' scores was recorded for each census, and the average light environment for each sapling was described by calculating the mean light index value for all the different censuses.

The reliability of this index was assessed for a subset of individual plants by comparing values with two other methods. Although these methods are potentially more accurate, they were unsuitable for use with large data sets over rugged terrain.

The first method consisted in a direct measurement of incoming solar radiation at the top of the sapling using a LI-190 quantum PAR sensor. An identical censor located less than three km away at an open site recorded data continuously to a data logger (CR10X, Campbell scientific Inc) and thus provided a measurement of incoming radiation at the top of the canopy. Photosynthetically active radiation (% PAR) received by the sapling was estimated by calculating the ratio between incoming radiation measured at the top of the sapling and the incident radiation measured simultaneously in the open adjacent site. Measurements were made under an overcast sky (no direct sunlight) from April to August 2009 to avoid the temporal and spatial variability associated with sunflecks.

For the second comparison, we used hemispherical photography (Jennings, Brown & Sheil 1999). Here, digital hemispherical photographs were taken above selected saplings (180° fisheye lens, Nikkor 8 mm f/2.8) before sunrise. The resulting image was analysed by gap light analyser software (GLA) (Frazer, Canham & Lertzman 1999) to calculate the global site factor (GSF), that is, the fraction of total radiation received relative to that received above the canopy integrated over a year.

Light index (LI) was seen to be closely correlated with both % PAR (Pearson correlation coefficient = 0.78, d.f. = 115, < 0.001) and with GSF (Pearson correlation coefficient = 0.87, d.f. = 20, < 0.001).

Whole-Plant Functional Traits

Diameter and height growth rates (GRDIA and GRHT)

Stem diameter was measured using Vernier callipers (precise to within one tenth of a millimeter) at a marked position on the stem 20 cm from the ground and in two orthogonal directions. The height of the main stem was determined using a measuring tape. Annual GR was assumed to be linear over the study period and was calculated as:

display math(eqn 1)

where G1 and G2 are diameter (mm) or height (m) at t1 (date of first census) and t2 (date of last census).

Annual GRs were further tested for sapling size effect and, when appropriate, corrected as follows: a linear model using LI (categorical variable) and initial diameter (or height) as predictors was fitted for each species. If the size effect was significant (< 0.05), the model was applied to the original data set replacing observed diameter (or height) by species median diameter (or height) for all individuals. LI was left unchanged. The residuals associated with the original data set were then added to model predictions. Adjusted GRs were used instead of the original observed GRs. The size effect was found to be significant in two species for diameter GR and in two species for height GRs. This size effect was in all cases weak, and the correction procedures only marginally affected the raw data values.

Plasticity in GRs

Median GRDIA and GRHT values were computed for three light environments (low-light LI = 1, medium-light LI = 2, high-light LI = 3) after rounding each individual Light Index score. As the 14 species were not evenly represented in LI–4 (Table 2), this light class was excluded from the analysis. Plasticity was quantified by the following index:

display math(eqn 2)

In most cases, maximum median GR was observed under high-light conditions and minimum median under low-light conditions. The sign of the trait variation following an increase in light levels is given in Table 3. The significance of the plasticity index was tested by Kruskall–Wallis test on whole-plant trait values observed in the two light classes corresponding to maximum and minimum median values. Standard deviations were computed for the plasticity index estimate using bootstrap resampling [boot package in r software (R Development Core Team 2011)].

