Trees

, Volume 23, Issue 1, pp 37–49

Morphogenetic trends in the morphological, optical and biochemical features of phyllodes in Acacia mangium Willd (Mimosaceae)

Authors

  • Céline Leroy
    • INRA, UMR AMAP
    • Université Paul Sabatier, UMR EDB
  • Michael Guéroult
    • INRA, UMR AMAP
  • Novi Sari Wahyuni
    • Universitas Brawijaya, Fakultas Pertanian
  • Jacques Escoute
    • CIRAD, UMR DAP
  • Régis Céréghino
    • Université Paul Sabatier, UMR Ecolab
    • CIRAD, UMR AMAP
    • CIRAD, UMR AMAP (botAnique et bioinforMatique de l’Architecture des Plantes)
  • Daniel Auclair
    • INRA, UMR AMAP
Original Paper

DOI: 10.1007/s00468-008-0252-5

Cite this article as:
Leroy, C., Guéroult, M., Wahyuni, N.S. et al. Trees (2009) 23: 37. doi:10.1007/s00468-008-0252-5
  • 140 Views

Abstract

Endogenous variations in the annual growth of trees suggest that similar trends would occur in phyllodes. In comparison to leaves, the characteristics of phyllodes are less well known, hence this study examines the effects of architectural position and age of tree on the phyllodes of Acacia mangium. Phyllodes were investigated on 1-, 2-, and 3-year-old trees from three axis positions within the crown. We focused on the morphological, optical and biochemical traits of the phyllodes. The increase in phyllode area and lamina thickness is more pronounced in the older trees. Leaf mass area (LMA), stomatal density, nitrogen and chlorophyll content increase with tree age. The values of these characteristics decrease from the main stem to the lower branches for the older trees. Phyllode light absorptance increased with tree age whereas reflectance was higher for the upper position compared to the lower position within the crown. Carotenoid content and chlorophyll a/b ratio were higher for the younger phyllodes of younger trees. Increasing tree size induced modifications in the phyllode characteristics which are influenced by both morphogenetic and light gradients within the crown. This study demonstrated pronounced changes in terms of morphological and functional indicators of photosynthetic capacity in relation to phyllode position within the crown and to tree age. These morphogenetic effects on the phyllode characteristics should be taken into account in studies on phenotypic plasticity.

Keywords

ChlorophyllPhyllode morphologyPhyllode anatomyNitrogenOptical propertiesPlant architecture

Introduction

Plants show a remarkable phenotypic variability in response to changes in environmental and ontogenetic factors (Sultan 2000; Wright and McConnaughay 2002). The intraspecific variability in leaf traits in response to light and water availability is well known (Niinemets et al. 2001), and several studies have examined leaf physiology and anatomy changes according to ontogeny or plant size (Thomas and Winner 2002; Ishida et al. 2005). Wright and McConnaughay (2002) suggest that phenotypic plasticity of plants may vary as a function of growth and development. Numerous studies on different plant species based on precise stem morphological observations have revealed that the features of a given elementary botanical entity (axis, growth unit, metamer) are predictable and strongly depend on both (1) their topological location in the comprehensive architecture of a plant and (2) the ontogenetic stage of the organism (Barthélémy and Caraglio 2007). The morphological and anatomical properties of elementary entities change qualitatively and quantitatively during tree development and are explained by differences in “physiological age” of a building meristem (Roggy et al. 2005; Heuret et al. 2006; Barthélémy and Caraglio 2007). Other studies examined the variability of functional leaf traits in relation to local light level and quality within an individual tree crown (Frak et al. 2002). To date, only few studies have collected data from different architectural positions within a tree at different tree ages.

Leaf structure can be a reliable marker of the different tree growth stages (Jones 1999; Poethig 2003). Leaf structure has been recognised to change not only during tree ontogenesis (Niinemets 1997; Ashton et al. 1998; Richardson et al. 2000; Roggy et al. 2005) or leaf senescence (Kitajima et al. 1997; Ishida et al. 1999; Kitajima et al. 2002; Yamashita et al. 2002), but also according to its position within the crown, for several coniferous and broadleaved species at different tree ages (Ashton et al. 1998; Ishida et al. 1999; Richardson et al. 2000; Leal and Thomas 2003; Yanez-Espinosa et al. 2003). Leaf structure can be considered as an indicator of functional leaf traits, and its contribution to leaf function is well documented (Garnier et al. 1999; Niinemets 1999; Roggy et al. 2005). Leaf morphological parameters such as leaf mass per unit area (LMA), leaf thickness and stomatal density have been reported to be positively related to leaf photosynthetic capacity per unit leaf area (Niinemets 1999; Lee et al. 2000). High proportions of chlorophyll and chlorophyll a/b ratio are considered to be an adaptation to increased red light absorption (Cao 1999). Leaf optical properties are driven both by leaf anatomical structure and by chlorophyll concentration (Lee et al. 1990; Baltzer and Thomas 2005). Carotenoid content is positively related to the photoprotection phenomenon (Foyer 1993). A positive relationship between Nitrogen and Amax (maximum foliar net assimilation capacity) has been found when expressed per unit area (Evans and Seemann 1989; Ellsworth and Reich 1993; Eamus and Cole 1997; Roggy et al. 2005). Thus, LMA, leaf thickness, leaf area and stomatal density are good morphological indicators of carbon assimilation. Based on the relationship between foliage chemical composition and photosynthetic capacity (see Niinemets 1999), we have chosen here to use nitrogen and chlorophyll content as functional indicators of carbon assimilation.

