Leaf anatomical traits are correlated with tree dominance in a Neotropical deciduous forest

Abstract

Deciduous forests from the neotropics are one of the most endangered forest types in the world due to the exploitation of their natural resources by mankind. Many aspects of these ecosystems have been studied; however, there is a lack of information about leaf structure and the effects of tree dominance on their structural leaf patterns. In this article, we examine leaf anatomy and specific leaf areas (SLA) in 13 tree species differing in their dominance in a Dry Forest site in Central Brazil, relating leaf anatomical traits with phytosociological aspects. Leaf anatomical traits differed according to tree dominance: greater leaf thickness (achieved through greater thickness of the mesophyll), low values of SLA and bigger stomata were found for the most dominant species, whereas the less dominant species showed thinner leaves with high SLA, as well as numerous and small stomata. These responses suggest that tree dominance is an important indirect effect associated with vertical light availability in the forest. These strategies are probably related to the accomplishment of a high performance in carbon gain and water economy, given the distinction in irradiance that the leaves of different species are subject to in the dry forest.

Keywords:

• deciduous trees
• dry forest
• functional traits
• leaf anatomy
• tree dominance

Introduction

Neotropical dry forest formations once occupied a wide range of Central and South American territories (Miles et al. 2006; Pennington et al. 2009), but now they are one of the most endangered of all types of tropical forests, mainly because of the extraction of wood and limestone, which are important resources for human activities (Quesada & Stoner 2004). The nature of these forests varies with the deciduous behaviour of their species (Sanchez-Azofeifa et al. 2005). This deciduousness is related to significant seasonality in rainfall distribution, resulting in several months of drought (Mooney et al. 1995). Although there is great variation in the soil profile of these forests (Pennington et al. 2006; Felfili et al. 2007), soil is generally shallow, presenting high amounts of calcium and magnesium (mesotrophic soil) (Mooney et al. 1995; Felfili et al. 2007). As a result of the variation in soil profile, semi-deciduous forests growing in well-developed soils with evergreen and deciduous species can be found (Pennington et al. 2009). However, fully deciduous forests can also be found in shallow soils comprising limestone outcrops (Felfili et al. 2007).

Patches of deciduous forest appear in the seasonal vegetation of the savannas in the Central region of Brazil (Pennington et al. 2006; Furley 2007). These forests have deciduous trees that grow on a shallow substrate characterized by limestone outcrops. Because of rainfall seasonality and edaphic aspects, plants in these forests are subjected to longer periods of drought, which last for 5 to 7 months (Felfili et al. 2007). During the dry season all species shed their leaves, whereas during the rainy season, when water is available, the forest canopy cover can reach 80% (Felfili 2001). Deciduous forests from Central Brazil have several unique tree species (such as Cavanillesia arborea, which resemble a ‘Baoba’ tree) reminding us of the ones occurring in the ‘Chaco’ and the ‘Caatinga’, the driest vegetational types of South America (Felfili et al. 2007). These forests also show a very distinct phytosociological structure, where the five most dominant species represent more than 60% of the forest's total basal area (Felfili et al. 2007).

The dominance of a given plant species relates to its relative importance with respect to the degree of influence that it can exert (based on competition for resources) on the other components of the community (Müller-Dombois & Ellenberg 1974). Recent studies provide evidence that the strategies of plant species can affect their performance in dry and wet Neotropical forests (Poorter & Bongers 2006; Poorter et al. 2008). To increase reproductive value or even growth and space occupancy, tree species differing in dominance and size may have traits that allow them to better deal with light variation along the vertical gradient in the forest. If a tree becomes dominant in a certain environment, it occupies greater space in the forest canopy and understorey than less common trees (Kabakoff & Chazdon 1996). These aspects may suggest that trees with low dominance compete with dominant species for light, and that distinct degrees of dominance may affect certain leaf traits linked to carbon gain and water economy. In spite of such knowledge, until now nothing has been known about the relationships between aspects of tree dominance and leaf strategies adopted by tree species that differ in dominance.