Table 3. Spearman's correlation coefficients between POP-RESP, degree of specialization, Hmax, median values in low-light and plasticity index for leaf mass per area (LMA), leaf life span (LLS), diameter growth rate (GRDIA) and height growth rate (GRHT). Levels of significance are shown with *< 0.05, **< 0.01, ***< 0.001
 Leaf traitsWhole-plant traitsLight niche parameters
LMALI1LMA plasticityLLSLI1LLS plasticityGRDIALI1GRDIA plasticityGRHT LI1GRHT plasticityPOP-RESPDegree of specialisation H max
Leaf traits
 LMALI1           
 LMA plasticity0.03          
 LLSLI10.83***−0.30         
 LLS plasticity0.76*−0.240.95***        
Whole-plant traits
 GRDIALI1−0.64*0.03−0.75**−0.26       
 GRDIA plasticity−0.64*0.04−0.66**−0.71*0.48      
 GRHTLI10.10−0.300.140.050.14−0.01     
 GRHT plasticity−0.590.01−0.71*−0.76*0.530.84**0.41    
Light niche parameters
 POP-RESP−0.67**−0.20−0.51−0.460.55*0.75**0.020.66*   
 Degree of specialisation−0.13−0.68**0.030.010.13−0.070.000.170.19  
H max −0.330.64*−0.41−0.240.180.270.110.100.04−0.46 

Leaf Functional Traits

Leaf life span

Leaf censuses were conducted on the main axis of each sapling. In Oxandra asbeckii, which has no leaves on the main axis, we selected actively growing branches (i.e. avoiding lower most branches) that were sufficiently old to show some leaf abscission (most apical branches were therefore excluded). A leaf sequence was defined from the youngest leaf (fully expanded at the first census) to the oldest leaf found at the base of the axis. In the first census (conducted in June 2007, November 2007 or February 2008), a record was made of the number of leaves on each monitored axis and the position of the youngest leaf in the sequence was marked using coloured adhesive tape. The number of leaves remaining in each sequence was further recorded (in July 2008 then in July or November 2009), yielding a sampling period of 5–30 months depending on axis lifetime.

Leaf life span (days) was estimated for each individual plant as the ratio of leaf number (N) to leaf death rate (T):

display math(eqn 3)

This approach has been successfully applied to LLS estimations in previous studies (Southwood, Brown & Reader 1986; Ackerly 1996; Wright, Westoby & Reich 2002; Navas et al. 2003). The model (Little 1961), assumes a steady-state system, meaning that the axis must be in a process of active leaf production and loss. Because of the discrete leafing (flushes) of some species, leaf loss rate (T) was used instead of leaf arrival rate in this model. A final leaf population of 8625 leaves was used in the analysis, with an average of 616 leaves per species.

Leaf mass per area

Five punches were taken between the main veins of leaves with a core (diameter = 16 mm) of standardized area in July 2008. LMA (g m−2) was calculated from leaf punch dry mass (oven-dried for 96 h at 65 °C) and punch area.

Plasticity in leaf traits

We used the same method as described for whole-plant traits.

Functional Trait Analysis

Species median values under low-, medium- and high-light conditions and the species plasticity index of functional traits formed the basis of the analysis of cross-species trait relationships. We computed Spearman's correlation coefficient (noted rs) to evaluate ranking consistency among all the traits. The strength of the correlation between the various functional traits was dependent on light conditions, with correlations always being stronger for trait values measured under low- or high-light conditions (LI = 1 and LI = 3). One possible reason might be that the middle-light index class (LI = 2) was less homogeneous, notably by including most cases of unstable LI over the monitoring period. As most of the saplings were located under closed canopy and because the species under high-light conditions showed an unbalanced distribution, only the correlation results obtained for saplings growing under low-light conditions (LI = 1) are presented and discussed herein. Nevertheless, plastic variations in functional traits across light environments were taken into account by considering the plasticity index.

Maximum Adult Stature (Hmax)

This parameter was taken from Favrichon (1995) for Tachigali melonii and from Herault et al. (2010) for the 13 other species. Adult stature was used in this study as a surrogate of the ontogenetic change in light availability experienced by a given species. Species Hmax ranged from 20 to 52 m (Table 1).