The genus Acacia forms a large group, widely distributed in Australia, characterised by phyllodinous leaves in the majority of species (Gardner et al. 2005). Phyllode anatomical structure is known to be an adaptation to arid habitat (typical of xerophyte plants) with stomata on both the adaxial and abaxial surfaces of the lamina and a symmetrical internal phyllode structure with palisade parenchyma tissues on both the adaxial and abaxial sides of the phyllode (Boughton 1990; Yates 1992).

The phyllodinous Acacia mangium Willd. (Mimosaceae) is a tropical lowland species tree native to northern Queensland (Australia), western Papua New Guinea, Irian Jaya and Maluku in Indonesia (Pinyopusarerk et al. 1993). This pioneer fast-growing evergreen tropical species has been widely introduced outside its natural range for its economic (timber, fuelwood, tanning) and ecological importance (soil improvement, reforestation). Acacia mangium is planted either in monospecific plantations or associated with tropical crops in agroforestry systems.

In the present study, we investigated morphology, anatomy, optical properties and chemical composition of phyllodes of A. mangium growing in agroforestry systems. To understand how phyllode properties change in relation to architectural position and plant size, we examined phyllodes produced (1) in three positions along the leafy axis, (2) on three axis categories within the crown and (3) at three tree ages.

Materials and methods

Study sites

The study was conducted in Pakuan Ratu (at 04°32′56″S, 104°5643″E), in north Lampung in the southern part of Sumatra Island in Indonesia. Elevation is around 60 m above sea level. Mean annual temperature is 28°C, varying between 22 and 33°C. Mean annual precipitation in this area is 2,200–2,500 mm with a pronounced dry season from May to October. The soils are very deep, well drained and very acid, with low soil fertility status. Iron concretions are often found within the soil profiles. The study site consisted of three agroforestry stands (2 × 4 m spacing between trees) of A. mangium close to one another. All the trees, native from Papua New Guinea, were planted from seed.

Sampling

Three 1-, 2- and 3-year-old stands were selected on the basis of their ecological similarities. Six trees of each age were selected as the most homogeneous (based on visual observations) and representative trees of the stands, based on total height, basal diameter, and crown shape and without any axis traumatism. The 1-year-old trees, characterised by 170 ± 4 cm total height and 1.8 ± 0.05 cm basal diameter, had two axis categories (A1: main stem and A2: branch). The 2-year-old tress, characterised by 543 ± 32 cm height and 8.6 ± 0.6 cm basal diameter, had four axis categories (A1, A2, A3 and A4), and the 3-year-old trees, characterised by 953 ± 44 cm height and 13.0 ± 2.2 cm basal diameter, had five axis categories and had developed the first inflorescences.

For 1-, 2- and 3-year-old trees (1, 2 and 3), samples of phyllodes located on the first order axis (A1 or main stem) and the second order axes (branches) borne by the median (A2M) and the lower (A2L) part of the first order axis were collected (Fig. 1a). For each of these three axis categories, samples of phyllodes were collected from the upper part (u), the median part (m) and the lower part (l). The upper, median and lower parts were located at approximately the 10th node, 25th node and 40th node, respectively, counted from the top of the axis. All collected phyllodes were fully extended. For each tree age, axis category within the crown and position along the axis, one phyllode was sampled on six trees concerning the main stem (A1u, A1m, and A1l) and five trees concerning the other categories present (less on 1-year-old, see Fig. 1a). A total of 124 phyllodes were collected for measurements.
https://static-content.springer.com/image/art%3A10.1007%2Fs00468-008-0252-5/MediaObjects/468_2008_252_Fig1_HTML.gif
Fig. 1

a Diagrammatic representation of the 1-, 2- and 3-year-old Acacia mangium selected for the present study. The circles indicate where the samples of phyllodes were collected. For each sample, a code was assigned: 1, 2, and 3 = age of the tree; A1 First order axis or main stem; A2 second-order axis (branch); M, and L middle and lower position of branch within the crown; u, m and l upper, middle and lower phyllode position on the axis. The brackets indicate the total number of phyllodes sampled for each axis category (5 or 6 trees sampled). b Morphological characteristics of the phyllode with the extra-floral nectary, which enables to differentiate two lamina sides