Here we aim to examine structural leaf traits in deciduous tree species according to their dominance in a Neotropical deciduous forest of Central Brazil. To do this, we selected 13 deciduous tree species differing in dominance in that the most dominant species have a higher basal area and greater coverage, shading the less dominant species. We expected that these deciduous tree species with distinct dominances would differ in their structural leaf traits. Considering the fact that aspects of tree dominance are normally related to the degree of exposure to light that a single tree experiences in the forest (Kabakoff & Chazdon 1996), we expected that the most dominant species would show thicker leaves with low values of specific leaf area (SLA) and big stomata, a set of traits that appears in sun leaves (Gratani et al. 2006). In contrast, the less dominant species would show thinner leaves, with higher values of SLA and high stomatal density, a set of traits normally appearing in shade leaves. We also expected that such leaf traits would correlate with values of tree dominance of the studied species.

Material and methods

Study site and selected species

This study was carried out in a 50-hectare area of seasonally deciduous forest at the ‘Fazenda Sabonete’ near the municipality of Iaciara in the Paranã valley, northeast Goiás state, Brazil (14°03′53.2″ S, 46°29′15.2″ W). Soils are shallow, dominated by limestone outcrops, which are rich in calcium, phosphoros and magnesium. Trees are tall and can reach 10 m in height. Annual precipitation is around 900 mm (Carvalho 2009) with a long dry season between May and September. Maximum canopy cover appears during the wet months (October–February), while during the dry season all species become leafless (Carvalho 2009).

In this study, we sampled 13 dominant tree species (Table 1). We selected the species based on their absolute dominances (Abs. Dom., which expresses the proportion of size of each species in relation to the total basal area of all species in the studied area), which were previously obtained from a phytosociological study (Felfili et al. 2007). All the 13 selected species were deciduous, shedding their leaves during the dry season, and were divided into three groups according to their dominance: high dominance (Abs. Dom.>8; four species); intermediate dominance (8<Abs. Dom.>3; five species) and low dominance (Abs. Dom.<3; four species).

Table 1 List of the most important tree species ordered by Absolute dominance (Abs. Dom.) in the Dry Forest on limestone outcrops in Central Brazil.

Species	Family	Abs. Dom.
Cavanillesia arborea K. Schum.	Malvaceae	19.9^Footnotea
Pseudobombax tomentosum (Mart. & Zucc.) Robyns	Malvaceae	14.3^Footnotea
Dilodendron bipinnatum Radlk.	Sapindaceae	11.2^Footnotea
Tabebuia impetiginosa (Mart. ex DC.) Standley	Bignoniaceae	8.3^Footnotea
Myracrodruon urundeuva Allemão	Anacardiaceae	7.7^Footnoteb
Sterculia striata A. St.-Hil. & Naudin	Malvaceae	7.3b
Aspidosperma subincanum Mart. ex A. DC.	Apocynaceae	3.7^Footnoteb
Commiphora leptophloeos (Mart.) J.B. Gillett	Burseraceae	3.2^Footnoteb
Acacia tenuifolia (L.) Willd.	Fabaceae	3.1^Footnoteb
Combretum duarteanum Cambess.	Combretaceae	2.4^Footnotec
Guazuma ulmifolia Lam.	Malvaceae	1.1^Footnotec
Bauhinia ungulata L.	Fabaceae	0.1^Footnotec
Cnidoscolus vitifolius (Mill.) Pohl	Euphorbiaceae	0.1^Footnotec

Data compiled from Felfili et al. (2007).

^a High dominance.

^b Intermediate dominance.

^c Low dominance.

Anatomical study

Leaves were collected in February 2009 from the crown edge in all sampled individuals. For anatomical procedures, we collected a middle fragment (between the main vein and the leaf margin) from healthy and fully expanded leaves from four chosen individuals per species (one leaf per plant). Values of the stem diameter at breast height ranged between 10 and 200 cm among the studied species, but were similar among plants of the same species. Each individual was randomly sampled in a different plot of the 25 plots used in a previous phytosociological study (Felfili et al. 2007). Samples were fixed in a 9 : 0.5 : 0.5 40% formalin : glacial acetic acid : 70% alcohol (FAA 70), dehydrated in a graded ethanol series, infiltrated and embedded in paraffin and cut into 8-µm sections (Johansen 1940). The cross-sections were stained with astra blue-basic fuchsine and were permanently mounted in Entellan® (Kraus et al. 1998). For paradermic observations, a middle leaf fragment from each individual (collected in the same leaves chosen for transversal cuts) was dissociated in a 1 : 1 glacial acetic acid : hydrogen peroxide solution, and then stained with safranin.

