Photosynthetic capacity is negatively correlated with the concentration of leaf phenolic compounds across a range of different species.

BACKGROUND AND AIMS: Phenolic compounds are the most commonly studied of all secondary metabolites because of their significant protective-defensive roles and their significant concentration in plant tissues. However, there has been little study on relationships between gas exchange parameters and the concentration of leaf phenolic compounds (total phenolics (TP) and condensed tannins (CT)) across a range of species. Therefore, we addressed the question: is there any correlation between photosynthetic capacity (A(max)) and TP and CT across species from different ecosystems in different continents?
METHODOLOGY: A plethora of functional and structural parameters were measured in 49 plant species following different growth strategies from five sampling sites located in Greece and Australia. The relationships between several leaf traits were analysed by means of regression and principal component analysis. PRINCIPAL
RESULTS: The results revealed a negative relationship between TP and CT and A(max) among the different plant species, growth strategies and sampling sites, irrespective of expression (with respect to mass, area or nitrogen content). Principal component analysis showed that high concentrations of TP and CT are associated with thick, dense leaves with low nitrogen. This leaf type is characterized by low growth, A(max) and transpiration rates, and is common in environments with low water and nutrient availability, high temperatures and high light intensities. Therefore, the high TP and CT in such leaves are compatible with the protective and defensive functions ascribed to them.
CONCLUSIONS: Our results indicate a functional integration between carbon gain and the concentration of leaf phenolic compounds that reflects the trade-off between growth and defence/protection demands, depending on the growth strategy adopted by each species.

Introduction

Data for various leaf traits encompassing many species are important in order to understand the different plant strategies and the adaptation of each species in a particular environment. The correlations between leaf traits provide insights into the selective pressures that have shaped the evolution of vegetation, and can help with the calibration of models predicting vegetation and productivity dynamics with respect to climate and land-use change (Reich ; Wright ; Shipley ; Westoby and Wright 2006; Kattge ).

A great variation in photosynthetic capacity (Amax) is evident both within and between species. Within-species variation in Amax has been ascribed to variations in leaf nitrogen concentration (Field and Mooney 1986; Evans 1989), due to the large fraction of leaf nitrogen that is invested in the photosynthetic apparatus. This is the reason that generally a strong positive correlation between Amax and leaf nitrogen concentration within species has been observed. Between-species variation in Amax occurs even among C3 species, although they share the same photosynthetic metabolism. A global survey dealing with 1 % of vascular plant species on Earth revealed that Amax varied by 120- and 40-fold when expressed on a dry mass and a leaf area basis, respectively (Glopnet; Wright ). Such a large variation is believed to be related to the growth strategy and/or the niche of a specific species. Regarding growth strategies, higher Amax is found in fast-growing species and in species with a shorter leaf lifespan than in slow-growing ones and species with a long lifespan. Regarding niches, higher Amax is found in sun than in shade species, and in early successional than in late successional species (Chabot and Hicks 1982; Gulmon and Mooney 1986; Poorter ; Lambers and Poorter 1992; Reich , 1997; Wright ; Hikosaka 2010). The magnitude of Amax of a species is associated with other functional and structural characteristics of the leaves. Therefore, strong relationships between Amax and other key leaf parameters among species are observed globally. For example, net photosynthetic capacity with respect to mass (Amax,m) is positively correlated with nitrogen content with respect to mass (Nm) and negatively correlated with leaf mass per area (LMA). Probably, these relationships are a consequence of the growth strategy of each species, e.g. rapid- versus slow-growing species (Coley ; Coley 1988; Lambers and Poorter 1992; Reich ).

Phenolic compounds are the most commonly studied of all secondary metabolites because of their significant concentration and their significant roles in plant tissues (Waterman and Mole 1994; Harborne 1997). The term ‘phenolic’ is used to define substances that possess one or more hydroxyl (OH) substituents bonded onto an aromatic ring. This highly diverse group of secondary metabolites includes mainly simple phenols, lignans, coumarins, flavonoids, tannins and quinines (Waterman and Mole 1994). These compounds fulfil at least three functions: (i) as defensive compounds—they inhibit the activity of herbivores or pathogens (Bennett and Wallsgrove 1994; Roberts and Paul 2006); (ii) as sun-screens—they reduce UV and visible-light penetration to sensitive tissues (Caldwell ; Middleton and Teramura 1993); and (iii) as antioxidants—they are involved in reducing damage by reactive oxygen species (Rice-Evans ; Close and McArthur 2002; Heim ; Sakihama ; Jaleel ). The biosynthesis of phenolic compounds requires energy, carbon skeletons and investment of additional nutrients such as nitrogen, which are diverted from primary metabolism. Therefore, allocation of photosynthetic products and nutrients must be balanced between normal growth processes and defense/protection demands (Herms and Mattson 1992).