Light measurements within the tree crown

Following a reconstruction of the three-dimensional tree architecture using the AMAP methodology (de Reffye et al. 1995; Leroy 2005; Barczi et al. 2008), the Archimed software (Dauzat et al. 1984) was used to determine the interception of diffuse photosynthetic photon flux density (PPFD) of the crown. The percentage of light intercepted by phyllodes in different positions within the crown (main stem, median and lower branches) was estimated (see Table 1).
Table 1

Estimation of the percentage of light received by the phyllodes on different axis categories studied within the crown (main stem, median and lower branches) using Archimed (Dauzat et al. 1984)

Tree age

Main stem (A1)

Median branches (A2 M)

Lower branches (A2L)

1-year-old

75.84 ± 3.22

56.84 ± 3.16

2-year-old

68.89 ± 3.84

25.10 ± 0.64

12.46 ± 1.92

3-year-old

64.81 ± 5.82

19.63 ± 1.16

4.83 ± 0.42

Phyllode optical properties

Because of the bi-facial structure of the phyllode (Yates 1992), the adaxial and abaxial sides are difficult to differentiate. However, based on the presence of an extra-floral nectary near the adaxial part of the base of the lamina, a side “A” was distinguished from a side “B” (see Fig. 1b). Optical properties were measured on phyllode side “A” in a spectral range between 400 and 700 nm, with a spectroradiometer (Unispec, PP Systems, USA). Reflectance (R) and transmittance (T) were measured with a model IC09 integrating sphere via an optical fibre cable (ANCAL Incorporated, USA), with a spectral resolution of 3.4 nm, and were expressed as percentage of incident light. Absorptance (A) was calculated as follows: A = 1 − (R + T).

Phyllode morphology and anatomy

For each sampled phyllode, a digital photograph was taken with a Nikon Coolpix 4500 digital camera. The area of each individual phyllode was then determined by image analysis by full pixel region (Optimas V 6.5, Media Cybernetics, Sylver Spring, MD, USA). Three 1-cm diameter lamina discs were taken from the middle of the phyllode, with care being taken to avoid the main veins, and dried at 50–55°C during 48 h in order to calculate the phyllode dry mass per unit area (LMA, g m-²). Because of a symmetric stomatal density of the two lamina sides (Boughton 1990), side “A” was chosen for the stomatal density study. The number of stomata was recorded under an optical microscope by using epidermal prints from transparent nail polish. Five replicates per sample were studied in order to estimate the stomatal density (number per mm²). The thickness of the central part of the phyllode was measured on 15 μm lamina sections made with a vibratome. Ten replicates per sample were used to estimate lamina thickness (μm).

Phyllode biochemistry

The phyllode pigment content was determined from 2 g fresh mass and extracted with 10 ml of buffered 80% aqueous acetone using a mortar. Pigments were extracted in the dark at 4°C for 24 h. Absorbance of the clear extract at 663, 646, 645, and 470 nm was recorded and concentrations of total chlorophyll, carotenoid and chlorophyll a/b ratio were computed. Pigment concentration of the extract and LMA were combined to compute chlorophyll and carotenoid concentration per unit phyllode area (mg m²) (Wellburn 1994). Phyllode nitrogen content (Na, g m²) was subsequently analysed after a Kjeldahl digestion.

Analysis procedure and presentation of results

Our data consisted of a small number of samples per position and of several variables, which vary and covary in a non-linear fashion. The commonly used Principal Component Analysis (PCA) method, which is based on linear principles for quantitative data, is not suitable in our case (Giraudel and Lek 2001). Therefore non-linear analysis methods such as artificial neural networks should be preferred for dealing with such data (Blayo and Demartines 1991). The dataset was submitted to an adaptive learning algorithm, the Self-Organising Map algorithm (SOM, Kohonen 1982) which simultaneously explores qualitative and quantitative variables. The SOM Toolbox (version 2) for Matlab® developed by the Laboratory of Information and Computer Science at the Helsinki University of Technology was used (see Vesanto et al. 1999 for practical instructions). The SOM algorithm is an unsupervised learning procedure (see Kohonen 2001 for theoretical considerations), which transforms the multi-dimensional input data into a two-dimensional map subject to a topological (neighbourhood preserving) constraint. This network consists of two layers, an input and an output layer. In the present study, the input layer was constituted by 124 nodes (one per sample) connected to the 14 measured variables (11 quantitative and 3 qualitative variables). The output layer was constituted by 54 neurons (visualised as hexagonal cells) organised on an array with 9 rows and 6 columns (Fig. 2). SOM plots the similarities of the data by grouping similar data items together, in a way that can be simply described as follows:
https://static-content.springer.com/image/art%3A10.1007%2Fs00468-008-0252-5/MediaObjects/468_2008_252_Fig2_HTML.gif
Fig. 2

