How Do Variation and Covariance of Leaf Functional Traits Influence Schinus terebinthifolia Raddi (Anacardiaceae) Acclimation to Light and Water Availability in Tropical Dry Ecosystems?

Abstract

Light availability in tropical forests varies spatially and temporally, strongly influencing plant acclimation. Understanding variation and covariation among functional traits associated with photoacclimation is essential for predicting plant responses to environmental change. Here, we investigated acclimatory responses of Schinus terebinthifolia Raddi (Anacardiaceae), a widespread Neotropical species adapted to heterogeneous light environments. We evaluated variation and covariation in morphological, anatomical, physiological, and nutritional traits under contrasting light conditions. Under high light, plants invested more resources in palisade parenchyma and subepidermal layers while maintaining water-use efficiency, indicated by higher δ¹³C values. Irregular adaxial cuticles and unstacked thylakoid membranes were also observed under high irradiance. The strongest covariation occurred among anatomical traits, especially spongy parenchyma and adaxial and abaxial cuticles. Overall, the relationship between trait variation and covariation was slightly negative but not significant, although patterns differed among functional groups. These findings demonstrate that photoacclimation in S. terebinthifolia involves coordinated functional strategies that optimize light modulation, water conservation, and photosystem II performance under variable tropical light environments.

1. Introduction

Variation in light quality and quantity is fundamental to the maintenance of several physiological processes in plants, such as photosynthesis, morphogenesis, phototropism, and flowering [1,2]. However, high visible light and/or UV radiation intensities can negatively affect plant performance [3]. Water availability is another crucial environmental factor that influences plant growth and survival. Water deficit can lead to stomatal closure, reduced Rubisco (ribulose-1,5-bisphosphate carboxylase/oxygenase) carboxylation efficiency, and increased production of reactive oxygen species (ROS), which compromise photosynthetic capacity and overall plant development [4,5].

Plants acclimate to temporal and spectral light variation and fluctuations in water availability by adjusting functional traits of different trait groups—morphological, anatomical, physiological, and nutritional [6,7]. Leaves are considered one of the best plant organ study models due to their rapid capacity to adjust to environmental variation, particularly in light and water availability [8].

Several leaf functional traits respond similarly to variation in light intensity and water availability, revealing integrated acclimation mechanisms to multiple stresses. Among these shared traits, the thickening of photosynthetic tissues, such as palisade parenchyma, stands out, as it contributes to both dissipating excess light and reducing water loss [9]. Increased deposition of cuticular waxes and the presence of trichomes are also common strategies that simultaneously enhance radiation reflection and limit transpiration [10]. An increase in the carbon isotopic ratio (δ¹³C) further indicates convergent responses, reflecting greater water-use efficiency under high irradiance and low relative humidity [7].

Additionally, an increase in leaf mass per area (LMA) is a typical response to both high light levels and water-restricted conditions. This reflects a conservative resource-use strategy that promotes greater structural robustness and leaf longevity in stressful environments [11]. On the other hand, responses more specific to light exposure include the reorganization of thylakoid membranes and an increase in the number of plastoglobules within chloroplasts—mechanisms that protect the photosystems, particularly PSII, from photooxidative damage [12].

Plants respond to environmental variation, such as changes in light intensity and water availability, by expressing new phenotypic traits. This process is termed phenotypic plasticity when it occurs within the same individual or clone [13], or intraspecific phenotypic variation when observed across multiple individuals of a species [14]. Although phenotypic variation depends partly on plasticity, as multiple phenotypes can arise from a single genotype [15], it is also shaped by long-term exposure to specific environmental pressures, such as persistent drought or light regimes, which may drive adaptive evolution [16].

Phenotypic covariation is another critical factor impacting the rate and manner in which species acclimate to environmental factors [17]. Covariation was initially understood to be a phenomenon that is negatively associated with phenotypic variation or plasticity [18]. However, recent studies have shown that such relationships may be positive, which can enhance the acclimation process of plants, particularly in stressful environments [7,19]. Most studies on covariation between functional traits have been conducted across multiple species or at the community level [20,21], whereas patterns of intraspecific covariation are a preliminary line of analysis regarding the combined expression of different traits [22].

An example of the importance of analyzing intraspecific covariation is that it gives insights into the functional relationships among traits involved in the leaf economic spectrum [23]. Within a leaf module, these relationships often reflect negative covariance due to resource allocation constraints. For instance, leaves with higher specific leaf area (SLA) tend to be thinner and less dense, which favors rapid resource acquisition at the cost of lower mechanical resistance and longevity [24]. Similarly, high photosynthetic capacity is typically associated with shorter leaf lifespan, as investment in rapid carbon gain trades off against tissue durability [25]. These trade-offs underpin the leaf economic spectrum, influencing plant responses to environmental gradients and resource availability [26].

Analyzing how plant functional traits vary or covary is crucial in the current context of climate change and rapidly evolving edaphoclimatic conditions [16]. Understanding these dynamics of phenotypic relationships and expression within vulnerable environments, such as dry tropical forests, is even more relevant, as it may help identify species that are tolerant or sensitive to climate change [27]. In this context, such studies must be applied to species that are widely present in tropical environments, such as Schinus terebinthifolia Raddi. (Anacardiaceae) [28]. S. terebinthifolia is considered an important stabilizer of pioneer environments in tropical habitats, as it creates suitable conditions for the later establishment of species in subsequent successional stages [29]. Although classified as heliophilic, S. terebinthifolia can also occur in shaded and water-limited environments, demonstrating a broad capacity to acclimate to both light and water availability—an ability that has yet to be thoroughly investigated in an integrated manner. It is important to note that in other regions of the world, this species is considered an aggressive invasive species with rapid growth, forming monodominant islands and causing severe ecological imbalances [30,31]. Thus, studies on how the functional traits of this species vary and/or covary have far-reaching relevance. It is important to acknowledge, however, that S. terebinthifolia exhibits significant genetic diversity, and distinct lineages may show variations in trait expression. While global studies often highlight its aggressive invasive potential, our research focuses on populations within their native range in Brazil. Therefore, we interpret our findings within this specific evolutionary context, acknowledging that trait-based patterns observed here may reflect site-specific adaptations and could differ from those found in non-native lineages.