Since there have been relatively few studies of the associations between photosynthesis and the concentration of total phenolic (TP) compounds and condensed tannins (CT) across species (see Kattge ), it would be interesting to examine whether there is any correlation between Amax (and probably other gas exchange parameters) and TP and CT among species. In order to test such a general concept it is important to include common plant species thriving in different climate zones and ecosystems. In the present study, we tested this hypothesis using 49 common plant species of different life forms from different sampling sites located in the east Mediterranean basin (Greece) and Australia.

Materials and methods

Plant material and study sites

Our data set comprises common vascular plant species from two different countries located in two different continents differing in climate, biogeography and soil conditions: two Greek and three Australian ecosystems (see Table 1). The general criteria for species selection included: (i) a species had to be common, because common species are probably better adapted to the local conditions and, also, have a reasonably good chance of impacting upon major ecosystem processes (Grime 1998; Diaz ) and (ii) the collection had to cover a wide range of growth forms, families and habitats. Our species selection was mainly local assembly based and restricted for practical reasons to common species possessing sufficient leaf size for gas exchange measurements. In Greece, samplings and field measurements were conducted during late spring–early summer (May–June) on Mount Parnitha, Attica, Central Greece, in a typical maquis and phrygana (garigue) formation from 2006 up to 2008, and at Domnista, Eurytania, Central Greece, in a typical temperate forest of southern Europe in 2007. In Australia, corresponding samplings and field measurements were conducted during late spring–early summer (November–December) of 2006 at the following sites: Snowy Plains (New South Wales, Eastern Australia), at a subalpine ecosystem, Britannia Creek—Yara Valley (Victoria, Southern Australia), in a typical temperate ecosystem representing the wettest–coldest edge of the Mediterranean-type ecosystems and Perth (Western Australia, South-western Australia), in a typical Mediterranean-type ecosystem with climatic conditions similar to those in Parnitha, Greece (Table 1). A total of 49 plant species, 32 native in Greece and 17 native in Australia, were studied (Table 2).

Table 1

Study site coordinates and climatic data.

Site	Description	Coordinates of meteorological station (lat.; lon.)	Altitude of meteorological station (m.a.s.l.)	Altitude of study site (m.a.s.l.)	Tmin (°C)	Tmax (°C)	Precipitation (mm)
Snowy plains	Subalpine Eucalyptus pauciflora woodlands	S36°17′38″; E148°58′21″	930	1400–1500	3.9	18.1	502.2
Britannia creek	Open messmate forests	S37°51′36″; E145°44′24″	189	400–600	7.0	18.5	1445.7
Perth	Botanic garden (King's Park & Botanic Garden)	S31°55′39″; E115°58′35″	15.4	50	12.1	24.3	781.9
Domnista	Deciduous broadleaf Quercus frainetto and Castanea sativa forests	N38°54′00″; E21°48′00″	690	1000	4.5	18	1255
Parnitha	Deciduous (Quercus macrolepis) and evergreen (Quercus coccifera and Pistacia lentiscus) open woodlands	N38°06′05″; E23°46′48″	235	200–400	6.5	27.8	446

Table 2

Studied plant species (per year and study site). The life form of the species is also presented.