a Distribution of sampled phyllodes on the self-organizing map (SOM) according to tree age, axis category and phyllode position on the leafy axis, and clustering of the trained SOM. Samples which are neighbours within clusters are expected to have similar leaf characteristics. Samples separated by a large distance from each other are distant in the output space. Codes (e.g. 79) correspond to one sampled phyllode. Clusters I–V were derived from a Ward linkage method and a Euclidean distance measure. b Proximity diagram of the clusters: Cluster I: 1A1l, 1A1m and 2A1l; Cluster II: 1A1u, 2A1u and 2A2Mu; Cluster III: 1A2Lm, 1A2Lu, 2A2Lu, 2A2Lm, 3A2Mu and 3A2Lu; Cluster IV 3A2 Mm, 3A2Ml, 3A2Lm and 3A2Ll; and Cluster V: 2A1m, 2A2Mm, 3A1Uu, 3A1m and 3A1l

  1. the virtual samples are initialized with random samples drawn from the input dataset;

     
  2. the virtual samples are updated in an iterative way: (1) a sample unit is randomly chosen as an input unit, (2) the Euclidean distance between this sample unit and every virtual sample is computed, (3) the virtual sample closest to the input is selected and called “best matching unit” (BMU), (4) the BMU and its neighbours are moved slightly towards the input unit.

     
The training was broken down into two phases:
  1. ordering phase (the 3,000 first steps): when this first phase takes place, the samples are highly modified in a wide neighbourhood of the BMU;

     
  2. tuning phase (7,000 steps): during this phase, only the virtual samples adjacent to the BMU are slightly modified.

     

At the end of training, the BMU is determined for each sample, and each sample is set in the corresponding hexagon of the SOM map. Samples which are neighbours on the grid are expected to represent neighbouring clusters of samples; consequently, samples having a large distance between each other (according to the measured variables) are expected to be distant in the feature space. The map size (i.e., the number of output neurons) is important to detect deviations of the data: the number of neurons should not be larger than the number of samples and not too small in order to optimise the differences. Consequently, different map sizes were tested and the optimal size was determined based on the minimum values for quantisation (map resolution) and topographic errors (data topology preservation), following the procedure described by Céréghino et al. (2005). To define clusters between the virtual units of the SOM map, a hierarchical cluster analysis with Ward’s linkage method and Euclidean distance was performed on the weight of each of the 14 measured variables in each of the 54 virtual units (Fig. 2). To analyse the contribution of each qualitative and quantitative variable to cluster structures of the trained SOM, the relative value of each input variable calculated during the training process was visualised in each neuron on the map on a grey scale.

In addition, for a better understanding, the raw data (mean and standard error) were graphically illustrated (1) for 2-year-old trees, according to axis category and phyllode position along the axis, and (2) for mature (“middle”) phyllodes, according to tree age and axis category. The differences between means for all the sampled cases were analysed using a Mann–Whitney–Wilcoxon test which enables comparison of small samples. Moreover, a two-way ANOVA test was performed to compare the effects of each factor (phyllode position on the bearing axis, crown position and tree age) and their interaction on the phyllode structure. The open source statistical software “R” was used.

Results

Phyllode optical properties

Values of reflectance ranged from 0.03 to 0.07 (Fig. 4). High reflectance values characterised Cluster I which corresponds to phyllodes from young trees, from the main stem (A1) and phyllodes located at the upper and middle position on the bearing axes (Figs. 2, 3). Lamina transmittance was higher for phyllodes located at the upper and middle position on the lower branches of the crown (Figs. 2, 3). With the increasing age of trees, the transmittance of phyllodes decreased. Transmittance ranged from 0.01 to 0.06 and increased significantly with phyllode position on the axis and axis category within the crown (Fig. 4a) and decreased significantly with tree age (Fig. 4b; Table 2). The absorptance map shows higher values on the left (0.93) for Clusters IV and V, and lower values on the right (0.88) for Clusters I, II and III (Figs. 2, 3). Phyllodes located at the middle and lower position on the bearing axis absorbed significantly more light than the upper ones (Fig. 4a; Table 2). Phyllode absorptance increased significantly with tree age (Fig. 4b; Table 2). The interaction between factors (phyllode position, axis category and tree age) was not significant for the three optical variables (Table 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs00468-008-0252-5/MediaObjects/468_2008_252_Fig3_HTML.gif
Fig. 3