The present study aimed to understand how S. terebinthifolia adjusts its leaf functional traits across different trait categories (cellular, morphological, anatomical, physiological, and nutritional) in tropical forest environments with contrasting light and water availability. The following specific research questions were established to address this central objective: Which functional traits exhibit variation and covariation? How do these functional traits contribute to acclimation to light and water availability? How can the linear relationship between variation and covariation help elucidate these processes? We hypothesize that despite the high degree of variation observed in certain functional traits in response to environmental heterogeneity, traits with broad functional roles may exhibit multiple covariations. Therefore, this study sought to contribute to a deeper understanding of the strategies that S. terebinthifolia uses to cope with environmental stress by clarifying how interactions between the variation and covariation in functional traits influence plant acclimation to light and water availability.

2. Materials and Methods

2.1. Study Site

This study was conducted in the Atlantic Forest biome. Although this biome is recognized for its humid forests, our research was conducted in a dry ecosystem, specifically in coastal sandbanks [32]. These environments extend along the Brazilian coastline and were formed by sea regression and/or the deposition of fluvio-marine sediments [33]. Coastal sandbanks are characterized by predominant aridity, with water availability limited to precipitation pulses [34], and soil with sandy and saline characteristics and a low capacity for water and nutrient retention [7].

This study was performed in the coastal sandbank area of “Reserva Particular do Patrimônio Natural Fazenda Caruara” (Private Natural Heritage Reserve Caruara Farm), which is considered the largest remaining coastal sandbank ecosystem in the northern region of Rio de Janeiro State, Brazil (21°43′42.18″ S, 41°1′34.25″ W), covering approximately 4800 hectares (Figure S1A,B). Although this region has a history of past land use, the specific sites where S. terebinthifolia individuals were sampled are characterized by an advanced successional stage (approximately 15–20 years of natural regeneration), indicating that the sampled individuals are well established within the successional dynamics of the reserve. The reserve includes diverse vegetation formations, such as beach grass, beach grass and shrub, Clusia shrub, and forest [34] (Figure S1C–F). This study was conducted in the beach grass and shrub and forest formations, where S. terebinthifolia occurs naturally under contrasting conditions of light and water availability (Figure S1C,F). The beach grass and shrub formation is a drier environment with higher light availability due to its sparse vegetation composed of scattered shrubs that do not form a continuous canopy (Figure S1D). In contrast, the forest formation is a more humid environment with lower light availability due to denser vegetation and tree species reaching up to five meters in height, which form a consolidated canopy (Figure S1E). Hereafter, the beach grass and shrub and forest formations will be referred to as dry–open and humid–shaded environments, respectively.

2.2. Climatic Conditions of the Study Site

The distance between the dry–open and humid–shaded environments is approximately 1 km (Figure S1C). Despite the proximity of the two study locations, their microclimatic conditions contrast markedly. Photosynthetically active radiation (PAR µmol·m⁻²·s⁻¹), temperature (°C), and vapor pressure deficit (VPD KPa) were higher in the dry–open environment, while humidity (%) and canopy cover (%) were higher in the humid–shaded environment (

Table 1. Paired analysis of microclimatic and phytosociological aspects of S. terebinthifolia in dry–open and humid–shaded environments. Means (±standard deviation) for the two environments were compared by t-test, with significant differences (p < 0.05) indicated in bold.

Variable		Dry–Open (Mean ± SD)	Humid–Shaded (Mean ± SD)	t-Test (p < 0.05)
Microclimate	PAR (μmol·m−2·s−1) a	1490.0 ± 395.4	49.7 ± 16.6	t(24.08) = −18.2 p < 0.001
	Temperature (°C) a	29.7 ± 0.7	28.2 ± 0.5	t(48) = −8.66 p < 0.001
	Humidity (%) a	35.9 ± 1.1	38.7 ± 1.1	t(48) = 9.11 p < 0.001
	Vapor Pressure Defict-VPD (Kpa) b	2.7 ± 0.2	2.3 ± 0.1	t(48) = −8.76 p < 0.001
	Canopy coverage (%) a	0	58.54 ± 1.5	t(48) = 25.2 p < 0.001
Species	Species habit	Shrub	Tree	-
	Species height (m) c	0.5	3	-
	Cover Value Index d	70.80	10.94	-
	Relative frequency (%) d	21.1	5.71	-

^a Values were recorded at 1.30 m above ground level in each cardinal direction around individuals, between 11:30 and 12:00 h. Temperature and humidity were measured with a thermohygrometer (HT-30, Instrutherm, São Paulo, SP, Brazil); irradiance was measured with a radiometer (LI-250A, Li-Cor Inc., Lincoln, NE, USA); and canopy cover was measured with a spherical densiometer (Model-A, Forestry Suppliers, Inc., Jackson, MS, USA). ^b VPD was calculated as es/ea (es: saturated vapor pressure of the air; ea: vapor pressure of the air). ^c Individual height was measured based on the length of the pruning pole, which was 1.80 m. ^d Cover Value Index and relative frequency were obtained from [34].

). According to the Köppen classification, the climate of the region is tropical with dry season in the austral winter (Aw) [35]. However, the climate is being reclassified as sub-humid or semi-arid due to temperatures above 24 °C, reduced rainfall, and increased evapotranspiration [36]. Annual precipitation ranges from 500 to 1200 mm, and mean annual temperatures range from 20 to 30 °C, with the months between May and August being the driest and coolest [7,34].

2.3. Study Species

Schinus terebinthifolia Raddi. (Anacardiaceae) was chosen as a model species because of its significant ecological role in Brazilian coastal sandbank environments, where it acts as a nucleating species, creating suitable microclimates for the germination of seeds of other species. It also has high coverage and relative frequency in both dry–open and humid–shaded environments (Table 1). Individual plants of S. terebinthifolia exhibit a shrub growth form with a mean height of 0.5 m in the dry–open environment and a tree form with a mean height of 3 m in the humid–shaded environment (Table 1; Figure S2). Five fully expanded and healthy leaves located in the upper portion of the plant canopy were chosen and collected from each of five adult individuals in each environment. Botanical material was analyzed and collected during the dry season of 2012.