No.	Species	Year	Site	Life form	Family
1	Derwentia derwentiana	2006	Snowy Plains	Herb	Scrophulariaceae
2	Eucalyptus pauciflora	2006		Tree	Myrtaceae
3	Acacia obliguinervia	2006		Shrub	Mimosaceae
4	Tasmannia xerophila	2006		Shrub	Winteraceae
5	Olearia megalophylla	2006		Shrub	Asteraceae
6	Daviesia mimosoides	2006		Shrub	Fabaceae
7	Eucalyptus sieberi	2006	Britannia Creek	Tree	Myrtaceae
8	Rubus sp.	2006		Climber	Rosaceae
9	Eucalyptus radiata	2006		Tree	Myrtaceae
10	Correa reflexa	2006		Shrub	Rutaceae
11	Correa lawrenciana	2006		Shrub	Rutaceae
12	Olearia lirata	2006		Shrub	Asteraceae
13	Pomaderris aspera	2006		Shrub	Rhamnaceae
14	Platylobium formosum	2006		Shrub	Fabaceae
15	Banksia menziesii	2006	Perth	Tree	Proteaceae
16	Corymbia calophylla	2006		Tree	Myrtaceae
17	Eucalyptus marginata	2006		Tree	Myrtaceae
18	Pistacia terebinthus	2006	Parnitha	Tree	Anacardiaceae
19	Quercus ithaburensis	2006		Tree	Fagaceae
20	Pistacia lentiscus	2006		Shrub	Anacardiaceae
21	Platanus orientalis	2006		Tree	Platanaceae
22	Rubus fruticosus	2006		Shrub	Rosaceae
23	Olea europaea	2006		Tree	Oleaceae
24	Styrax officinalis	2006		Shrub	Styraceae
25	Rosa cannina	2006		Shrub	Rosaceae
26	Pyrus amygdaliformis	2006		Tree	Rosaceae
27	Smilax aspera	2006		Climber	Smilaceae
28	Phlomis fruticosa	2006		Shrub	Lamiaceae
29	Quercus coccifera	2006		Shrub	Fagaceae
30	Malva sylvestris	2007		Herb	Malvaceae
31	Thapsia garganica	2007		Herb	Umbelliferae
32	Echinops viscosus	2007		Herb	Asteraceae
33	Securigera securidaca	2007		Herb	Fabaceae
34	Bituminaria bituminosa	2007		Herb	Fabaceae
35	Lotus ornithopodoides	2007		Herb	Fabaceae
36	Castanea sativa	2007	Domnista	Tree	Fagaceae
37	Clematis vitalba	2007		Climber	Ranunculaceae
38	Quercus frainetto	2007		Tree	Fagaceae
39	Juglans regia	2007		Tree	Juglandanceae
40	Ostrya carpinifolia	2007		Tree	Betulaceae
41	Rubus sp.	2007		Climber	Rosaceae
42	Tussilago farfara	2007		Herb	Asteraceae
43	Fragaria vesca	2007		Herb	Rosaceae
44	Platanus orientalis	2007		Tree	Platanaceae
45	Ballota acetabulosa	2008	Parnitha	Herb	Lamiaceae
46	Cercis siliquastrum	2008		Tree	Fabaceae
47	Cionura erecta	2008		Shrub	Apocynaceae
48	Anchusa sp.	2008		Herb	Boraginaceae
49	Arbutus unedo	2008		Tree	Ericaceae

For each species, sampling and measurements were conducted on three adult individuals and two fully expanded, current growth season's leaves per individual. All three individuals were within an area of 100-m radius and had a similar age with leaves accessible for in planta measurements (regarding the limitations of the instruments). Measurements were conducted on fully expanded and sun-morphotype leaves: south-east facing (in the Northern Hemisphere) and north-east facing (in the Southern Hemisphere). There were no indications of temporary shading during their expansion. Leaves with obvious symptoms of herbivore or pathogen attack and senescent leaves of the previous growth period were excluded. Laboratory measurements were conducted on the same six leaves that were used for gas exchange measurements. The leaves were collected after measurements of gas exchange had been completed (see below), wrapped in sealed plastic bags and immediately transported to the laboratory in a portable coolbox.

Morpho-anatomical measurements

For calculation of the LMA (g m−2), leaves were oven-dried at 70 °C for 48 h. Leaf lamina area was determined from photographs of the leaves by image analysis using Image-Pro Plus (version 3.01, Media Cybernetics, Silver Spring, MD, USA). Image spatial calibration was ensured by the incorporation of a ruler and samples were photographed from a position perpendicular to the sample plane to avoid geometric distortion of the images. Leaf mass per area was estimated as the ratio of leaf dry mass to leaf area (g m−2). For total leaf thickness (LT) measurements, hand-cut cross-sections were made on fresh leaves (replicates as above) of all samples. Leaf density (LD; g cm−3) was calculated according to Witkowski and Lamont (1991) as the ratio between LMA and LT.