Gradient analysis of the phyllode characteristics for each qualitative (a) and quantitative (b) variable on the trained SOM, with visualisation in grey scale (dark = high value, light = low value). Each small map representing one variable can be superimposed on the map representing the distribution of samples presented in Fig. 2, thus showing the relative values of each variable (in shades of grey) within each sub-area of the SOM

https://static-content.springer.com/image/art%3A10.1007%2Fs00468-008-0252-5/MediaObjects/468_2008_252_Fig4_HTML.gif
Fig. 4

Reflectance, transmittance and absorptance of (a) upper, middle and lower phyllodes according to axis category (A1, A2 middle and A2 lower) of 2-year-old trees and (b) of the middle phyllodes according to tree age (1-, 2- and 3-year-old) and axis category

Table 2

Results of F tests (two-way ANOVA) comparing the characteristics of the 124 sampled phyllodes according to position on the axis (“upper”, “middle” and “lower”), axis category (“A1”, “A2 Middle” and “A2 Lower” within the crown) and tree age (1-, 2- and 3-year-old)

Variables

Phyllode position

Axis category

Tree age

Phyllode position × Tree age

Tree age × axis category

F

P

F

P

F

P

F

P

F

P

Optical properties

Reflectance

2.28

Ns

10.92

***

11.03

**

0.09

Ns

1.21

Ns

Transmittance

19.58

***

7.87

***

68.62

***

0.23

Ns

0.03

Ns

Absorptance

15.88

***

2.80

Ns

76.27

***

0.50

Ns

0.01

Ns

Morphology

Phyllode area

1.47

Ns

23.56

***

0.12

Ns

3.78

*

6.21

**

Lamina thickness

1.00

Ns

78.32

***

0.92

Ns

0.12

Ns

33.11

***

LMA

22.03

***

1.80

Ns

4.34

*

1.88

Ns

5.34

**

Stomatal density

3.63

*

19.61

***

19.13

***

0.18

Ns

7.56

***

Biochemistry

Chl total

24.93

***

3.76

*

18.89

***

1.57

Ns

4.60

*

Chl a/b

14.31

***

5.25

**

61.43

***

8.72

***

1.58

Ns

Carotenoid

10.80

***

4.26

**

43.05

***

6.53

**

3.39

*

Na

4.82

**

3.35

*

18.31

***

2.07

Ns

4.66

*

The interactions between phyllode position and tree age or between tree age and axis category are also presented. F values (F) and level of significance (P) are given. The degrees of freedom are 2 for phyllode position, axis category and tree age and 4 for the interaction effects. Phyllode position × axis category interaction is non-significant for all the studied variables

Significance levels: ns = P > 0.05; * P < 0.05; ** P < 0.01, *** P < 0.001

Phyllode morphology

The maps of the phyllode area, lamina thickness, dry mass per area (LMA) and stomatal density showed similar trends (Fig. 3). The higher values of these four variables characterised Cluster V which grouped mainly middle position phyllodes on the main stem of 2- and 3-year-old trees (Figs. 2, 3). Whatever may be the branching order and the category of bearing axes, phyllodes located at the upper position on the axes had a lower LMA compared to those located at the middle and lower axis positions (Fig. 5a).
https://static-content.springer.com/image/art%3A10.1007%2Fs00468-008-0252-5/MediaObjects/468_2008_252_Fig5_HTML.gif
Fig. 5

Phyllode area, mass per unit area (LMA), stomatal density and lamina thickness of (a) upper, middle and lower phyllodes according to axis category (A1, A2 middle and A2 lower) of 2-year-old trees and (b) of the middle phyllodes according to tree age (1-, 2- and 3-year-old) and axis category

The phyllode area increased from branches in the lower crown to branches in the middle crown and on the main stem (Fig. 5b). No significant difference was found for phyllode area with tree age (Table 2). LMA increased with tree age for the phyllodes from the main stem, with values ranging from 73 to 106 g m² (Fig. 5b). A slight difference was observed on stomatal density according to the position of the phyllodes on their bearing axes (Table 2). The number of stomata per mm² increased with tree age for the phyllodes located on the main stem and on the median branches and the differences in stomatal density values increased according to axis category with tree age (Fig. 5b). The lamina thickness varied significantly according to the axis category with values ranging from 235 to 285 μm (Figs. 5b, 6; Table 2). The different values of lamina thickness between the main stem and the lower branches increased with tree age, indicating that vertical gradients became significantly steeper with increasing height (Fig. 5b; Table 2). Interaction of phyllode position with axis category and tree age was not significant for any of these morphological variables (Table 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs00468-008-0252-5/MediaObjects/468_2008_252_Fig6_HTML.gif
Fig. 6