Light Niche Characterization

Light niche parameters at the sapling stage of the 14 species examined in the study were derived from the data set and from results of a previous study conducted at the Paracou field station (Vincent et al. 2011). These authors investigated the relative roles of habitat specialization and dispersal limitation in shaping the spatial distribution of species. They provided a ranking of 49 tropical tree species based on how previous canopy disturbance affected the likelihood of saplings of each species being present. They quantified this sensitivity to canopy disturbance as the minus log of the odds ratios associated with a unit increase in log-distance to the nearest canopy disturbance. This index is termed POP-RESP (population response to canopy disturbance) in this study. POP-RESP ranged from −0.23 to 0.56 for the 49 species found to be sensitive to disturbance and ranged from −0.08 to 0.08 for the subset of species considered in this study (Fig. 1). Of the 14 species studied here, five had a value greater than 0, indicating that species abundance decreased with distance from disturbed areas, five had values below 0, indicating an opposite trend, and four had values not significantly different from 0, suggesting that disturbance had no – monotonic – effect on sapling abundance. POP-RESP was used as a proxy for the light niche optimum.

Figure 1.

Distribution of 49 French Guianan tree species according to POP-RESP value (minus log of OddDist from Vincent et al. 2011). The position and the range of the 14 study species is indicated by a black arrow.

Niche breadth was assessed by determining the degree of specialization to canopy disturbance regime. Hereafter, we use the expression ‘light niche breadth’ to refer to this variable degree of specificity in light requirement. We re-analysed the sapling inventory data collected by Vincent et al. (2011) to produce estimates of relative sapling abundance per class of disturbance. The original model was rerun after converting log-transformed distance to nearest canopy disturbance into a categorical variable (four-level factor corresponding to 1 <2.2 m, 2 <7.2 m, 3 <40 m and 4 >40 m from canopy disturbance area). The predicted relative frequency of quadrats containing the target species per disturbance class was used to compute the degree of specialization as the ratio of maximum to minimum relative frequencies. Lower degree of specialization values was indicative of a broader niche (Thompson, Hodgson & Gaston 1998).

To analyse the multivariate association of leaf, growth trait and niche parameters, we conducted a principal component analysis based on a correlation matrix of median values obtained for low-light traits and trait plasticity in all the species. The median value of GRHT was excluded from the correlation matrix as it did not show a species effect (tested by a one-way anova). Niche parameters (POP-RESP and degree of specialization) and Hmax were later correlated with the PCA axes. Statistical analyses were performed by r software (R Development Core Team 2011) on untransformed values of LMA, LLS, GRDIA and GRHT.

Results

Light Niche Characterization

Canopy disturbance had a non-significant effect on four species. Two of these (Symphonia species1 and Licania alba) showed a clear hump-shaped response (distance class effect P value <0.05), whereas the two others (Dicorynia guianensis and Sextonia rubra) did not (Fig. 2). None of the 14 species studied showed the monotonic and systematically decreased abundance with distance to canopy disturbance typical of pioneer species. Pattern variability in each response group (Fig. 2) indicated a great diversity of habitat specialization, with some species preferring intermediate light conditions (Symphonia species1 and L. alba). The different species also exhibited a variety of light niche breadths, with degree of specialization ranging from 1.2 to 3.5 (Table 1).

Figure 2.

Plot of relative sapling occurrence according to distance to canopy gap (as a categorical variable with four levels corresponding to 1 <2.2 m, 2 <7.2 m, 3 <40 m and 4 >40 m from canopy disturbance area) for 14 tropical tree species. Species name and degree of specialization (dg) are indicated above each plot. Note that the vertical axis log scale differs among species. Dotted red lines indicate 95% confidence interval (lines appear dark grey in print). Species-specific plots are grouped in three columns according to the sign of the population response to canopy disturbance (POP-RESP).

Across-Species Correlation between LLS and LMA

Median LMA values ranged from 41 to 87 g m−2 in low-light conditions, and median LLS values ranged from 29 to 121 months (Table 2). LMA was seen to be closely correlated with LLS (LI = 1, rs = 0.83, d.f. = 12, < 0.001, Table 3).