2.4. Morphological Traits

Morphological traits were measured on the five leaves collected from each of the five individuals of the dry–open environment and of the humid–shaded environment, totaling 25 samples per environment and 50 samples overall. Discs measuring 0.5 cm in diameter (d) were removed with a cork punch from the middle portion of the leaves, while avoiding vein regions. Disc area was determined as Area_disc = π·r², where r = d/2. The disks were kept for 24 h in Petri dishes with distilled water until they reached maximum saturation. Saturated mass (M_saturated, g) was measured using a digital balance with 0.0001 g precision (AY220, Shimadzu Corporation, Kyoto, Japan), while thickness (Thic, mm) was measured immediately thereafter using an electronic digital caliper with ±0.02 mm precision (Stainless Hardened, Findmall, Walnut, CA, USA). The discs were then dried in an oven at 60 °C for 72 h (TE-393/1, Tecnal, Piracicaba, SP, Brazil) and weighed again to determine dry mass (M_dry, g). The following traits were calculated from these parameters: leaf mass per unit area (LMA, g·m⁻²) = M_dry/Area_disc [37]; leaf saturated water content (LWC, g·m⁻²) = (M_saturated − M_dry)/Area_disc [37]; leaf density (Den, mg·mm⁻³) = LMA/Thic [38]. Leaf area (LA, cm²) was measured on the same leaves by scanning them at 300 dpi resolution using an Epson Perfection V19 (Nagano, Japan) scanner, followed by analysis with ImageJ software (version 1.52a) [39].

2.5. Anatomical Traits

Anatomical traits were measured on sections taken from the middle third (central region between the base and tip) of the same leaves used in the morphological analysis. Sections were taken by freehand with a razor blade, mounted on slides with 50% glycerin, and covered with cover-slips. The slides were observed and images captured using an image capture system (Moticam Pro 282B, Hong Kong, China) coupled to a light microscope (Axioplan, ZEISS, Oberkochen, Germany). Five different fields were captured from each of the 50 prepared slides from which the thickness of adaxial (Adep) and abaxial (Abep) epidermis; adaxial (Adcut) and abaxial (Abcut) cuticles; subepidermal layer (Subep); and palisade (Pal) and spongy (Spon) parenchyma were measured using Image Pro Plus 4.5 software (Media Cybernetics, Silver Spring, MD, USA).

2.6. Chlorophyll a Fluorescence

Chlorophyll a fluorescence emission was analyzed to determine changes in quantum yield of photosystem II (PSII). Five completely expanded leaves, without signs of necrosis and from the same branches as those selected for morphological and anatomical analyses, were chosen, for a total of 25 samples per environment and 50 samples overall. The SPAD index (CCM-200 Plus, Opti-Sciences, Inc., Hudson, NH, USA) was previously determined to avoid leaves with heterogeneous chlorophyll content, thus avoiding the inclusion of physiologically atypical leaves. The leaves were adapted to the dark for 30 min with clips to dark-adapted samples so the quinone pool reached the oxidized state. Dark-adapted leaf regions were initially exposed to light (6 μmol·m⁻²·s⁻¹ at 660 nm) to measure initial fluorescence (F₀), followed by a saturating flash of light (3000 μmol·m⁻²·s⁻¹) to measure maximum initial fluorescence (F_m). An actinic light was then applied to measure the steady-state value of fluorescence (F_t) before another saturating flash was used to measure final maximum fluorescence (F_m’). After turning off the actinic light, a new pulse of far-red light was applied to measure the final minimum fluorescence (F₀’). These variables were used to estimate the following: maximum quantum yield of PSII: F_v/F_m = (F_m − F₀)/F_m; photochemical quenching: qP = (F_m’ − F_t)/(F_m’ − F₀’); and non-photochemical quenching: NPQ = (F_m − F_m’)/F_m’ [40]. These analyses were performed at midday with a portable fluorometer (OS5p+, Opti-Sciences, USA).

2.7. Photosynthetic Pigments

The content of total chlorophyll and carotenoids was determined using leaf discs (0.5 cm diameter) taken from the same regions of the leaves where the chlorophyll a fluorescence analyses were performed. The disks were kept in plastic tubes containing 5 mL of dimethylsulfoxide (DMSO) and protected from light to avoid oxidation of photosynthetic pigments. After four days, 1 mL of the supernatant was removed from the total extract and read using a spectrophotometer (Shimadzu 1240, Japan) at absorbances of 480, 649, and 665 nm. Total chlorophyll (Chlo) and carotenoid (Car) concentrations were determined using the formulas proposed in [41].

2.8. Nutritional Traits

Nutritional analyses were performed on the same five leaves as previous analyses. The isotopic compositions of δ¹³C (‰) and δ¹⁵N (‰), N¹⁴ (N) and P³⁵ (P) elemental, and the C/N ratio were determined. The leaves were dried at 60 °C for 72 h in a drying oven (TE-393/1, Tecnal, Brazil), crushed in a Willey knife mill (SL-31, Solab, Piracicaba, SP, Brazil), and then macerated to avoid very granular samples. Isotypic concentrations of δ¹³C, δ¹⁵N, and N¹⁴ and the C/N ratio were determined by placing approximately 1.5 mg of each sample in its own tin capsule, followed by analysis using a mass spectrometer (Thermo Finnigan Delta V Advantage, Thermo Fisher Scientific, Waltham, MA, USA) coupled to Flash 2000 Organic Elemental Analyser (Thermo Fisher Scientific, Waltham, MA, USA). Pee Dee Belemnite (PDB) and atmospheric N were used as standard values for C and N, respectively. The analytical precision was ±0.1‰ for δ¹³C and ±0.2‰ for δ¹⁵N, and the certification standard (Protein OAS/IsotopeCert 114859; Elemental Microanalysis, Okehampton, UK) was used. The remaining portion of the samples was used to determine P³⁵ concentration. The analyses were carried out through the Universidade Federal Rural do Rio de Janeiro, Brazil, using the methodology established in [42].