Photomicrographs of 15 μm sections from A. mangium phyllodes. Sections represent phyllodes from (a) upper phyllode of the A1 of a 1-year-old tree (1A1u), (b) upper phyllode of the A1 of a 3-year-old tree (3A1u), (c) upper phyllode of the lower A2 of a 3-year-old tree (3A2Lu). c.p.m. Central parenchymateous mesophyll; u.c. “upper” cuticle; u.e. “upper” epidermis; l.c. “lower” cuticle; l.e. “lower” epidermis; p.p. palisade parenchyma

Phyllode anatomy

There were no structural anatomical differences whatever the phyllode sampling position and tree age. Each had a heavily cutinised epidermis and two layers of palisade parenchyma on both sides. The middle part of the lamina was characterised by a central parenchymateous mesophyll. Changes in total lamina thickness occurred mainly by difference in the central parenchymateous mesophyll thickness (Fig. 6).

Phyllode biochemistry

Whatever the axis category, the phyllodes located at the middle position on the bearing axis had significantly higher chlorophyll concentration compared to the two other positions (Fig. 7a). Maps of the total chlorophyll and nitrogen (Na) concentrations showed the same pattern (Fig. 3). The highest values characterised the phyllodes of the main stem of the 3-year-old trees (Cluster V, Figs. 2, 3) whereas the lowest values were associated with the 1-year-old trees and the lower branches within the crown (Fig. 7b). Phyllodes located at the middle position on the leafy axes had lowest chlorophyll a/b ratio and carotenoid concentration compared to those located at the upper and lower positions (Fig. 7a). Chlorophyll a/b ratio and carotenoid content had a similar map pattern (Fig. 3) and decreased significantly with increasing tree age (Table 2; Fig. 7b). Phyllodes from the lower branches had the highest average values of chlorophyll a/b ratio and carotenoid content compared to the phyllodes of the main stem. The interaction of phyllode position with tree age was significant for chlorophyll a/b ratio and carotenoid content (Table 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs00468-008-0252-5/MediaObjects/468_2008_252_Fig7_HTML.gif
Fig. 7

Total chlorophyll, chl a/b ratio, carotenoids and nitrogen content of (a) upper, middle and lower phyllodes according to axis category (A1, A2 middle and A2 lower) of 2-year-old trees and (b) of the middle phyllodes according to tree age (1-, 2- and 3-year-old) and axis category

Discussion

Age and phyllode characteristics

A recent study showed that A. mangium seedlings are characterised by a succession of different types of juvenile leaves before the phyllodes develop (Leroy and Heuret 2008). The development of the first phyllode occurs between nodes 8 and 16, depending on the seedling. Thus, phyllodes appear rapidly and the stages when there are juvenile leaves are limited to 0–5 months old seedlings. In the present study, we focused on highlighting morphogenetic trends by comparing phyllode characteristics for 1-, 2- and 3-year old trees. A further comparison of juvenile leaves and phyllodes would be interesting, as suggested by Brodribb and Hill (1993) who studied this aspect from a physiological point of view.

According to distal, middle and proximal positions on the axis, phyllodes may be associated to different ages. In our study, we have shown that absorptance was higher and transmittance was lower in relation to phyllode age on the axis whereas no variation appeared for the reflectance whatever the phyllode age. Studies on leaves of Carya illinoensis (Qi et al. 2003) and Corymbia gummifera (Choinski et al. 2003) also have shown the same pattern. On the other hand, Kitajima et al. (1997) did not reveal any differences in absorptance rate according to Anacardiumexcelsum and Antirrhoeatrichantha leaf age.

Phyllode area, lamina thickness and LMA were higher for older A. mangium phyllodes than for young ones. On various tropical species, LMA has been reported to increase with leaf age (Kitajima et al. 1997; Kitajima et al. 2002). The increase of LMA was explained in Urera by ash accumulation and was due to both ash and carbon accumulation in Cecropia (Kitajima et al. 2002).

