Leaf architecture and symmetry of understory tree species of an Araucaria forest

Shade plants of ombrophilous forests are subjected to light-limiting conditions and need to invest in architectural structures associated with leaf symmetry to increase light capture. This study investigated the leaf architecture of six Araucaria forest tree species with distinct symmetry: Cupania vernalis, Casearia sylvestris, Schinus terebinthifolius, Piper gaudichaudianum, Roupala brasiliensis and Cedrela fissilis. We hypothesized that symmetry, associated with other traits, minimizes self-shading. Asymmetry index, petiole length, total leaf area, leaf angle, internode length and stem diameter were measured. The asymmetry index did not indicate a clear distinction between asymmetric and symmetric leaves. Leaves classified as asymmetric had higher values for the asymmetry index in the median and basal regions of the leaf, while symmetrical leaves had higher values in the apical region. The results also indicated an adjustment among structural leaf traits that facilitated a three-dimensional organization that produced an advantageous arrangement for light capture, which seems to be a response to selective pressure by the heterogeneous light conditions of the ombrophilous forest understory.


Introduction
Small amounts of diffuse radiation punctuated by sun flecks of short duration (Valladares & Niinemets, 2008) results in less than 2% of active photosynthetic radiation reaching the understory stratum of ombrophilous forests.In such limiting light conditions forest and, especially, understory plants need to maximize light capture by investing in structures with low construction and maintenance costs (Givnish, 1988).
Shaded leaves tend to be thinner and larger than sun leaves.Other traits associated with lamina expansion have been related to the process of light capture such as leaf angle, petiole length and specific leaf mass, as well as leaf shape and phyllotaxis (Valladares & Brites, 2004;Boeger, Garcia, & Soffiatti, 2009).The organization of these traits defines leaf architecture and optimizes carbon gain by individual leaves, thus contributing to the total performance of the plant (Ackerly, 1999).The combination of these traits can influence the degree of self-shading and minimizing leaf overlap, which in turn can cause an increase in photosynthetic rates (Givnish, 1984), especially in ombrophilous forests.
In addition to the aforementioned traits, selfshading can be reduced by leaf asymmetry (Givnish, 1984), especially in leaves with asymmetric bases.Leaves with total or partial asymmetry occur in some plant species of the genera Begonia (Begoniaceae), Columnea (Gesneriaceae) and Tilia (Tiliaceae), and in species with compound leaves in ombrophilous forests (Givnish, 1984;Boeger, Alves, & Negrelle, 2004).According to Chitwood et al. (2012), asymmetry can be due to the asymmetric transport of auxin to different leaves of the same branch, thereby forming leaves with distinct sides.
As a functional attribute, however, asymmetry is poorly studied.Asymmetry can lead to a reduction in leaf area of one side of a leaf and consequently influence photosynthetic rates (Givnish, 1984).On the other hand, if asymmetry occurs at the base of the lamina it minimizes auto-shading, and may be compensated for by the greater area of the other side of the leaf.A similar phenomenon is observed with anisophyllous leaves, with the leaf area of the smaller leaves being compensated for by the leaf area of the larger leaves (Muelbert, Varassin, Boeger, & Goldenberg, 2010).
Leaf area, shape, size, length and width are often used as ecological functional traits for measuring functional diversity in ecosystem dynamics (Pérez-Harguindeguy et al., 2013).Variation in a certain functional trait can influence species fitness, with less variation in the trait indicating less functional diversity in a given community (Rosenfeld, 2002).However, as a functional trait leaf symmetry is still poorly understood, as well as its relationships with other functional traits.
This study investigated the leaf architecture of six understory species with distinct symmetry of an Araucaria forest.We hypothesized that asymmetry, associated with other traits such as leaf angle, leaf shape and petiole and internode length, minimizes self-shading without compromising leaf surface.These traits are important for increasing light capture and maintaining photosynthetic rates, especially in understory plants.