2.9. Microscopic Analysis

Microscopic analysis was carried out to observe qualitative variation in anatomical, micromorphological, and ultrastructural aspects of leaves between environments. Five leaves were randomly chosen from among the five individuals in the dry–open environment and the five individuals in the humid–shaded environment (N_area = 5 leaves, N_total = 10 leaves).

The leaves were fixed in a solution containing 2.5% glutaraldehyde, 4% formaldehyde, and 0.05 M sodium cacodylate buffer with pH = 7.2 [43]. The samples were then subjected to three washes of 45 min each in the same buffer before being subjected to post-fixation with 1% osmium tetroxide and 0.05 M sodium cacodylate buffer for one hour. The samples were rewashed in the same buffer to remove excess osmium and dehydrated in an increasing ketone series (10%, 30%, 50%, 70%, 90%, and 100% three times) for 1 h at each step. After dehydration, the samples destined for light and transmission electron microscopy were infiltrated and included in Epoxy resin (Epon^®, Leica Microsystems, Wetzlar, Germany) and kept in a drying oven at 60 °C for 72 h to polymerize and obtain blocks. The blocks were subsequently sectioned by ultramicrotome (Reichert Ultracuts, Leica Instruments, Germany) using a diamond knife (Diatome^® Ultra 45°, Biel/Bienne, Switzerland). For light microscopy, semi-thin sections of 0.70 µm were obtained and distributed on slides, stained with toluidine blue and 1% borax buffer for 1 min on a hot plate at 60 °C, and mounted on glass slides and covered with cover-slips using Entellan^® (Merck KGaA, Darmstadt, Germany). Observations were made using a light microscope (Axioplan, ZEISS, Oberkochen, Germany) coupled to an image capture system (Moticam Pro 282B, Hong Kong, China). For transmission electron microscopy, ultrathin sections of 50 nm were obtained and collected on 300 mesh copper grids, which were subjected to post-contrast in 5% uranyl acetate for 40 min and 1% lead citrate for 5 min. The samples were analyzed using a transmission electron microscope (JEM 1600 Plus, Jeol Ltd., Akishima, Tokyo, Japan) with an accelerating voltage of 80 KV, and images were captured using a CCD digital camera (Gatan Bioscan Camera, Model 792, Gatan, Inc., Pleasanton, CA, USA).

For scanning electron microscopy, dehydrated samples were subjected to CO₂ critical point drying (CPD 030, BAL-TEC AG, Balzers, Liechtenstein) before being mounted on stubs with double-sided carbon tape for metallization with palladium gold (SCD 050, Baltec, Switzerland), forming a layer approximately 20 nm thick. The samples were analyzed using a scanning electron microscope (EVO 40, Zeiss, Germany) at an accelerating voltage of 15 KV.

2.10. Statistical Analyses

Statistical analyses were conducted using the R version 4.5.1 programming language and the RStudio version 2025.05.0 code editor [44]. The constructed scripts are detailed in Figure S3.

Initially, the dataset was tested for normality by environments with the Shapiro–Wilk test, using the “byf.shapiro” function of the “RVAideMemoire” package [45]. Homogeneity of variances by groups was tested by Levene’s test, using the “leveneTest” function of the “car” package [46]. Leveraging outliers were identified by groups, with environment (dry–open and humid–shaded) as a grouping factor, using the “identify_outliers” function of the “rstatix” package [47]. Outliers considered to be leveraged were excluded from the dataset.

Traits that presented a non-parametric distribution and heteroscedasticity were transformed by a base 10 logarithmic function (Log₁₀) to approximate a normal distribution, with correction through adjustments of degrees of freedom using the Welch test. Differences between mean values of functional traits for dry–open and humid–shaded environments were evaluated by the t-test (p < 0.05) for two independent samples, using the “t.test” function of the “stats” package [48].

The effect size of trait variation across functional groups was quantified by employing Hedges’ g, a standardized measure that corrects for small sample bias, using the “effectsize::hedges_g” and “effectsize::interpret_g” functions of the “effectsize” package [49]. Application of Hedges’ g in this context allows for a nuanced understanding of phenotypic variability, aligning with methodologies employed in previous studies.

Hedges’ g is calculated as follows:

$$g={\displaystyle \frac{X_1-X_2}{s^{*}}}x\left(1-{\displaystyle \frac{3}{4\left(N-2\right)-1}}\right)$$

where X₁ and X₂ are the means of the two groups being compared, and s* is the pooled standard deviation, which accounts for within-group variability, calculated as follows:

$$s^{*}=\sqrt{{\displaystyle \frac{\left(n_1-1\right)s_1^2+\left(n_2-1\right)s_2^2}{n_1+n_2-2}}}$$

where $s_1^2$ and $s_2^2$ are the variances of each group, and $n_1$ and n₂ are the sample sizes.

The correction factor $\left(1-{\displaystyle \frac{3}{4\left(N-2\right)-1}}\right)$ adjusts g to prevent the overestimation of effect sizes for small samples, where N is the total number of observations (in this study, N = 50 with 25 per environment).