Across-Species Correlations between Leaf Traits and Whole-Plant Traits

Median GRs under low-light conditions varied more than fivefold across species for GRDIA (0.2–1 mm year−1) and 20-fold for GRHT (0.5–9.5 cm year−1) (Table 2).

Canopy openness increased diameter GR (as indicated by a ‘+’ before plasticity values in Table 2): the effect of LI on diameter GR (GRDIA) was significant for all species except Pradosia cochlearia but the strength of the growth response (quantified by a GR plasticity index) differed among species. GR plasticity was positively correlated with high-light diameter GR (rs = 0.85, d.f. = 12, < 0.001). Canopy openness also increased height GR for all species except Bocoa prouacensis (Table 2). Additionally, GRHT plasticity varied markedly (0.6–40.3 cm year−1) and was higher for saplings growing faster under high-light conditions (Spearman's correlation coefficient of 0.93 between GRHT plasticity and high-light GRHT).

Leaf mass per area was negatively related to diameter GR and to diameter GR plasticity (Table 3). Species with a faster and more responsive diameter GR had lower LMA. No significant relationship was detected between LMA and height growth strategy. LLS was significantly and negatively correlated with diameter GR and with diameter and height GR plasticity indices (Table 3).

Across-Species Correlations between Functional Traits, Light Niche Parameters and Adult Stature

POP-RESP quantified the combined effects of differential growth, recruitment and survival for a period of a decade after a disturbance in alternate light environments and was used as a proxy for light niche optimum. POP-RESP and LMA were significantly correlated (see Table 3 for Spearman's coefficient correlation, Pearson's correlation coefficient = −0.59, d.f. = 12, < 0.05) although no significant relationship was found between POP-RESP and LLS (Table 3). Diameter GR plasticity showed the strongest rank correlation with POP-RESP (rs = 0.75, d.f. = 12, P < 0.01) and explained 38% of POP-RESP variation (Pearson's correlation coefficient = 0.62, d.f. = 12, < 0.05, Fig. 3). The rank correlation between POP-RESP and height GR plasticity was less close than with diameter GR plasticity (rs = 0.66, d.f. = 12, P < 0.05), but the predictive value of GRHT plasticity was higher (Pearson's correlation coefficient = 0.69, d.f. = 12, < 0.01). The species studied showed differences in LMA and LLS plasticity (Table 2). LMA increased monotonically with LI for all species, while LLS decreased monotonically for seven of the 14 species (Table 2).

Figure 3.

Across-species correlations between canopy disturbance response (POP-RESP) and diameter growth rate (GRDIA) plasticity. Species are abbreviated as in Table 1. Regression line, coefficient of determination and significance level are shown.

Leaf mass per area plasticity was significantly correlated with degree of specialization (rs = −0.68, d.f. = 12, < 0.01) and with adult stature (rs = 0.64, d.f. = 12, < 0.05). It accounted for 25% of interspecific variations in degree of specialization (Pearson's correlation coefficient = −0.5, d.f. = 12, < 0.01) and for 38% of interspecific variations in adult stature (Pearson's correlation coefficient = 0.61, d.f. = 12, < 0.05). By contrast, LLS, height GR and diameter GR plasticity did not show any significant relationship with degree of specialization or adult stature (Table 3).

Associations among the traits were analysed by principal component analysis (Fig. 4). The first principal component analysis axis explained 59% of the variation and the second axis 17% of overall trait variation (Fig. 4). The first axis reflects components of the species' successional status: negative coordinates were indicative of the most shade-tolerant species showing lower GRs, higher LMA and LLS and a negative response to canopy disturbance; positive coordinates were indicative of species taking advantage of canopy disturbance with the highest and most responsive GRs and the lowest LMA and LLS values. The second axis (dominated by LMA plasticity) is less readily interpretable but seems to mirror light niche breadth (expressed in terms of degree of specialization) and to a lesser extent adult stature, with generalist and large-statured species at the bottom experiencing a wide range of irradiance regimes at the juvenile stage or along their life cycle and displaying high LMA plasticity.