Studied species
Six understory species were selected based on their importance value as determined by a phytosociological survey (Neto et al., 2002) and based on the presence of least ten individuals in the forest (Table 1).The limited number of chosen species is due to the low number of species with asymmetrical leaves in the study area (Silveira, Boeger, Maranho, Melo, & Soffiatti, 2015).The chosen species are as follows: Roupala brasiliensis Klotzsch (Proteaceae) is a native tree species of forests in South Brazil, and is commonly known as carvalho or carvalho-brasileiro.The leaves are compound, odd-pinnate, and alternate with elliptical and asymmetrical leaflets with an acuminate apex (Lorenzi, 1992).
Casearia sylvestris Swartz (Salicaceae), commonly known as guaçatunga, is a perennial species with symmetrical, simple and alternate leaves.The lamina is elliptical and has an acuminate apex, a cuneate base and crenate margins (Lorenzi, 1992).
Cupania vernalis Cambess (Sapindaceae) is a native tree species known as camboatá or camboatãvermelho.It is found from the state of Minas Gerais to the State of Rio Grande do Sul.It occurs in different strata of secondary forests as well as the understory.Its leaves are alternate, compound and even-pinnate.The leaflets are symmetric and oblong, with an obtuse apex, an acute base and serrate margins (Lorenzi, 1992).
Cedrela fissilis Vellozo (Meliaceae) is known as cedro rosa and is widely distributed in all tropical vegetation except the Cerrado.Its leaves are compound with asymmetrical and oblong leaflets, with an obtuse base and an acute apex (Carvalho, 2006).
Schinus terebinthifolius Raddi (Anacardiaceae) is known as aroeira vermelha or pimenta rosa.It is widely distributed throughout South America and occurs from the State of Rio Grande do Norte to the State of Rio Grande do Sul in Brazil.It occurs in many forest types including Araucaria forest, Atlantic forest, deciduous forest and mangroves (Lorenzi, 1992).It possesses compound leaves with asymmetrical and elliptical leaflets with an acute apex and a decurrent base.
Piper gaudichaudianum Kunth (Piperaceae) is a native shrub that is very common in montane Atlantic forest where it occupies forest edges.It is commonly known as jaborandi.Its leaves are simple, alternate and elliptical with an asymmetric base, an acuminate apex and entire margins (Albiero, Souza, Mourão, Almeida, & Lopes, 2005).

Methodological procedures
Ten individuals of each species were selected for study.The median branches were marked and removed with pruning shears and three leaves between the 3 rd and 6 th nodes from the apex were selected.For compound leaves, median leaflets were selected and treated as simple leaves (Bongers & Popma, 1990).According to some authors, leaflets can be functionally similar to single leaves when they are in the same size class because of their origin, patterns of differentiation (Sinha, 1997) and function (Bongers & Popma, 1990).
The following measurements were made: internode length (between 3 rd and 6 th nodes); petiole length; median leaflet angle; and the width of each side of the leaflet from the central midrib to the margin of lamina in three points (Figure 1-I): apical region (one cm from the base; Figure 1-I a and b); median region (Figure 1-I c and d) and basal region (one cm from the apex; Figure 1-I e  and f).Linear measurements were made with a digital caliper and leaf angles with a protractor.
Total leaflet area (cm2) was calculated from a digital image obtained using a flatbed scanner and Sigma Scan software (version 4.0, SPSS Inc., Chicago, IL, USA).Asymmetry of median leaflets was calculated by an asymmetry index (Palmer & Strobeck, 1986)

Statistical analysis
Mean values and respective standard deviations were calculated for all quantitative variables.Leaf sides were compared using the T-test (α = 0.05) and Person correlations among leaf traits were performed using Past version 1.34 (Hammer, Harper, & Ryan, 2001).

Results
The asymmetry index varied among regions of the studied leaves (Table 1).The leaves of Roupala brasiliensis and Piper gaudichaudianum had the highest leaf asymmetry indexes, while Cedrela fissilis and Schinus terebinthifolius had the lowest (Table 1).For asymmetrical leaves, the asymmetry was greater in the median and basal regions of the leaves; for P. gaudichaudianum the asymmetry was greater in the basal region of the leaf (Table 1) and for R. brasiliensis it was greater in the median region of the leaf (Table 1).The asymmetry of P. gaudichaudianum was significant (p < 0.001) and represented a difference of 15% between the areas of the two sides of the leaf, while in R. brasiliensis the difference was greater, around 35% (p < 0.001).The differences between the areas of the two sides for C. fissilis, S. terebinthifolius, C. vernalis and C. sylvestris were 3, 7 and 7% and 12%, respectively (Figure 2), but these values were not significant (p > 0.05).
The leaves of S. terebinthifolius had the smallest total leaf area while P. gaudichaudianum and C. fissilis had the largest; however, leaves with larger areas (P.gaudichaudianum, C. fissilis and C. vernalis) also had greater standard deviations (Table 1).Roupala brasiliensis and P. gaudichaudianum had the greatest difference between side A and B indicated, as indicated by the A/B ratio (Table 1).The A/B ratio was lower and similar among the other species, independently of the leaf symmetry (Table 1).The longest petioles/petiolules occurred in leaves of C. sylvestris and the shortest in R. brasiliensis and C. fissilis (Table 2).The internode/rachis lengths did not vary among species, except in P. gaudichaudianum, which had the highest mean value (Table 2).The leaf angles also exhibited little variation, except for R. brasiliensis, which had the lowest mean (Table 2).