Trait analysis

We evaluated 10 anatomical leaf traits that are relatively easy to measure, are common in the literature of ecological anatomy and represent functional characteristics that are normally related to environmental adaptation (see Rossatto & Kolb 2009; Bedetti et al. 2011). Measurements of leaf traits (thickness of epidermis on abaxial and adaxial surfaces, palisade and spongy parenchymas and total leaf thickness) were taken 40 times using a 10×objective lens for each species, with 10 measurements taken per tissue per individual. The stomata frequencies, as well as the length of guard cells and width of stomatal complexes, were obtained using epidermal prints of dissociated material. Most of the leaves are hypostomatic (with the exception of Cnidoscolus vitifolius), and stomata counts were made only on the abaxial surface. The counts and measurements were made in three fields per sample using a 10×objective lens. The stomata size was measured in 30 stomata per sample. The area occupied by the stomata (SA) was calculated assuming the stoma was an ellipse following the equation:

where: SW=stomatal width and SL=stomatal length.

For the measurements of anatomical traits we used the Image Pro 4.0 software. The SLA was measured using five fully expanded leaves, or leaflets, without the petiole, collected from the same individuals used for the anatomical study. Each leaf was scanned on a flatbed scanner and its area was determined using the free software Area (Caldas et al. 1992) and then dried at 70°C for 3 days and weighed.

Data analysis

A principal component analysis (PCA) was performed to visualize the differences in leaf structure between trees showing distinct dominance. All traits were standardized and log₁₀ transformed before the analysis. For the PCA, we used the variance–covariance matrix method (Gotelli & Ellison 2004). Only the two most significant axes were presented in this analysis. The PCA was done using free software Past 2.17b (Hammer et al. 2001). To verify the relationship between leaf traits and tree dominance, average values of leaf anatomical traits were related with the absolute dominance of species in the studied site (retrieved from Felfili et al. 2007) using simple linear regression in the Sigma Plot 11.0 software. In all analyses, we used α=0.05.

Results

Leaf anatomical traits and tree dominance

There was great variation in the thickness of leaf tissues in the studied group of species (Fig. 1 and Table 2). Both adaxial and abaxial surfaces of the epidermis consisted of one layer of cells (Fig. 1A–C); however, dominant species had a thicker epidermis than species with intermediate and low dominance (Table 2). All species presented dorsiventral leaves, with distinct palisade and spongy parenchyma (Fig. 1A–C). Ergastic substances occurred: mucilage was mainly stored in adaxial surface epidermis (Fig. 1A), while calcium crystals (Fig. 1B) and phenolic compounds (Fig. 1C) appeared in various regions of the leaf blade. Average values of mesophyll tissue thickness and of total leaf thickness varied among groups of species with distinct dominances, where dominant species presented higher values in relation to species with low dominance (Table 2). The SLA also differed, with low dominance species showing higher values of SLA (around 250 cm² g⁻¹) than intermediate and high dominance species (around 180 cm² g⁻¹) (see Table 2). Stomatal density ranged from low densities in most dominant species, such as 200 stomata mm⁻² in Cavanillesia arborea, to higher values in less dominant species, such as 1500 stomata mm⁻² in Bauhinia ungulata. There was also a variation in the guard-cell length and stomatal complex width, with dominant species showing guard-cell length values between 13 and 39 µm and stomatal width between 9 and 23 µm. Less dominant species showed stomata with a guard-cell length varying between 14 and 23 µm and stomatal width between 7 and 15 µm. The area occupied by the stomatal complex ranged from 88 to 603 µm² and differed in the groups of studied species. Stomatal traits differed such that the group of species with a high dominance showed bigger stomata and lower stomata density than intermediate and low dominance species (Table 2).

Figure 1 Photomicrographs from cross-sections of leaf blades from tree species of Dry Forest, from less (A) to more dominant (C) species. A, Bauhinia ungulata; B, Aspidosperma subincanum; and C, Cavanillesia arborea. Abbreviations: m, mucilaginous substance; cr, calcium crystal; ad, epidermis of adaxial leaf surface; pp, palisade parenchyma; sp, spongy parenchyma; ab, epidermis of abaxial leaf surface; phe, phenolic compounds. Scale bar 100 µm.

Table 2 Measurements of leaf anatomical traits (mean±standard deviation; n=4 or n=5) for the group of trees according to their dominance in the studied Dry Forest (Goiás, Brazil).