Principal Component Analysis (PCA) was employed to identify the main contributing traits and explore correlation patterns underlying plant acclimation to spatial heterogeneity in light and water availability. Initially, a covariance matrix was constructed, where n represents the number of traits analyzed per individual (n = 10 in the present study) and k the total number of trait observations included in the analysis (k = 23). From this matrix, eigenvalues were estimated using the “eig.val” function of the “factoextra” package [50] to obtain a set of independent vectors that best explain the variability in the PCA dimensions (Table S1). The explanatory power of each PCA dimension was visualized by creating a scree plot using the “fviz_eig” function of the “factoextra” package [50], taking variance and eigenvalues into account (Figure S4). Based on this analysis, the dimensions Dim1 and Dim2 were selected for further interpretation. A covariance matrix was then built using the “corrplot” function of the “corrplot” package [51] to assess how each functional trait relates to Dim 1 and Dim 2 (Figure S5). The five traits most correlated with Dim1 and Dim2 (total = 10 traits) were selected using the “fviz_contrib” function of the “factoextra” package [50], in line with the general rule established in [51], which requires at least five observations per independent variable [52] (Figure S6). Finally, a PCA biplot was constructed using the “fviz_pca_biplot” function of the “factoextra” package [50], with a confidence ellipse (confidence interval = 0.95) added using the “ellipse.level = 0.95” function of the same package to illustrate the formation or overlap of clustering groups.

Visualization of covariance among functional traits was made clearer by constructing an interaction network with Pearson correlation values (p < 0.05) using the “qgraph” function of the “qgraph” package [53], selecting a minimum cut-off point for correlations above 0.65 using the “minimum = 0.65” function of the same package. This specific threshold was implemented to exclude weak, potentially spurious correlations (noise) while retaining moderate-to-strong biological interactions that represent true functional coordination, thereby preventing network over-fragmentation. Pearson’s correlation coefficient was used because it standardizes the relationship between two covarying traits that differ in measurement scale, thus allowing direct comparisons across different traits.

Pearson’s correlation coefficient (p < 0.05) was also used to construct a correlation matrix using the “corrplot” function of the “corrplot” package [51] to illustrate the relationships between traits associated with the leaf economic spectrum and water-use efficiency (δ¹³C) during the formation of leaf tissues that most contributed to acclimation to light and dry (Thic, Pal, and Subep).

The relationship between variation and covariation was examined by first calculating Hedges’ g value for each functional trait, following the previously described formula, to quantify the magnitude of trait variation between environments. Trait covariation was then estimated by calculating the proportion of significant Pearson correlations (p < 0.05) that each trait had with all other traits in the dataset. This covariation index (Ci) was obtained using the following formula established in [54]:

$$Ci={\displaystyle \frac{N°SignificantCorrelations}{N°TotalCorrelation}}$$

With these two values, one for variation (Hedges’ g) and one for covariation (Ci), assigned to each trait, linear models were constructed using the “stats” package of R to evaluate the relationship between trait variation and covariation within and across functional trait groups (morphological, anatomical, physiological, and nutritional).

3. Results

3.1. Leaf Functional Traits

Variation in light and water availability between two coastal Atlantic Forest phytophysiognomies led to distinct acclimation responses in the leaf functional traits of Schinus terebinthifolia.

Morphological traits revealed that individuals established in the dry–open environment had significantly higher values for Thic, LWC, LMA, and Den than those in the humid–shaded environment (

Table 2. Paired analysis of the functional traits of S. terebinthifolia leaves in dry–open and humid–shaded environments. Means (±standard deviation) for the two environments were compared by t-test, followed by Hedges’ g effect size test, with significant differences (p < 0.05) indicated in bold. The “Trend” column indicates the environment with the higher mean value for each trait (↑ Dry: higher in dry–open; ↑ Humid: higher in humid–shaded). Additionally, the Hedges’ g values include directional signs indicating the magnitude and direction of the difference: a positive sign (+) denotes a higher mean value in the dry–open environment, while a negative sign (−) denotes a higher mean value in the humid–shaded environment. These directional indicators are based on the comparison of mean values, regardless of statistical significance.

Trait Groups	Traits	t-Test			Effect Size
		Dry–Open	Humid–Shaded	p-Value	Interpret	Trend
Morphological	Thic (mm)	0.25 ± 0.004	0.19 ± 0.01	<0.001	large	↑ Dry
	LWC (g·m−2)	472.4 ± 63.4	400.0 ± 32.8	0.043	large	↑ Dry
	LMA (g·m−2)	309.8 ± 27.9	203.8 ± 8.3	<0.001	large	↑ Dry
	Den (g·mm−3)	1248.6 ± 118.0	1111.4 ± 41.2	0.04	medium	↑ Dry
	LA (cm−2)	9.6 ± 2.5	10.9 ± 0.50	0.304	medium	↑ Humid
Anatomical	Pal (μm)	125.9 ± 20.1	97.2 ± 15.7	0.036	medium	↑ Dry
	Spon (μm)	86.3 ± 16.1	81.5 ± 16.9	0.657	small	↑ Dry
	Subep (μm)	33.7 ± 2.8	29.4 ± 2.9	0.046	medium	↑ Dry
	Adep (μm)	14.5 ± 1.5	13.5 ± 1.5	0.353	small	↑ Dry
	Adcut (μm)	4.2 ± 1.3	4.1 ± 1.0	0.862	very small	↑ Dry
	Abep (μm)	14.1 ± 1.7	14.2 ± 1.9	0.983	very small	↑ Humid
	Abcut (μm)	5.9 ± 2.2	5.1 ± 1.4	0.512	very small	↑ Dry
Physiological	Fv/Fm	0.84 ± 0.01	0.87 ± 0.01	<0.001	large	↑ Humid
	qP	0.96 ± 0.02	0.93 ± 0.02	0.094	medium	↑ Dry
	NPQ	0.10 ± 0.02	0.13 ± 0.03	0.135	large	↑ Humid
	Chlo	6.8 ± 0.86	7.8 ± 1.12	0.157	small	↑ Humid
	Car	5.2 ± 2.2	6.2 ± 0.001	0.347	small	↑ Humid
Nutritional	N (%)	1.26 ± 0.21	1.41 ± 0.21	0.282	medium	↑ Humid
	C (%)	37.7 ± 2.9	35.9 ± 1.7	0.277	small	↑ Dry
	δ13C (‰)	−29.7 ± 0.61	−30.7 ± 0.55	0.024	large	↑ Dry
	δ15N (‰)	0.31 ± 0.11	0.74 ± 0.62	0.167	medium	↑ Humid
	P (%)	1.13 ± 0.28	1.05 ± 0.32	0.673	very small	↑ Dry
	C/N	35.7 ± 4.95	30.8 ± 5.30	0.166	medium	↑ Dry