Gas exchange parameters

Measurements of gas exchange were conducted between 0900 and 0012 h. Gas exchange parameters were: photosynthetic capacity (Amax), transpiration (E) and stomatal conductance (gs) were measured in two leaves per individual; three individuals (six samples) using a portable photosynthesis system LI-6400 (Li-Cor Inc., Lincoln, NE, USA). Amax,a was measured at ambient CO2 atmospheric concentration under saturating photosynthetic photon flux density (∼1850 μmol m−2 s−1 PPFD). Leaves were acclimated to the saturating light intensity until rates of Amax,a stabilized. Amax,m was calculated as the ratio of Amax,a to LMA. E was also expressed with respect to mass (Em).

Total nitrogen concentration

After weighing, dried plant material of two leaves per individual for three individuals (six samples) was ground to a fine powder with a ball mill. Nitrogen and carbon concentrations were determined by Dumas combustion. Aliquots of 50 mg of the finely ground foliage samples were combusted to N2 and CO2 in the presence of O2, and quantified by means of thermo-conductivity (LECO CHN2000, St Joseph, MI, USA).

The total nitrogen concentration of samples from Greek species was measured by the micro-Kjeldahl digestion method, properly modified for accurate measurements of small amounts of leaf samples and analysed colorimetrically (Mills and Jones 1996).

To assess the variability due to analysis with two different methods, comparisons were made between selected samples from Greece and Australia. The difference between the two analytical methods was constant (∼1.5 %) and thus values were adjusted accordingly. Total nitrogen concentration was expressed per total leaf area (Na) and per dry mass (Nm).

Total phenolic compounds and CT determination

Total phenolic compounds were measured in two leaves per individual for three individuals (six samples) according to the Folin–Ciocalteu method as described by Waterman and Mole (1994). Tannic acid (Sigma, USA) was used for a reference curve. Although the reagent also reacts with substances other than phenolic compounds, we used this method because it is recommended for corresponding field studies, it is the most popular and therefore, the results are comparable to the majority of studies (see Harborne 1989; Waterman and Mole 1994). Condensed tannins were determined according to the proanthocyanidin method as described by Waterman and Mole (1994). Delphinidin (Extrasynthese S.A., Genay, France) was used for the reference curve. The concentrations of TP compounds, CT and their sum (TP + CT) were expressed per total leaf area, per dry mass and per nitrogen content (TPN, CTN).

Stable carbon and nitrogen isotope signatures

All samples from two leaves per individual for three individuals (six samples) were analysed with a ThermoScientific Delta V Plus mass spectrometer. The samples were introduced into a Thermo-Flash EA elemental analyser where CO2 and N2 gas were produced by combustion at 1020 °C. The gases, moved along in a continuous flow of helium, were separated by a GC column, and then introduced into a continuous flow gas source mass spectrometer for carbon and nitrogen isotopic ratio determination. The isotopic ratios are expressed for carbon as δ13C versus Pee Dee Belemnite (a marine carbonate), and for δ15N versus N2 (atmospheric N2):

where X is the δ13C or δ15N value and R = 13C/12C and δ15N/δ14N, respectively.

The isotopic analyses were performed in the Stable Isotope Unit of the Institute of Materials Science (NCSR Demokritos), accredited according to EN ISO/IEC 17025:2005. Repeated measurements were made for each of the samples. Analytical precision was 0.1 ‰ for δ13C and 0.2 ‰ for δ15N values.

Data analysis

Spearman bivariate correlations among the pairs of all 20 initial parameters [see Additional Information Table S1] were performed with SPSS Statistics (version 17.0, IBM® SPSS® Statistics, New York, NY, USA) at a 95 % level of significance and correlation coefficients were recorded [see Additional Information Table S2]. Only eight parameters were selected for further analysis (Table 3) after elimination of derivatives and parameters that were different expressions of the same trait and the bivariate correlations between them had no physiological meaning.

Table 3

Abbreviations and units of the eight leaf traits examined.

Leaf trait	Abbreviation	Units
Leaf mass per area	LMA	g m−2
Leaf density	LD	kg m−3
Net photosynthetic capacity with respect to mass	Amax,m	nmol CO2 g−1 s−1
Transpiration rate	E	mmol H2O m−2 s−1
Nitrogen isotopic composition	δ15N	‰
Nitrogen content with respect to mass	Nm	mg N g−1
Concentration of total phenolic compounds with respect to mass	TPm	mg tannic acid mg−1 d.w.
Concentration of condensed tannins with respect to mass	CTm	mg delphinidin mg−1 d.w.