	Dominance
Leaf trait	Low	Intermediate	High
Adaxial epidermis (µm)	16.98±1.41	19.36±2.30	27.80±3.26
Palisade parenchyma (µm)	37.58±4.35	46.48±4.60	52.11±3.97
Spongy parenchyma (µm)	44.45±5.79	51.58±4.47	74.10±12.99
Abaxial epidermis (µm)	9.02±0.47	9.64±0.65	14.86±1.50
Leaf thickness (µm)	99.84±7.25	124.62±9.05	168.12±18.70
Specific leaf area (cm^2 g^−1)	248.11±18.52	184.77±15.67	184.39±8.68
Stomatal density (no. mm^−2)	846.50±152.26	605.46±71.20	482.24±79.18
Guard-cell length (µm)	19.34±1.48	17.00±0.81	24.63±2.44
Stomatal width (µm)	11.87±0.88	11.08±0.49	16.33±1.45
Stomatal area (µm^2)	127. 49±28.46	149.68±16.94	354.87±93.12

The first two axes of the PCA explained 61.11% and 24.06% of the variation in the 10 analysed traits (Fig. 2). The first axis was defined primarily by stomatal density, thickness of spongy parenchyma and leaf thickness. In the first axis, the most dominant species (Cavanillesia arborea, Pseudobombax tomentosum, Dilodendron bipinnatum and Tabebuia impetiginosa) appeared on the right side of the axis, whereas the less dominant species (Combretum duarteanum, Guazuma ulmifolia, Bauhinia ungulata and Cnidoscolus vitifolius) appeared on the left side of the axis (Fig. 2). Species showing intermediate values of dominance appeared on both sides of the first axis (Fig. 2). Along the second axis, there was also a clear shift in the position of most dominant species relative to less dominant species. This shift along axis 2 is attributed to the thickness of epidermis from adaxial and abaxial leaf surfaces and the thickness of palisade parenchyma (Fig. 2).

Figure 2 Position of deciduous forest species on the first two axes of the principal components analysis for all the anatomical leaf traits. Straight lines indicate the influence of anatomical leaf traits on the separation of tree species. AB, thickness of abaxial epidermis; AD, thickness of adaxial epidermis; GCL, guard cell length; LT, leaf thickness; PP, thickness of palisade parenchyma; SA, stomatal area; SD, stomatal density; SP, thickness of spongy parenchyma; SW, stomatal width; SLA, specific leaf area. Filled symbols: most dominant species; semi-filled symbols: intermediate dominant species and open symbols: less dominant species.

Relationship between dominance and leaf anatomical traits

The absolute dominance of tree species was significantly related to important leaf anatomical traits; however, these relationships varied according to the type of leaf trait (Fig. 3). This relationship was significant and positive when considering the total leaf thickness (r ²=0.42, P=0.018) (Fig. 3A) and the mesophyll thickness (r ²=0.32, P=0.043) (Fig. 3B); however, it was significant and negative (r ²=0.33, P=0.038) when SLA was used (Fig. 3C). Despite a negative trend in the relationship (Fig. 3D), absolute dominance was not related to stomatal density (r ²=0.125, P=0.23); however, the area occupied by individual stoma was positive and significantly (r ²=0.48, P=0.009) related to the tree's absolute dominance (Fig. 3E).

Figure 3 Relationship between absolute dominance (Abs. Dominance) of tree species and their leaf anatomical traits. A, Absolute dominance and leaf thickness; B, absolute dominance and mesophyll thickness; C, absolute dominance and specific leaf area; D, absolute dominance and stomatal density; and E, absolute dominance and stomata area.

Discussion

Consistent with our hypothesis, we found that deciduous tree species did not converge in their leaf structure, producing leaves with a distinct structure according to their dominance in the forest. We also found relationships between tree dominance and leaf traits in the studied species. Depending on their dominance, species presented a syndrome of leaf traits, where the most dominant trees showed thicker leaves, with greater thickness in epidermis and parenchymas and a fewer big stomata; whereas the less dominant trees had thinner leaves, with thinner epidermis and parenchymas and a more small stomata.