Total chlorophyll content and Na were lower for both young and old A. mangium phyllodes and higher for mature ones. The opposite pattern was observed for chlorophyll a/b ratio and carotenoid content which were lower for mature phyllodes. This unusual profile of biochemical variables could be explained by an incomplete cellular structure. Indeed, Kitajima et al. (2002) have shown that the maximum leaf photosynthetic capacity was often not observed until 7–14 days after the date of full leaf expansion for Cecropia longipes and for Urera caracasana. Thus, the A. mangium phyllode biochemical content may still evolve after full lamina expansion. Total chlorophyll content and Na increased up to a maximum rate before decreasing with the senescence of the phyllode, as previously described for A. mangium by Yu and Ong (2000), as well as for other species (Bertamini and Nedunchezhian 2002; Choinski et al. 2003). The decrease in Na from mature to old phyllodes was similar to results reported for Anacardium excelsum and Antirrhoea trichantha leaves (Kitajima et al. 1997), which is presumably due to nitrogen reallocation from old to new leaves. The decrease of Na can also be assumed to contribute to the decline of photosynthetic capacity, as the light saturated rate of photosynthesis Amaxa (Amax related to the leaf area) and Na are known to be strongly correlated (Evans and Seemann 1989; Roggy et al. 2005). Yu and Ong (2000) showed that the photosynthetic capacity of phyllodes increased as they matured and decreased as they aged further. The rate at which nitrogen is reallocated from a leaf could depend on its relative position within a branch, on branch architecture, and on shading by the other branches (Kikuzawa 1995; Ackerly 1996). Nevertheless, in our case, the position of the phyllode within the crown did not affect the rate of N reallocation as the interaction between leaf age and axis category was non-significant. A decrease of carotenoid content from young to mature leaves has commonly been observed (Krause et al. 1995). Carotenoids have an important role in plant protection against oxidative damage due to high level of solar radiation. However, the increase of carotenoid content from mature to old phyllodes could be explained by modifications of the insertion angle of branches resulting in a self-shading phenomenon, even if the carotenoid degradation is known to be lower compared to total chlorophyll.

Phyllode characteristics and axis category within the crown

Significant changes in phyllode reflectance, phyllode area, lamina thickness, stomatal density, chlorophyll content, and Na were found between the main stem (A1), middle branches (A2M) and lower branches (A2L). These trends are more pronounced with the increase of tree age and the increase of crown size which results in a decrease of phyllode irradiance within the crown and particularly for lower branches.

Reflectance decreased according to axis order and with tree age. The higher reflectance rate found on phyllodes from the upper crown participates to reduced lamina overheating and transpiration as the light quantity is higher (Givnish 1984). However, A. mangium lamina reflectance rate was 2–3 times lower compared to other phyllodineous Acacias, which are reported to be very densely covered with trichomes (Yates 1992). Also at similar incident irradiance, an increase of phyllode absorptance and a decrease of transmittance may consequently be seen as a more efficient light interception, as phyllode area increases as well with tree age. Larger phyllode area and higher lamina thickness were observed on the main stem compared to the order 2 axes and were more pronounced with increasing tree age. A classic response to reduction of light is larger leaf area and thinner lamina (Lee et al. 2000; Uemura et al. 2000), but on young Betula papyrifera (Richardson et al. 2000) and young Picea (Ashton et al. 1998), leaves from the upper crown were, as in our present study, larger than those from the lower crown position. However, for mature trees of the same species, these authors observed the opposite, which might result from a limitation of water availability in the higher part of the crown (Ashton et al. 1998; Kock et al. 2004). So, in our case, variation in the phyllode area can reflect a difference of physiological age of the meristem rather than a variation of leaf irradiance. This result on leaf area is opposite to other studies on sun/shade leaves and in our case it can be seen as a different meristem expression between upper and lower crown positions.

Similarly, numerous studies pointed out an increase of leaf thickness and LMA according to crown position (Ashton et al. 1998; Richardson et al. 2000; Leal and Thomas 2003) and according to the quantity of light received (Sims and Pearcy 1994; Roggy et al. 2005). LMA variations between upper and lower crown were also more pronounced with tree age. An increase in LMA may enable the concentration of photosynthetic compounds per unit area (Gutschick and Wiegel 1988), and enhance the resistance to water limitation in leaves with more compact cells as suggested by Givnish (1988). Moreover, the higher lamina thickness could be explained as a response to a xeromorphic environment with a low relative humidity and a high transpiration demand (Poorter et al. 1995).

Also, stomatal density was found to be higher for phyllodes from the upper crown compared to the lower crown positions, as observed on Betula papyrifera by Ashton et al. (1998). The increase in stomatal density may be related to the increase in light, promoting higher carbon gains, or to the increase in temperature and air dryness, maximising transpiration rates and evaporative cooling (Lee et al. 2000; Roggy et al. 2005).

