Testing the correlations between leaf life span and leaf structural reinforcement in 13 species of European Mediterranean woody plants

Introduction

The large differences in leaf longevity found among different terrestrial plant species and their effects on plant productivity have been the focus of intensive research. Many studies have reported that long leaf longevity should decrease photosynthetic rates per unit leaf mass (Reich, Walters & Ellsworth 1992; Reich et al. 1999), which has been interpreted as being the consequence of a trade-off between leaf traits that confer persistence and those that maximize instantaneous productivity (Reich, Walters & Ellsworth 1997; Warren & Adams 2000). Leaf mineral nutrients, in particular leaf nitrogen concentration, are generally known to affect the photosynthetic capacity when both nutrients and photosynthesis are expressed on a leaf mass basis (Reich et al. 1992, 1997). In turn, the negative effects of a high leaf mass per area (LMA) on photosynthetic rates per unit leaf mass have been observed across a wide range of life-forms, ecosystems and biomes (Reich et al. 1997, 1999). Thus, a high CO₂ assimilation rate on a leaf mass basis tends to be associated with short leaf longevity because a high assimilation rate requires a high N concentration per unit leaf mass and a low LMA, which increases leaf vulnerability to herbivory and other physical hazards (Wright & Cannon 2001; Shipley et al. 2006).

The high LMA typical of long-living leaves can be achieved either through increased thickness or density (Witkowski & Lamont 1991; Niinemets 1999). A high concentration of structural components contributes to increasing leaf density and mechanical strength (Turner 1994; Read & Sanson 2003; Takashima, Hikosaka & Hirose 2004). Cellulose, hemicellulose and, in some cases, lignin constitute a large fraction of the cell wall (Burgert 2006) and the stiffness of the cell wall varies according to the amount of these components (Vincent 1999; Burgert 2006). The presence of higher concentrations of lignin, cellulose and hemicellulose in leaves with a long life span has often been postulated (Chabot & Hicks 1982; Chapin, McKendrick & Johnson 1986; Williams, Field & Mooney 1989; Takashima et al. 2004). However, the experimental evidence for this assumption is scarce. A higher concentration of structural components per unit leaf mass would contribute to diluting N concentrations, and this should reduce photosynthetic rates and N use efficiency (Vitousek, Field & Matson 1990; Lloyd et al. 1992). In addition, several authors have reported higher construction costs per unit leaf mass in evergreen leaves because of their putative higher concentrations of expensive protective compounds such as lignin (Poorter & Villar 1997; Eamus et al. 1999). Accordingly, it is clear that – both through their implications for construction costs and through their effects on the assimilation rate – leaf structure and chemical composition are crucial if we are to explain the differences in leaf life span among different species.

In the present study, we analysed LMA, leaf thickness and the amount of fibre (cellulose, hemicellulose and lignin) in several woody species differing in leaf life span, in a Mediterranean climate. Although the literature contains many references to the relationships between leaf longevity and other leaf traits, such as LMA, nutrient concentrations and photosynthetic capacity (Reich et al. 1997, 1999; Warren & Adams 2000; Mediavilla & Escudero 2003a; Gulías et al. 2003; Wright et al. 2004), there are considerably fewer data available concerning fibre concentrations and their relationships with leaf life span or leaf habits (Castro-Díez et al. 1997; Damesin, Rambal & Joffre 1997; Villar et al. 2006). We are not aware of any study where fibre concentrations and their relationships with leaf longevity and other related leaf traits have been analysed in a set of species including such a large range of leaf life spans as that included in the present study.

Here, our main goal was to examine the way in which various leaf traits related to structure and chemical composition vary between species differing in their leaf life span. Consistent with previous hypotheses (Chabot & Hicks 1982; Coley, Bryant & Chapin 1985; Aerts & Chapin 2000; Hikosaka 2004), we predicted that, in addition to a greater thickness, a higher fibre concentration per unit leaf mass is also necessary to confer persistence to the leaves, and that increased requirements for leaf protection should lead to a steady increase in fibre concentrations as leaf longevity increases.

Materials and methods

Results

The results of the anova revealed that there were significant differences among species for all leaf features analysed (Table 2). As expected, LMA and thickness tended to increase with leaf longevity (Fig. 1), and both traits showed minor differences between different locations for the same species (Table 3). Fibre concentrations increased more than twofold between minimum and maximum values (Fig. 2), and again tended to be similar for the same species located in different plots (Table 3). Although the interspecific differences persisted within each leaf habit (deciduous or evergreen), the variability in fibre concentrations proved to be higher among the deciduous species, for which a greater coefficient of variation, especially in hemicellulose and lignin, was obtained despite their lower range of leaf-life spans. The lowest values in fibre concentration were always recorded among the deciduous species, while the highest value was always recorded in an evergreen. However, many evergreens attained fibre concentrations lower than some deciduous species (Fig. 2). On comparing the two groups of species, we observed that the differences were only significant for structural carbohydrates (mean concentration of hemicellulose + cellulose in deciduous species, 223 vs. 307 mg g⁻¹ in evergreens, F = 9·35, P = 0·0109), whereas there were no significant differences between the average of the deciduous and evergreen species in the case of lignin (mean concentration in deciduous species, 115 vs. 139 mg g⁻¹ in evergreens, F = 2·47, P = 0·14).