Regression analyses were performed to determine the type of relationship that exists between pairs of defined parameters, the strength of the curve and coefficients of determination (r), and the statistical significance of correlation coefficients was recorded. Regression analysis was performed using Statgraphics Plus v. 4, (StatPoint Technologies, Inc., Warrenton, VA, USA) at a 95 % level of significance on the means of six samples per species. Correlations were displayed graphically as scatter graphs using SigmaPlot 11.0 (Systat Software, Inc., San Jose, CA, USA).

The principal component analysis (PCA) for eight parameters was conducted with CANOCO (version 7.0.61.0, StatSoft, Inc., Tulsa, USA) after log transformation.

Results

Significant negative relationships of Amax,m with the concentration of TP compounds, the concentration of CT, and the sum of TP and CT were found irrespective of how these concentrations were expressed (with respect to mass, area or nitrogen content [see Additional Information Table S2]). However, these relationships became stronger when the concentration of phenolic compounds was expressed per unit of nitrogen [see Additional Information Table S2]. Although the strength of the correlations was trait specific, in general, the concentrations of phenolic compounds were negatively correlated to gas exchange parameters and positively correlated to leaf structural parameters (Table 4). Regression analysis showed that the best-fitting model to the prediction of the relationships between Amax,m and TPm (Fig. 1A, r = 0.52) and CTm (Fig. 1B, r = 0.59 at P < 0.01) was the reciprocal one (type of equation: Y=1/(a + bx) [1]). Moreover, all the relationships between TP, CT and gas exchange traits [Amax,m, transpiration rate (E), stomatal conductance and intrinsic water-use efficiency] [see Additional Information Table S2] were described by the same model [though correlation and regression (data not shown) coefficients were lower].

Table 4

Spearman rank correlations for each pair of the eight traits examined.

	LMA	LD	Amax,m	E	δ15N	Nm	TPm
LD	0.595**
Amax,m	−0.657**	−0.588**
E	−0.288*	−0.409**	0.837**
δ15N	−0.601**	−0.508**	0.461**	n.s.
Nm	−0.323*	n.s.	0.362**	n.s.	n.s.
TPm	n.s.	0.351*	−0.415**	−0.346*	n.s.	n.s.
CTm	0.336*	0.350*	−0.521*	−0.455	−0.330*	n.s.	0.616**

Correlations with coefficient rs > 0.5 are in bold.

*P < 0.05.

**P < 0.01.

Fig. 1

Fit of reciprocal model and regression coefficients. (A) For net photosynthetic capacity with respect to mass (Amax,m) and concentration of phenolic compounds with respect to mass (TPm); r = 0.52, at P < 0.01, (B) for net photosynthetic capacity with respect to mass (Amax,m) and concentration of condensed tannins with respect to mass (CTm); r = 0.59, at P < 0.01, (C) for net photosynthetic capacity with respect to mass (Amax,m) and LMA; r = 0.57, at P < 0.01.

The matrix of rank correlation and coefficients of determination (rs) among all possible pairings of the eight examined traits (Table 4) confirmed some already known positive correlations, such as the ones between Amax,m and E, between Amax,m and Nm, and the negative correlation between Amax,m and LMA (Table 4, Fig. 1C). The results from the regression analyses showed that the relationship between LMA and Amax,m, (r = 0.57, at P < 0.01) was also reciprocal (Fig. 1C). Nitrogen isotopic composition (δ15N) was negatively correlated with leaf structural traits (LMA, LD) and positively correlated with Amax,m (Table 4).

In the PCA (Fig. 2), the first two axes accounted for 77.9 % of the total variation. Axis 1 (first PC), which explained 59 % of the total variation, was well associated with traits related to growth (C, N gain and water losses—negative side of the axis) and protection (water saving and defense/protection—positive side). According to the eigenvector values of the traits on the first PC (Table 5), Amax,m, E, LMA and CTm had the highest scores. Axis 2 (second PC: 18.9 % of the total variation) was associated with TPm and CTm, the traits that had higher eigenvector values on this axis (Table 5).

Fig. 2

Principal component analysis biplot of plant samples (Greece: The percentage of total variance explained by each PC (first PC: 59 %; second PC: 18.9 %) and eigenvector values of the leaf traits are shown in Table 5.