Discussion
As expected, the highest asymmetry indexes were for P. gaudichaudianum and R. brasiliensis, with the highest values being at the leaf base.This high degree of asymmetry is further supported by differences between absolute measurements of the major and minor sides of the leaves of these species.This difference represents a reduction of area (Figure 2) in P. gaudichaudianum (17%) and in R. brasiliensis (35%).The highest leaf area reduction in the median and basal regions of the leaves in species with asymmetric leaves (Figure 2) indicates that asymmetry contributes to selfshading avoidance in the leaf below and subsequent leaves or leaflets, which minimizes the negative effect of leaf area loss.Furthermore, overlapping is a respiratory drag for shaded leaves and a potential source of abrasion for all leaves.Plants that avoid this overlap have an additional advantage for light competition (Givnish, 1984).
Asymmetry is also a function of hormonal action on primordial leaves due to biased auxin distribution during phyllotactic patterns.In spiral systems, the peak of auxin concentration is displaced towards the descending side of the center of mass of leaf primordia and is influenced by architecture of the shoot apical meristem, as observed in tomato (Déb, Marti, Frenz, Kuhlemeier, & Reinhardt, 2015).Studies on Vitis spp and Hedera helix showed that the asymmetric flux of auxin in the shoot apical meristem accentuates differences along the left-right axis and is influenced by phyllotaxis (Martinez, Chitwood, Smith, & Sinha, 2016).
The results presented here also indicated an adjustment between structural characteristics, such as asymmetry, leaf area, petiole length, internode length, foliar angle, phyllotaxis and leaf shape, that allowed a three-dimensional arrangement of the leaves for each species, as suggested by the correlation analysis.
The architecture of branches and leaves is influenced by several characteristics that are important in maximizing light capture in heterogeneous environments and/or artificial environments such as crop fields (Yuan, Yong, & Peng, 2015).The combination of leaf architectural features seems to be the main element for success in capturing light in these environments, because these architectural traits are very plastic, and plants usually adapt leaves to their environment (Valladares, Gialoni & Gómez, 2007).The entire lightharvesting component of the ecosystem seems to be based on relationships between specific leaf traits despite extensive intraspecific variation in these traits (Xu, Guo, Xu, Wei & Wang, 2009).
The set of processes that enable a plant to avoid shading by neighboring plants is known as shade avoidance syndrome (Smith & Whitelam, 1997) and it confers higher fitness in dense stands of plants (Franklin & Whitelam, 2005).Shade avoidance syndrome is one the most important competitive strategies for shaded plants and its effectiveness depends the combination of multiple responses they develop, such as internode and petiole elongation, leaf angle, leaf size and shape, phyllotaxis and asymmetry (Smith & Whitelam, 1997;Valladares & Niinemets, 2008).
Leaf area is a very plastic and highly variable attribute among species (Boeger et al., 2006) and among individuals of the same species (Vieira, Boeger, Cosmo & Coan, 2014).In the study area, species occurring in the understory varied from 9.7 cm 2 (Myrcia rostrata) to 43.7 cm 2 (Luehea divaricata) (Boeger et al., 2006;Vieira et al., 2014).The average values for leaf area of the studied species, both those with asymmetrical and symmetrical leaves, are similar to the values for other species occurring in this area, except for S. terebinthifolius.Additionally, all the studied species were classified as notophylls, which is the most common size class in shaded environments, as well as a response to light conditions, humidity and temperature (Boeger, Alves & Negrelle, 2004).
Leaves with larger surfaces are expected in shaded environments because they favor the capture of light (Givnish, 1984), as observed in Cupania vernalis and P. gaudichaudianum (> 40 cm 2 ).Furthermore, diffuse light in shaded environments represents 10% of the total sun radiation, and can be important for shade leaves, since this light may be harvested from both sides of the leaf (Valladares & Brites, 2004).
The angle formed by two subsequent leaves varied from 106.8º to 135.2º (Table 2), which are close to the Golden Angle (136.7º).The Golden Angle is considered the best angle for allowing several leaves to be arranged around the stem without a complete overlap (Valladares & Brites, 2004), thus improving light capture efficiency (Sarlikioti, de Visser, Buck-Sorlin & Marcelis, 2011) with minimal cost of investment and tissue maintenance (Givnish, 1987).Leaf angles can vary within an individual plant, with those in the upper canopy being more vertical, while those in the lower canopy and low light habitats being more horizontal.This variation in leaf angle can result in a more uniform distribution of light (Niinemets, 2010).Leaf angle is also influenced by phyllotaxis.Spiral/alternate phyllotaxis is considered selectively favorable for light interception in light-limiting environments (Valladares et al., 2002;Valladares & Niinemets, 2008).Spiral/alternate phyllotaxis facilitates a more efficient use of space near the branch (Givnish, 1984), and ensures greater photosynthetic activity and thus maximal carbon gain (King, Beck, & Lüttge, 2004).
The combination of internode length, petiole length and leaf area also favor light capture, since arrangements of these attributes can maintain a greater distance between subsequent leaves.Internode length can compensate for inefficient leaf angles and types of phyllotaxis (Brites & Valladares, 2005) by increasing the distance between leaves.Because leaves with smaller areas require a smaller investment in structural tissues for leaf support, the investment in longer petioles is considered an effective way to avoid shading by neighboring leaves (Brites & Valladares, 2005).Investment in larger internodes is related to larger leaf areas (e.g.P. gaudichaudianum, R. brasiliensis and C. vernalis), while leaves with smaller areas, such as in C. Sylvestris, investment more in the length of the petiole and less in internodes.
Elliptically-shaped leaves are considered the most efficient blade type for avoiding self-shading.An elliptical shape, as with the studied leaves, diminishes the area of overlap between subsequent leaves, mainly in the basal region, and represents an additional advantage in heterogeneous light environments (Givnish, 1984).
Leaf asymmetry, petiole and internode length, leaf shape, leaf angle and leaf area were correlated and form an architectural organization of leaves.This architecture facilitates an advantageous spatial arrangement for light capture and appears to be a response to selective pressure from the heterogeneous conditions of ombrophilous forests.The elements that influenced leaf architecture did not vary between asymmetric and symmetric leaves, when analyzed individually.However, when they occur in different combinations they result in species-specific architecture that increases light capture throughout the growth and development of individual plants.Furthermore, this architectural organization is adjusted by the growth dynamics of individual trees that occupy multiple different forest strata with distinct light conditions.