Differences in the thickness of leaf tissues and in stomatal traits may be associated with the fact that trees with distinct dominances normally differ in their height, crown area and basal area (King 1990; Küppers 1994), and are subjected to distinct irradiance levels (Murphy & Lugo 1986; Niinemets & Kull 1994). Therefore, there is great variation in light availability that occurs along the vertical continuum through the canopy (where most dominant species expose their leaves) and understorey (where the less dominant species occur) (Poorter 2002). Dominant species would produce typical sun leaves, which are thicker, presenting a well-developed epidermal tissue and a well-developed mesophyll (Gratani et al. 2006; Rossatto & Kolb 2010), while the less dominant species (which are normally smaller in size) would produce shade leaves to deal with the low and diffuse light availability present in the understorey (Valladares & Niinemets 2008). In consideration of the above, our results suggest that tree dominance is an important indirect effect, related to light availability, and that tree dominance correlates with leaf anatomy.

A common assumption is that dry environments should promote the selection of leaf anatomical traits linked to the reduction of water loss, primarily through the decrease of leaf transpiration (Seddon 1974). Such traits include small protected stomata, thick leaves presenting tissues with compact cell arrangement and elevated thickness of protection tissues, such as the epidermis with its cuticle (Fahn & Cutler 1992). As all the species studied here are deciduous, this behaviour may be the most important aspect in their adaptive response to deal with water seasonality and seasonal drought (Reich & Borchert 1984); however, their canopy strategy alone cannot explain the variation found for the quantitative values of the studied leaf traits, especially stomatal traits and SLA.

Great efficiency in gas exchange can be achieved with big stomata and elevated stomata density in thick leaves (Pearce et al. 2006; Galmes et al. 2007), as found here for the most dominant species. This suggests that the most dominant tree species may substantially increase their photosynthetic rates in the conditions of high light irradiance to which they are subjected (Pearce et al. 2006). In contrast, the strategy of presenting a great number of small stomata may be useful to the less dominant species, as small stomata are very responsive to environmental variations (Aasamaa et al. 2001; Hetherington & Woodward 2003), and can increase conductance and carbon assimilation in shaded conditions.

Deciduous trees have leaves that lack development of non-photosynthetic tissues, producing thin leaves with high values of SLA (Zhang et al. 2007). We confirmed this, as leaves of the species studied here presented high SLA values (90–270 cm² g⁻¹), which are higher than what is normally reported for other savanna (40–90 cm² g⁻¹) and forest (70–130 cm² g⁻¹) systems (Hoffmann et al. 2005). Differences found here in SLA between dominant and less dominant species are foreseeable given the differences found for thickness of parenchymal tissues in their leaves (Fig. 2). For the most dominant species, both parenchymas showed greater thickness in comparison with less dominant species (Fig. 2, Table 2). Thicker palisade parenchyma in dominant species may help to optimize the capture of direct light that reaches the sun leaves (Vogelmann et al. 1996), whereas the thicker spongy parenchyma might increase the internal leaf conductance to CO₂ (see Terashima et al. 2011). These strategies, coupled with the presence of bigger stomata in dominant species, could significantly improve photosynthetic rates in such sun leaves (Sefton et al. 2002; Terashima et al. 2011) and suggest that these dominant trees can achieve greater efficiency in water use, especially during the wet season when water is available (Greenwood et al. 2008). In the less dominant species, a thick spongy parenchyma may optimize CO₂ assimilation by increasing the capture of the diffuse light that predominates in the forest understorey (Vogelmann et al. 1996).

Based on our results, we can affirm that, despite the common behaviour of deciduousness, the studied species drastically differ in the quantitative aspects of their leaves. Such differences are probably linked to differences in the dominance and occupancy that they present in the deciduous forest environment. Hence, thickness of leaf tissues, total leaf thickness, specific leaf area and the stomata size increase with the tree dominance in association with the higher light availability. Further studies are needed to better clarify whether the pattern observed here was the product of species-specific dominance or if it is linked only to relative tree height regardless of species affiliation, or even if both factors can work together.

Acknowledgements

This work was funded by a FAPESP grant (proc. 2011/23112-3) and from PROPE/UNESP (14/2012/Renove). The authors would like to thank Dr Jeanine Felfili (in memoriam) and Dr Fabricio Alvim Carvalho (UFJF) for the opportunity to travel to the region of the dry forest; Sr Silvio Lacerda, the owner of ‘Fazenda Sabonete’ in Iaciara for permission to work in his property and Dr Stephen Hartley and two anonymous reviewers for their helpful comments.