, Figure 1A,B). For anatomical traits, the thicknesses of Pal and Subep were significantly greater for individuals of the dry–open environment (Table 2, Figure 1A,B). Differences in physiological traits were noted only for F_v/F_m, which was significantly higher for the humid–shaded environment (Table 2). Among nutritional traits, only the isotopic signature of δ¹³C was significantly elevated in the dry–open environment (Table 2). Among trait groups, morphological traits exhibited the greatest intraspecific variability, with a large effect size (Hedges’ g = 2.61), followed by nutritional traits (Hedges’ g = 0.63) and physiological traits (Hedges’ g = −0.51), which showed a medium effect size; anatomical traits demonstrated a small effect size (Hedges’ g = 0.37) (Figure 2A–D).

The patterns of epicuticular wax deposition exhibited high functional variation between environments (Figure 1C–J). Individuals of the dry–open environment displayed epicuticular waxes on the adaxial surface arranged in crust-like formations, while those of the humid–shaded environment exhibited waxes as a smooth layer (Figure 1C,E). Epicuticular waxes on the abaxial surface appeared as fissured layers in both environments (Figure 1G,I). The cuticle on the adaxial surface presented a rough surface in the dry–open environment and a smooth surface in the humid–shaded environment (Figure 1D,F). The cuticle on the abaxial surface exhibited localized roughness around the stomata in both environments (Figure 1H,J). Despite these topographical differences, the thicknesses of Adcut and Abcut did not differ between environments (Table 2).

Chloroplasts were distributed along the cell wall of the palisade parenchyma cells (Figure 3A). The chloroplasts displayed large oil drops and starch grains scattered throughout the stroma in both environments (Figure 3B–F); however, plastoglobules were only observed in individuals of the dry–open environment (Figure 3C). The primary disparity in chloroplasts between the two environments was in the organization of thylakoid membranes. Chloroplasts of the dry–open environment displayed uncompressed and unstacked thylakoid membranes, resulting in stromal gaps between them (Figure 3C,E), while chloroplasts of the humid–shaded environment exhibited thylakoid membranes organized in grana structures composed of 10 to 12 stacked thylakoids (Figure 3D,F).

3.2. Functional Trait Relationships

The relationships between the leaf functional traits of S. terebinthifolia were analyzed using Principal Component Analysis (PCA) and the interaction network of functional traits of the dry–open and humid–shaded environments (Figure 4A,B). The first two dimensions of the PCA explained 90.8% of the data variability, with 66.1% accounted for by Dimension 1 (Dim 1) and 24.7% by Dimension 2 (Dim 2) (Figure 4A). The traits Pal, Subep, Adep, Abcut, and LMA were more correlated with Dim 1, while traits like Abep, Adcut, Chlo, F_v/F_m, and Spon were more related to Dim 2 (Figure S6).

The interactions between the functional traits separated individuals of the dry–open environment and those of the humid–shaded environment into two respective clusters. The traits Pal, Subep, and LMA were more influential in forming the dry–open cluster, while F_v/F_m and Chlo were determinants for the humid–shaded cluster. Strong significant correlations among Abcut, Adep, Spon, Adcut, and Abep, evidenced by the clustering and orientation of their respective vectors in the PCA, led to a slight overlap between the clusters, demonstrating the sharing of functional traits between dry–open and humid–shaded environments (Figure 4A).

The general interaction network, constructed between the functional traits of the two studied environments (Figure 4B) using Pearson correlation (coefficient > 0.65; p < 0.05), displayed 23 nodes with 185 identified correlations, including 149 positive and 36 negatives. Anatomical traits and δ¹³C showed strong significant correlations above 0.90. Furthermore, the analysis of functional covariations revealed that traits related to the leaf economic spectrum, such as LMA, Thic, Pal, Subep, and δ¹³C, exhibited significant correlations (Figure 5).

3.3. Relationship Between Variation and Covariation

The variance and covariance of functional traits, as analyzed through the linear relationships established between Hedges’ g values and the covariance index (Figure 6), were significantly negatively related for the morphological trait group (R²_adj = 0.81; slope = −0.08; p = 0.024). In contrast, the anatomical trait group (R²_adj = 0.74; slope = 0.31; p = 0.008) and nutritional trait group (R²_adj = 0.84; slope = 0.42; p = 0.006) exhibited positive correlations between the variance and covariance of functional traits, while the physiological trait group showed no significant correlation (R²_adj < 0.01; slope = 0.05; p = 0.543). The differing directions of the slopes observed across the functional trait groups resulted in a non-significant linear relationship (R²_adj < 0.01; slope = −0.03; p = 0.380) between variance and covariance for the overall set of functional traits.

4. Discussion

This study explored the acclimation of S. terebinthifolia to variation in light and water availability between contrasting environments. The findings demonstrate that S. terebinthifolia adjusts its leaf functional traits in response to environmental heterogeneity involving light and water availability. This acclimation includes investment in traits that enhance light distribution and dissipation and mechanisms that sustain photosynthetic performance. Acclimation to these environmental differences also involves coordinated variation among distinct groups of functional traits, particularly those associated with water-use efficiency. Finally, we emphasize the importance of analyzing patterns of variation and covariation separately for each trait group, as these patterns may provide insights into the mechanisms driving phenotypic variability under environmental stress.

4.1. Functional Variability of Leaf Traits in Response to Light and Water Availability

The amount and quality of light and availability of water in an environment can profoundly impact leaf functional traits, particularly leaf thickness [7,55,56]. Plants exposed to high light availability and limited water availability tend to adapt by investing in the development of more robust mesophyll tissues in leaves, with elongated cells organized into multiple layers [57,58]. The present study showed that S. terebinthifolia had significantly thicker leaves when in the dry–open environment than when in the humid–shaded environment, mainly due to the increased thickness of Pal and Subep. This thickening of Pal is associated with cell elongation, such that cells act like optical fibers, allowing greater light penetration into the mesophyll and more uniform light distribution [59]. Additionally, as light passes through these cells, some energy is dissipated as heat, preventing leaves from overheating internally [7]. Subep performs functions like the velamen found in the roots of species of Orchidaceae, such as water storage, prevention of moisture loss, and reflection of infrared and ultraviolet radiation [60].