Table 5

Eigenvector values of eight leaf traits on the first two PCA axes in Fig. 2.

	1st PC	2nd PC
	59 %	18.9 %
Amax,m	−0.937	0.317
E	−0.799	−0.153
CTm	0.703	0.610
LMA	0.664	−0.296
LD	0.642	−0.044
TPm	0.558	0.675
Nm	−0.406	−0.213
δ15N	−0.364	−0.045

Values are ranked in the order of absolute magnitude along the first PC. The higher value for each parameter between the two axes is in bold.

Increasing values on the first PC indicated a trend for higher water saving and defensive/protective demands (high LMA, LD, TPm and CTm). Therefore, Australian plants that have higher LMA values are clearly separated from Greek plants (Fig. 2).

Discussion

The most important findings of the present study concern the negative correlations between Amax,m and TP and/or CT, irrespective of expression. Although the two curves Amax,m–TP and Amax,m–CT are described by the same model, the different strengths of these correlations may indicate a differentiation in the functional roles of TP and CT within plant tissues. The direct relationship between Amax and TP has not been detected earlier, probably since ecological studies using a large number of species did not include the measurement of phenolic compounds, whereas studies in which TP or CT was measured did not include an efficient number of species or life forms. In a recent study, Ishida found that leaf TPN or CTN was positively correlated to LMA and negatively correlated to Amax,m and Amax,a. However, the correlations were weak (r < 0.5), probably because all the plants examined were drought tolerant and no herbs were included. Our results confirmed these correlations.

The results of the PCA indicate an interaction between growth (parameters associated with C and N gain, such as Amax, Nm, E and δ15N, which is an indicator of soil N availability; see Schmidt and Stewart 2003; Craine ) and defense/protection demands (parameters associated with mechanical and chemical reinforcement, such as LMA, LD and phenolic compounds), which is in accordance with the hypothesis of Herms and Mattson (1992). Additionally, the implication of growth strategy limitations in the interaction between structure and the concentration of phenolic compounds is also indicated since the analysis showed that high TP and CT are associated with thick, dense leaves with low N. Species with this leaf type are slow growing and are common in environments with low water and nutrient availability, and high temperatures and light intensities (Wright , 2004; Poorter ), conditions that may also increase both the risk of photodamage and herbivory, and subsequently justify high TP and CT. This particular leaf type represents an indicator of the growth strategy of each species, and this is of use from the curves between Amax,m and TP and CT. Indeed, the majority of herbs are positioned on the left part of the curve TPm–Amax,m (Fig. 1A). On the other hand, species that are characterized by low Amax,m and high TP and CT (mainly evergreen trees and shrubs) are positioned on the right part of the curve. This is in accordance with the alreadyknown trend that fast-growing species (mainly herbs) possess low levels of leafdefensive compounds, including TP, whereas slow-growing species show high levels of TP (Coley 1983, 1988; Coley ; Bryant ; Herms and Mattson 1992; Endara and Coley 2011).

Conclusions and forward look

Our results indicate a functional integration between carbon gain and the concentration of phenolic compounds concentration that reflects the trade-off between growth and defence/protection demands, depending on the growth strategy adopted by each species. Further investigation is needed in terms of sample size and/or meta-analysis in order to obtain a more integrated picture of the relationship between the concentration of phenolic compounds and carbon gain. It would also be interesting to investigate whether this relationship is strengthened in the case of species of the same genus. This remains to be answered in a future study.

Additional information

The following additional information is available in the online version of this article:

File 1. Table 1. List of abbreviations for all 20 leaf traits initially examined.

File 2. Table 2. Spearman rank correlations for each pair of the 20 traits initially examined.

Sources of funding

This work was conducted within a Greek–Australian bilateral co-operation project, funded by the Greek Ministry of Development, General Secretariat of Research and Technology. Financial support from the Greek Scholarship Foundation to S.S. as a post-graduate student is gratefully acknowledged.

Contributions by the authors

The project was conceived and planned by G.K. in collaboration with all co-authors. In Australia, the main experimental part of the study was carried out by M.N.F., G.T., V.L. and M.A.A. In Greece, the main experimental part of the study was carried out by S.S., as part of her PhD thesis, D.N., G.L. and E.D. The final data analysis was done by P.B. G.K. was the supervisor of S.S.