Figure 4.

Plot of principal component analysis (PCA) ordination diagram showing following traits (arrows): median values in low-light and plasticity index for leaf mass per area (LMA), leaf life span (LLS), diameter growth rate (GRDIA) and plasticity index for height growth rate (GRHT). The first two axes jointly capture 76% of total variation in trait data. Population response to canopy disturbance (POP-RESP), degree of specialization (Deg. of. specialization) and Hmax were not included in the PCA but were subsequently correlated with the PCA axis.

Discussion

This study aimed to test whether the pattern of trait covariation observed previously at the entire community level of tropical moist forests held true among the subset of species with intermediate light requirements and whether median values and plasticity in functional traits were related to niche parameters. Study results show first that LLS was closely correlated with LMA, even within this limited range of species light requirements. Second, whole-plant traits (and to a lesser extent leaf traits) correlated with the light niche optimum. Third, LMA plasticity in response to light was predictive of light niche breadth and adult stature.

General Trends in Trait Covariations among Species with Intermediate Light Requirements

In spite of subtle differences regarding population responsiveness (POP-RESP), all 14 species studied showed differences at leaf and whole-plant levels that could be meaningfully linked with environmental factors. The interspecific variability noted in the leaf traits (twofold for LMA, four-fold for LLS) clearly demonstrates a marked functional diversity within the group of species studied. In comparison, Sterck, Poorter & Schieving (2006) reported that LMA varied by a factor of c. 4 and LLS by a factor of c. 10 among saplings of 50 sympatric tree species in a Bolivian rain forest spanning the whole shade-tolerance gradient. We found that the LMA-LLS axis, which is a major dimension in plant ecological strategy (Westoby et al. 2002), emerged in 14 tropical tree species with intermediate light requirements. The evolutionary coordination of LLS and LMA is believed to have led to a higher LMA in the most shade-tolerant species (Lusk et al. 2008) despite the carbon-balance advantages of a low LMA in shaded conditions (lower construction costs per area and lower maintenance costs). However, we also observed that the relatively long LLS of two species (Eperua falcata and Gustavia hexapetala, see Fig. S1 in Supporting Information) was associated with low to medium LMA. This may be explained by the differential contribution of LMA components to LLS. LMA can be described as the product of lamina depth and tissue density. Long LLS may therefore be achieved by a medium LMA through low leaf depth and relatively high tissue density or a large proportion of structural components that provide protection against herbivores and physical stress. Lusk et al. (2010) reported that leaf cell wall fraction and punch strength were more robust correlates of juvenile light requirement than LMA. And additionally to mechanical resistance, long LLS can result from a great diversity of defence strategies: to date, no consistent syndrome of defensive investment in terms of chemical compounds, leaf toughness (component of LMA) or leaf nitrogen content has been found (Fine et al. 2006).