Total chlorophyll and Na content were higher for the upper crown and increased according to tree size. These results are consistent with those reported for Dicorynia guianensis (Roggy et al. 2005) and various Quercus species (Mediavilla and Escudero 2003). Similarly, higher values of Na for upper crown positions have been previously found for many species (Ellsworth and Reich 1993; Niinemets 1997; Le Roux et al. 1999). The observed vertical gradient of Na is a direct consequence of a vertical variation of LMA (Ellsworth and Reich 1993). Most studies have shown a strong correlation between high Na content and high local light level (Stenberg et al. 1998; Leal and Thomas 2003), and Na content and the light saturated rate of photosynthesis, Amaxa (Evans and Seemann 1989; Ellsworth and Reich 1993; Roggy et al. 2005). On various Acacia species, of which A. mangium, Eamus and Cole (1997) found a good correlation between Amaxa and Na. Thus, phyllodes from the high radiation upper crown might have a higher photosynthetic capacity compared to those from the low radiation lower crown position. Contrary to the phyllode total chlorophyll and Na content, carotenoid content and chlorophyll a/b ratio decreased with tree age. Also the higher contents were, for both variables, at the base of the crown compared to the top. Our results are different from those of Young (1991) who observed that sun leaves have a higher carotenoid content than shade leaves, and interpreted this as a way of avoiding photoinhibition. The particular structure of phyllodes, which has been described as being adapted to high irradiance and hot environment (Hansen 1996), might not require to synthesise carotenoids for photoprotection. We suggest also that the higher carotenoid content of phyllodes from the lower crown could increase light harvesting ability, as already discussed by Valladares et al. (2000).

Use of the self organising map

We used the SOM visualisation as an indirect gradient analysis to point out relationships between sample locations and biological variables in a single analysis. Combining ordination, classification, and gradient analysis functions, the SOM is a powerful tool used to visualise such data, without prior transformation. Specifically, the input data were reduced to two dimensions, preserving as much as possible of the information on the relationships between input signals. Through this process, neighbouring input values were mapped into the same or neighbouring nodes according to the metrics defined in the output space, and neighbourhood topology was preserved. Additionally, the SOM averages the input dataset in weight vectors through the learning process and thus removes noise (Vesanto et al. 1999). Specifically, given the conditions and aims of our study, SOM offers a clear bi-dimensional visualisation of the data, which considerably facilitates their analysis. During the learning process of the SOM, neurons that are topographically close in the array activate each other to learn from the same input vector. This procedure results in a smoothing effect on the weight vectors of neurons (Kohonen 2001). Thus, these weight vectors approximate the probability density function of the input vector. Subsequently, the visualisation of elements of these vectors for different input variables is convenient for understanding the contribution of each input variable with respect to the clusters on the trained SOM. The SOM helped to take a maximum of data aspects into account (qualitative, quantitative) and to clearly visualise the results in a single two-dimensional projection.

Conclusion

In the present analysis, we showed that phyllodes from large trees display more morphological and biochemical plasticity than phyllodes from small trees. Local leaf irradiance can be interpreted as a factor changing quantitatively the phenomenon of meristem maturation. Indeed, our study indicates a strong light gradient in 1-, 2- and 3-year-old A. mangium trees. Also, from a former study, a linear relationship has been established between internode length and associated phyllode length or area, which is strongly dependant on the topological location within the entire crown (Leroy 2005). Thus, the phyllode characteristics in A. mangium reflect the physiological age of the meristem. Phyllode structure can thus be a reliable marker of different stages of tree development. However, in our study it is difficult to dissociate light irradiance of leaves from the physiological age of the building meristem. A way to discriminate these two factors could be to analyse phyllode characteristics in both sun and shade conditions. Alternatively, from an experimental point of view, homologous axes on a same tree crown could be studied with or without a sun shield. Finally, results from the present study provide a clearer understanding of structural and physiological development and can be used as a support for leaf or phyllode sampling protocols. Indeed, many studies focus on plant responses under global change scenarios, and leaf plasticity within an individual plant has to be taken into account to avoid mixing different leaf growth phases.

Acknowledgments

The authors thank S. M. Setompul, Christina Putri and Evi for their help in field- and laboratory work. We are grateful to J. Orivel for his help with the SOM Toolbox and H. Cochard and Y. Caraglio for their comments on the manuscript. This research was carried out as part of the SAFODS (Smallholder AgroForestry Options for Degraded Soils) collaborative research project, partly funded by the EU: contract number ICA4-CT-2001-10092. AMAP (Botany and Computational Plant Architecture) is a joint research unit which associates CIRAD (UMR51), CNRS (UMR5120), INRA (UMR931), IRD (R123), and Montpellier 2 University (UM27); http://amap.cirad.fr/. The supports are gratefully acknowledged.

Copyright information

© Springer-Verlag 2008