Across all species means, there was a strong positive correlation between leaf longevity and LMA and thickness (Table 4). Leaf longevity was also positively correlated with structural carbohydrates. Only the leaf lignin concentration seemed to be independent of leaf longevity. The same conclusions were reached using PICs.

The correlations between leaf life span and the other variables across all species might be due either to the direct effects of the variation in leaf longevity or to the contrast between the two leaf habits: deciduous or evergreen. LMA and thickness increased steadily with leaf longevity for the whole set of species, irrespective of leaf habit (Fig. 1). By contrast, the concentrations of cellulose and hemicellulose increased with leaf longevity when deciduous species were compared with evergreen ones, but seemed to be independent of leaf longevity for the set of evergreen species alone (Fig. 2). To better differentiate the effects of leaf habit from those of leaf longevity, we repeated the correlation analyses, this time restricting the comparisons to within each leaf habit. Among the deciduous species, all the relationships disappeared (data not shown), which may logically be attributed to the reduced range of variation in longevity and other leaf traits for this set of species, although the number of data was too low to establish significance levels in PIC correlations. Among all the evergreen species, the relationships between leaf longevity, LMA and thickness persisted both in TIPs and in PICs, but the relationships between leaf longevity and the fibre concentration disappeared (Table 4).

The concentration of structural carbohydrates seemed to be deeply affected by phylogeny: it was rather high (for a given leaf longevity) in all Quercus species, whereas the cellulose concentration was especially low, among the evergreens, in I. aquifolium (Fig. 2). When the comparisons were restricted to evergreen Quercus (Q. coccifera, Q. suber and Q. ilex) and Pinus (P. halepensis, P. pinea, P. pinaster and P. sylvestris) species, a significant positive correlation between leaf longevity and the cellulose concentration emerged, despite the low number of data available. This relationship was apparent only in PICs (Table 4).

Discussion

Clear differences were seen among the five oak species, five pine species, and three additional tree species studied here as regards both the morphology and the chemical composition of their leaves. By contrast, although most plants can show considerable phenotypic plasticity in leaf traits in response to environmental variability (Read & Stokes 2006), the intraspecific variation in the present study was low, probably because of the slight climatic differences between the plots inhabited by the same species. In most cases, the interspecific differences in leaf traits seemed to be related to differences in leaf longevity. Consistent with previous studies (Reich et al. 1992; Aerts & Chapin 2000), longer-living leaves tended to have higher thickness and mass per unit area than shorter-living leaves, which supports the hypothesis that longer-living leaves require mechanical protection to guarantee their persistence (Chabot & Hicks 1982; Coley et al. 1985; Turner 1994; Westoby, Warton & Reich 2000).

It has also been proposed that in longer-living leaves the allocation of biomass to structural components is greater, which contributes to their increased LMA and may favour an enhanced leaf longevity through leaf protection (Chabot & Hicks 1982; Williams et al. 1989; Turner 1994; Takashima et al. 2004). In our study, the structural carbohydrate concentration (cellulose and hemicellulose) indeed tended to increase with leaf longevity when all the species were compared (Table 4). However, in contrast to our own starting hypothesis, and unlike what has been proposed by many other authors (Aerts 1995; Damesin et al. 1997; Warren & Adams 2000; Villar & Merino 2001), we failed to observe any trend associated with leaf longevity in the case of lignin. Accordingly, it is possible that the interspecific differences in lignin concentration could be due to other requirements not strictly related to leaf duration, such as transport functions associated with the venation system of the leaf (Roth-Nebelsick et al. 2001). However, for us the reasons why the different kinds of leaves studied had such different amounts of lignin are difficult to unravel since these differences do not seem to be related to variations in water availability among the different sites (Table 3).

The results concerning lignin, however, should be taken with caution because of the uncertainties associated with the existing methods for lignin determination. Although Van Soest's method has proved to be a satisfactory alternative to characterize the carbohydrates in the plant cell wall (Hindrichsen et al. 2006), ‘lignin’ fractions may be contaminated by substances such as protein, cutin and suberin, while part of the lignin content of the plant material is potentially soluble in the acid detergent solution (Fukushima & Hatfield 2004). In fact, different confounding effects are also present in the other methods available, such that none of these methods should be considered a standard unambiguous method for all samples (Hatfield & Fukushima 2005). The acid detergent method used in the present paper shows good correlations with other known methods (Hindrichsen et al. 2006), and we are confident that it provides comparable results for the different species, although additional data on other sets of species and using other analytical methods would be necessary to confirm the trends (or rather, the lack of trends) in lignin concentrations observed by us.