Leaf traits and whole-plant traits provide a coarse yet consistent ranking of species in terms of degree of shade tolerance. To date, close correlations among functional traits have been established in most cases across different ecosystems (Reich, Walters & Ellsworth 1992). Regarding tropical moist forests, such trait-based approaches have been applied at the community level among species spanning the whole shade-tolerance gradient (Poorter & Bongers 2006) although (Wright et al. 2003) showed in a Panamanian tropical forest that few species are located at the extremities of the gradient and that most species have intermediate light requirements. We demonstrated here in this study that the syndrome of traits driving the demographic trade-off involved in the niche-based theory of coexistence held indeed among a set of 14 species from which all pioneer and truly shade-tolerant species had been excluded. As expected, we found that species with higher LMA and longer-lived leaves displayed lower diameter GR and lower responsiveness to canopy gap at both the whole-plant and population levels (Fig. 4), thus suggesting an overall adaptation to an understorey habitat. For example, long LLSs are required under low-light conditions to pay back the investment made in leaves. This is apparently achieved by a high LMA that should limit GR and responsiveness to gap opening. Conversely, species with lower LMA and short-lived leaves have a faster diameter GR and higher responsiveness to canopy openness. Low LMA and fast turnover of plant parts permit an efficient light capture and a more flexible response to light availability. Our study detected few significant correlations between leaf traits and height GR across species. This result is not entirely consistent with the study by Poorter & Bongers (2006) involving 53 pioneer and shade-tolerant rain forest tree species at the sapling stage, which reported a significant linear trend between height GR and LMA (r2 = 0.19, < 0.01) and LLS (r2 = 0.19, < 0.001). Height GR might be less relevant in the present study because of the absence of pioneer species for which height growth is crucial to outcompete neighbours and survive.

Prediction of Light Niche Parameters and Ecological Significance of Plasticity

Whole-plant traits correlated more closely with light niche optimum than did leaf traits (Table 3). Height GR plasticity accounted for 49% of light niche optimum (POP-RESP). Contrary to the hypothesis put forward by Violle & Jiang (2009), we observed that whole-plant traits plasticity was related to niche optimum, not to niche breadth. This result is consistent with previous findings among non-pioneer species (Poorter & Arets 2003) and among pioneer and shade-tolerant rain forest tree species (Valladares et al. 2000). One possible explanation might be that species adapted to low-resource habitats may be less plastic in traits directly related to fitness, for example, GR and this was a by-product of specialization to this environment (Lortie & Aarssen 1996). Shade-tolerant species, which are unable to make full use of abundant light and have a low average GR, are expected to show smaller differences in GR between high- and low-light environments. Therefore, restricting performance traits to vegetative biomass, reproductive output and plant survival, as proposed by Violle et al. (2007) and as applied in our study, clearly showed its limits: it might be more relevant to consider diameter and height GRs as performance traits rather than functional traits.

Median LMA accounted for 35% of the interspecific variation in light niche optimum (POP-RESP). In a study, over 53 tree species spanning the whole range of shade-tolerance gradients, Poorter & Bongers (2006) found specific leaf area (1/LMA) (r2 = 0.50) to be more predictive than we did. This difference in LMA predictive capacity might result from the wider range of LMA and niche optimum values investigated in the Poorter & Bongers study (2006). Again, in contrast with this study, we found LLS and light niche optimum to be only marginally correlated (rs = −0.51, < 0.06), perhaps for the same reason, that is, that the variance in niche optima in our study was limited, thus reducing the power of the analysis. Predictions of plant or population performance along an environmental gradient might be further improved by considering traits related to stem economic spectrum (Chave et al. 2009) and shown to vary independently of leaf economic spectrum (Baraloto et al. 2010). (Poorter et al. 2010) demonstrated that wood density was also linked with growth and survival in large rain forest trees and that wood spectra were related to regeneration light requirement. Wood density is likely to shape plant functioning through its impact on carbon gain, biomechanical and hydraulic safety and defence. In support of this, Herault et al. (2011) found that wood density was a significant predictor of ontogenetic variation in diameter GR among 50 rain forest tree species in French Guiana.