The greater thickening of mesophyll tissues in dry and open environments correlates directly with higher leaf water content. According to the resource investment theory, our results suggest that increased water storage capacity is a fundamental acclimation strategy for S. terebinthifolia, aiming to mitigate water stress and high irradiance [61]. This is aligned with the broader consensus that maintaining a high leaf water content is essential for effective thermal regulation through evapotranspiration, without limiting CO₂ uptake [61,62]. Furthermore, higher Den found in the dry–open environment also contributes to increased water content in leaves, as greater cell compaction reduces intercellular spaces, decreasing water loss by transpiration; however, this compaction may also reduce mesophyll conductance of CO₂ [38,63,64]. Also, the increased thickening of leaf mesophyll increases chloroplast surface area exposure, facilitating CO₂ uptake and compensating for the difficulty of CO₂ diffusion within the leaves [6].

“The substantial variations observed in morphological traits indicate that these attributes exhibit a high degree of variation, consistent with Arnold’s paradigm regarding their influence on plant performance and fitness [65]. Although this variation does not directly confirm selection, it may contribute to fitness advantages in heterogeneous environments [66]. The pronounced variation in morphological traits found in the present study, compared to anatomical, physiological, and nutritional traits, suggests that these attributes may be particularly responsive to contrasting environmental conditions. While this high variation indicates a potentially relevant role in helping S. terebinthifolia cope with dry conditions and variable light regimes, we emphasize that such variation does not, by itself, confirm adaptive value. Additional evidence, such as fitness differentials or heritability estimates, would be required to explicitly infer selection or adaptive evolution.”

Although no significant differences were observed for the thicknesses of Adcut and Abcut, the topography of Adcut exhibited marked variation between the dry–open environment and the humid–shaded environment. The enhanced ornamentation of epicuticular waxes under high light conditions may result from the light-induced expression of genes such as CER1, WAX2, and LACS2, which coordinately regulate both cutin formation and wax biosynthesis and deposition [67,68]. Once deposited on the cuticle surface, the major components of epicuticular waxes—primarily very long-chain fatty acids (C₂₀–C₃₄)—can undergo partial melting followed by recrystallization. This temperature-sensitive process reorganizes these molecules into more ordered microstructures, enhancing the morphological complexity of the surface [69].

The primary functions of the cuticle and epicuticular waxes include light reflection and water loss restriction [10]. These protective roles are largely determined by surface topography and chemical composition. For instance, smooth adaxial surfaces reflect about 4.5% of incident light, while ornamented surfaces can reflect 20%–50%, which enhances light scattering [70]. In addition, the cuticle dissipates high levels of UV radiation, primarily due to phenolic compounds covalently bound to cutin [71]. Although not directly quantified in the present study, the leaf tissues of S. terebinthifolia are known to contain high levels of phenolic compounds, suggesting a potential role in UV protection [72,73]. Epicuticular waxes also hinder non-stomatal water loss, especially under drought conditions. Their removal increased transpiration in Brazilian semiarid species [74], and higher wax content has been linked to greater water retention [75]. Together, these findings reinforce the dual role of epicuticular waxes in photoprotection and water conservation, which are critical for acclimation in heterogeneous environments.

Using chlorophyll a fluorescence, we found that individuals in the humid–shaded environment exhibited higher Fv/Fm values, reflecting a greater efficiency in utilizing absorbed light for photochemistry [40]. Our results are consistent with the established paradigm that shade-acclimated leaves prioritize photosynthetic efficiency; however, it is notable that values in both environments (0.83 ± 0.03 in dry–open and 0.87 ± 0.01 in humid–shaded) remained above 0.75, which is the range considered optimal for PSII function even under stress [76]. This performance is further corroborated by our observation of high qP (close to 1) and low NPQ, a physiological pattern that aligns with [76], indicating that S. terebinthifolia effectively directs most absorbed energy toward photochemistry rather than dissipating it as heat, even when exposed to the more challenging dry–open conditions.

These physiological patterns are consistent with the anatomical and ultrastructural adjustments observed under a dry–open environment. The increased thickening of Pal in dry–open conditions likely promotes chloroplast repositioning within cells, reducing excessive light exposure and minimizing photoinhibitory damage [77]. Furthermore, palisade cell elongation allows chloroplast alignment along cell walls, increasing CO₂ assimilation efficiency [9,78].

At the subcellular level, chloroplasts of S. terebinthifolia in the dry–open environment presented an unstacked arrangement of thylakoid membranes, a configuration often associated with photoprotective responses under stress conditions [57,79]. This organization is believed to facilitate the access of repair proteins to photodamaged PSII complexes, particularly for the turnover of protein D1, and to improve electron transport by increasing the connectivity between PSII-LHCII and PSI-LHCI [12,80]. Furthermore, the increased abundance of plastoglobules and starch grains observed in chloroplasts of plants in the dry–open environment suggests a reinforcement of antioxidant defenses and energy storage capacity. Plastoglobules have been implicated in mitigating oxidative damage and protecting thylakoid membranes against lipid peroxidation [81]. Altogether, these structural and biochemical adjustments likely contribute to maintaining PSII functional activity under high light and low water availability, as evidenced by Fv/Fm values sustained above the photoinhibition threshold.