Conclusion
All the species studied presented some degree of asymmetry, based on the asymmetry index, including the species described morphologically as symmetric.The asymmetry in asymmetric leaves, such as those of P. gaudichaudianum and R. brasiliensis, is striking in the median and basal regions of the blade, which is associated with self-shading avoidance.Besides asymmetry, leaves can adjusted certain leaf traits, such as petiole length, leaf angle and leaf form, in different spatial arrangements.These distinct architectural organizations allow shade species to increase light capture and occupy multiple different forest strata.Thus, this architecture seems to be a response to selective pressure.
. The asymmetry index (AI) was calculated for each leaflet region [apical (a-b), median (c-d) and basal (e-f)], (Figure 1-I), using mean values as: AI = ( [ i − i] )/ , where Ai indicates the larger side and Bi the smaller side for the three regions of the leaflet (Figure 1-II), and N is the number of the total individuals measured per species.

Figure 1 .
Figure 1. Figure 1.Schematic drawing indicating leaf/leaflet measurements.I. Measurements between midrib and leaf/leaflet margin in apical region (a and b); median region (c and d) and basal region (e and f).II.Letter A indicates the larger side of the leaf and B the smaller side, as used for calculating leaf area.

Table 1 .
Asymmetry index and leaf areas for leaves of the studied species.Different letters represent significantly different means according to the T-test (p < 0.05).

Table 2 .
Leaf traits and means and standard deviations for stem diameter, internode/rachis length, leaf angle and petiole/petiolule length for the studied species.