As expected, the study described herein provided evidence of a negative relationship between LMA plasticity and degree of specialization along the light gradient (rs = −0.68, P < 0.01), despite the exclusion of gap and understorey specialists from our subset of species. Species with a low degree of specialization (occurring with an even frequency over the light gradient) exhibited greater LMA plasticity than species more specialized at one end of the light gradient. This underlines the adaptive value of LMA response to light level. One hypothesis underlying this adaptive response is that interception is optimized in low-light through a greater leaf area per unit leaf biomass; under high-light conditions, photosynthesis rates are increased by greater leaf biomass for a given unit of leaf area. This has been corroborated by comparisons between sun and shade leaves (Onoda, Schieving & Anten 2008); the results showed sun leaves to contain larger amounts of palisade mesophyll associated with a higher photosynthetic capacity (Niinemets 1997). Overall, this finding suggests that LMA plasticity might enable species to enlarge their light niche breadth. In the past, the literature on LMA and other leaf trait plasticity in tropical trees mainly focused on the correlation with light niche optimum. These studies (Valladares et al. 2000; Rozendaal, Hurtado & Poorter 2006; Lusk et al. 2010) did not result in a clear consensus but postulated that plasticity is greater in pioneer species as these experience more pronounced and more predictable light variations. This lack of consistency might stem from differences between studies in terms of ontogenetic stage (plasticity is likely to depend on ontogenetic stage, see (Thomas & Winner 2002), observational conditions (field versus controlled environment) and methods of plasticity quantification. These conflicting results may also suggest that there is no significant relationship between LMA plasticity and light niche optimum. Our findings point towards phenotypic plasticity in functional traits that warrant further investigation in relation to niche breadth. The present study also supported the working hypothesis that leaf plasticity in response to light level is higher for species subjected to a major ontogenetic change in light availability, reflected by high adult stature (Table 3). This result suggests that LMA plasticity expressed at the sapling stage may reflect adaptation to the spatial heterogeneity of light conditions encountered both at the sapling stage and along several ontogenetic stages.

Conclusion

This study provides insights into the spectrum of trait values found under natural conditions in a set of 14 co-occurring rain forest tree species from which all pioneer and truly shade-tolerant species had been excluded. The LMA–LLS differentiation axis captured important features of adaptation along the gap-understorey continuum and provided an accurate prediction of whole-plant traits. We confirmed that in spite of the variety of trait combinations, tropical tree species with intermediate light requirements can be ranked along a continuum of leaf traits that scale with response to canopy disturbance (Wright et al. 2003). Also, in line with the assumptions made by Violle & Jiang (2009) and with recent articles promoting the ecological significance of intraspecific variability and functional trait plasticity (Berg & Ellers 2010; Albert et al. 2011; Violle et al. 2012), this study clearly demonstrated the usefulness of taking functional trait plasticity into account when quantifying the niche parameters of tropical tree species over environmental gradients.

The functional equivalence hypothesis is the keystone of neutral theory (Hubbell 2005): it states that species differences are independent of traits influencing their fitness or their demography and their interactions with the biotic and abiotic environment (Leibold & McPeek 2006). Our findings did not support this hypothesis, at least for tropical tree species, as we found that leaf traits, GRs, growth response to light and spatial distribution correlated significantly with respect to light across species with intermediate light requirements. Hence, differences among species, reflecting the majority of tropical tree species, in terms of light requirement, were related to differential responses to environmental heterogeneities that are expected to promote coexistence by stabilizing mechanisms (Chesson 2000). Our results constitute a significant step towards validation of the assumptions made in the niche-based theory of coexistence in the whole community of tropical forests and underline the important role played by canopy disturbance in promoting the coexistence of tropical tree species, as reported by (Molino & Sabatier 2001). By characterizing species distribution along an environmental gradient from functional traits, this approach provides a basis for predicting the effects of human- or climatic-induced changes of canopy disturbance regimes on species assemblages and thus potentially contributes to ‘rebuilding community ecology from functional traits’ (Mc Gill et al. 2006).

Acknowledgements

This project received financial support from a European institution (project DYGEPOP, Population dynamics and managements of trees in French Guiana, convention no. 2828 dated 24/10/08 from European Regional Development Fund). We are grateful to Lourens Poorter, Nick Rowe and Daniel Sabatier for their helpful comments on an earlier version of this manuscript. This is a publication of Laboratoire d'Excellence CEBA (ANR-10-LABX-25).

Ancillary

Advertisement