4.2. Variation and Covariation in Functional Traits: A Complex Framework in the Acclimation to Light and Water Availability

Principal Component Analysis (PCA) and interaction network analysis revealed key traits involved in the acclimation of S. terebinthifolia to heterogeneous light and water conditions. The present study found Pal, Subep, and LMA to be the primary traits contributing to the cluster of individuals of the dry–open environment. This highlights the necessity of developing mechanisms that enable efficient photon distribution and thermal dissipation while limiting water loss [7,59,68]. The most influential traits for the cluster formed by individuals of the humid–shaded environment were Fv/Fm ratio and Chlo content, suggesting that these individuals invested in strategies to maximize light capture due to the lower light availability [82]. This acclimation is evident from the relatively higher performance of PSII in a humid–shaded environment compared to dry–open environments. However, it is important to note that under dry–open conditions, PSII becomes more vulnerable to photo-oxidative damage, which may impair its proper functioning [83]. One important finding was that the anatomical traits with the least variation between the two studied environments (e.g., thickness of Spon, Adep, Abep, Adcut, and Abcut) exhibited a broad network of functional interactions, slightly overlapping the two clusters. This result demonstrates that the coordination among traits reduces the need to express new phenotypic traits to tolerate light and water heterogeneity [7,18]. We understand that this functional covariation reflects mechanisms that are more directed toward imposing water restrictions, which are inherent to the dry ecosystem where the species is established.

The trait relationships of S. terebinthifolia indicate that a significant portion of carbon is allocated to the development of Pal and Subep, as evidenced by strong correlations between these traits and LMA. LMA is a widely used metric in leaf economic spectrum studies, representing dry mass investment per unit area for light interception [26]. The higher mean LMA values observed for the dry–open environment suggest substantial carbon allocation to thicker Pal and Subep.

Additionally, the leaf structure of S. terebinthifolia prioritizes water-use efficiency (WUE), as indicated by strong correlations of these traits with δ¹³C. δ¹³C was the only nutritional trait with statistical significance and high effect size in the dry–open environment. This characteristic suggests that individuals exhibit greater WUE under conditions of high light and low water availability than those in shaded and more humid environments. Previous studies indicate that high light levels and low water availability can increase the composition of δ¹³C in leaf tissues [84,85]. This phenomenon occurs because Rubisco (ribulose-1,5-bisphosphate carboxylase-oxygenase) has a higher affinity for ¹²C; however, when stomatal conductance is reduced to minimize water loss, the internal concentration of ¹²C in leaves tends to decrease [86]. In this situation, Rubisco begins incorporating ¹³C into carbon structures. This mechanism was likely adopted by S. terebinthifolia in leaf construction, especially in the dry–open environment.

The relationships between variation and covariation among functional traits have been widely discussed but remain a subject of many questions and controversies [87,88]. Initially, it was proposed that these relationships tend to be negative, aiming to minimize the costs associated with the expression of new phenotypic traits [18]. However, more recent studies suggest that, under stress conditions, these relationships may be positive [7,19]. In most of these studies, these relationships are analyzed by considering only a single group of functional traits, and when more than one group is included, the relationships are generally not investigated separately for each group, which may obscure important associations. Since the present analyses of the acclimation capacity of S. terebinthifolia involved different groups of functional traits, we considered this a valuable opportunity to clarify these relationships.

The results showed that the overall linear relationship between variation (Hedges’ g) and covariation (covariation index) had a slightly negative, but non-significant, slope. Examination of the slopes of the different functional trait groups (morphological, anatomical, physiological, and nutritional) separately revealed slopes that varied from negative to positive to null. We believe that the different directions of these slopes, along with their significance levels, resulted in an overall slope that was nearly null.

Functionally, these divergent slopes may reflect environmental pressures acting on specific trait groups. For example, the positive slope for the anatomical trait group was driven by an environmental pressure that required greater variability in the thicknesses of the palisade parenchyma and subepidermal layer. However, these traits maintained functional covariations with other anatomical traits, resulting in a positive slope. A similar phenomenon occurred with the nutritional trait group, whose relationship was driven by δ¹³C. In contrast, the negative slope of the morphological trait group was influenced by leaf area, which had limited involvement in acclimation to light and water availability, while the other traits actively participated. The null slopes of the physiological trait group indicate that environmental factors did not significantly impact these traits. We interpret this as a compensatory system, where changes or covariations among structural and nutritional traits were prioritized, ensuring that heterogeneities in light and water availability did not compromise photosynthetic processes.

We understand that this functional covariation reflects mechanisms that are more directed toward imposing water restrictions, which are inherent to the dry ecosystem where the species is established. Because these patterns were evaluated during the dry season, they likely represent species responses under conditions of greater water limitation. Seasonal fluctuations in water availability may influence the expression of some functional traits, particularly those associated with resource acquisition, photosynthetic performance, and water-use strategies, potentially modifying the strength of the trait associations observed here. Consequently, evaluating trait coordination under contrasting seasonal conditions may provide additional insights into the temporal dynamics of acclimation and the stability of these functional relationships in seasonally dry tropical forests.

5. Conclusions

The acclimation of S. terebinthifolia to light and water availability primarily involved variation in structural traits, such as LMA, thickness of Pal and Subep, topographic irregularities of Adcut, and thylakoid membrane arrangement in chloroplasts. These traits are associated with efficient mechanisms for distributing and dissipating thermal energy from light and for maintaining water balance, ensuring that the plants’ photosynthetic performance is not compromised under heterogeneous environmental conditions. In contrast, the traits that exhibited the greatest covariations were anatomical traits, such as the thickness of Spon, Adep, Abep, and Adcut—structures directly related to strategies that minimize water loss through the leaf surface and support drought tolerance.

The linear relationship between variation and covariation during acclimation proved quite complex, as it was insignificant when analyzing the total dataset. However, separate examinations for the different groups of functional traits (morphological, anatomical, physiological, and nutritional) revealed slopes with varying vector directions. These results suggest that the acclimation of S. terebinthifolia to heterogeneous light and water availability prioritized variations and covariations primarily between structural and nutritional traits to prevent compromising the photosynthetic processes. This highlights the importance of analyzing trait groups independently, rather than in aggregation, to avoid misleading or incomplete interpretations.

Future studies incorporating trait measurements across contrasting seasonal conditions may further improve our understanding of the temporal stability of trait coordination patterns and acclimation processes in seasonally dry tropical forests.