Unlike the case of lignin, a significant correlation was observed between the structural carbohydrate concentration and leaf longevity (Table 4). However, much of the strength of this relationship seemed to be due to the contrast between evergreen and deciduous species, since it disappeared when compared only across the evergreen species. Accordingly, when contrasted across species, without regard to lineage (TIP correlations), the concentration of structural carbohydrates in our set of species is more closely related to the dichotomy between evergreen and deciduous habits than to leaf life span. This result is contrary to what could be expected: since a high fibre concentration is assumed to contribute to enhanced leaf resistance to herbivory and other physical hazards (Chabot & Hicks 1982; Turner 1994; Wright & Cannon 2001), it would be reasonable to assume that even within the evergreen species greater leaf longevity should be accompanied by a greater concentration of structural compounds that help to confer the leaves mechanical protection.

The differences in the concentrations of structural carbohydrates between evergreen and deciduous species might be easily explained by the fact that in a seasonal climate deciduous leaves are maintained only during seasons with relatively mild conditions (mainly spring and part of the summer), and hence the needs for structural protection may be also relatively low. In contrast, evergreen leaves must survive unfavourable seasons, and this requires a structural reinforcement (Givnish 2002), achieved with a higher concentration of cell wall components, which contributes to increasing freezing tolerance in plants (Rajashekar & Lafta 1996; Solecka & Kacperska 2003). Accordingly, in terms of the concentration in structural carbohydrates per unit leaf mass, physical leaf reinforcement varies, apparently, depending on whether the leaves must or must not confront unfavourable climatic conditions during their lives, although it is independent of the number of unfavourable seasons that the leaves of evergreens must survive, at least when one compares across species rather than looking at within-lineage divergences. Our results are at odds with those of Villar et al. (2006), who observed differences in lignin but not in total structural carbohydrates between evergreen and deciduous species, exactly the opposite pattern to what we have found. Those authors studied seedlings cultivated in a greenhouse, instead of mature trees under field conditions, and these conflicting conclusions may have resulted from the ecophysiological differences between seedlings and mature trees reported by different authors (Cornelissen et al. 2003). However, the differences in the analytical methods used cannot be discarded as a potential confounding factor.

However, when we look at the evolutionary divergences analysed by means of PIC correlations, we find evidence for correlated evolution between the concentration of structural carbohydrates and leaf longevity. As also reported by Villar et al. (2006), phylogeny appears to be a more important factor determining leaf fibre concentrations in our species than the functional differences imposed by natural selection. Here, I. aquifolium showed a very low cellulose concentration for an evergreen species, and the phylogenetic contrasts in which this species was involved were strongly affected by its extreme values. When we eliminated I. aquifolium, restricting the PIC correlation analysis to evergreen Quercus and Pinus species, a significant positive correlation was observed between leaf life span and cellulose concentration, whereas the correlation coefficient for hemicellulose was almost significant (P = 0·0583), despite the low number of data. This is also apparent from the data points shown in Fig. 2. This suggests that, within a given lineage, increases in leaf duration are accompanied by increases in cell-wall components, as could be expected (Hikosaka et al. 1998; Warren & Adams 2000; Hikosaka 2004).

The interspecific variations in fibre concentrations, thus, differ from the trends observed for other leaf traits, such as LMA and thickness, which increased steadily with the increase in leaf longevity, and this occurred irrespective of phylogeny, since this trend was also observed across species (Table 4). This steady change in these leaf traits gives rise to the cumulative negative effects of a long leaf life span on the instantaneous gas-exchange rates per unit leaf mass observed in many other papers (Reich et al. 1992, 1997, 1999; Hikosaka & Hirose 2000; Mediavilla & Escudero 2003a). In contrary to what had been previously suggested (Chabot & Hicks 1982; Chapin et al. 1986; Takashima et al. 2004), the structural reinforcement due to the increase in fibre concentration per unit leaf mass is not strictly related to leaf longevity, although the concentration of structural carbohydrates depends on the leaf habit. Thus, evolutionary shifts between deciduous and evergreen leaf habits may have profound effects on the needs for a structural reinforcement with greater amounts of structural carbohydrates. Nevertheless, within the evergreen species co-occurring in a particular environment, leaf duration may apparently be increased with no need for cumulative investments in structural components per unit leaf mass (Table 3), although obviously the amount of structural material per unit leaf area tends to increase as LMA increases with leaf longevity. However, within a particular lineage, reinforcement with both more structural carbohydrates per unit mass and a higher LMA is apparently necessary to permit increases in leaf life span. In view of the phylogenetic influence on the relationships between leaf life span and fibre concentrations, more research is needed to check whether the trends observed in the present work might be generalizable to other phylogenetic groups and other plant communities. In addition, it is also necessary to investigate other leaf traits that may help to determine leaf protection, such as leaf shape, architecture, anatomy and alternative anti-herbivory defences (Niklas 1999; Balsamo et al. 2003; Read, Sanson & Lamont 2005; Read & Stokes 2006), as well as their interactions with the leaf chemical composition.