By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Please be advised that we experienced an unexpected issue that occurred on Saturday and Sunday January 20th and 21st that caused the site to be down for an extended period of time and affected the ability of users to access content on Wiley Online Library. This issue has now been fully resolved. We apologize for any inconvenience this may have caused and are working to ensure that we can alert you immediately of any unplanned periods of downtime or disruption in the future.
1. Multiple functions of plant silicon are known from agricultural species where silicon fertilizer alleviates biotic and abiotic stress impacts. In contrast, the function of plant silicon in natural ecosystems is often overlooked and is relatively poorly understood.
2. We investigated a potential integration of silicon with the leaf dry mass economics framework. We examined the relationship between leaf longevity, a trait considered to optimize leaf carbon use, and leaf silicon concentration across plant functional types and phylogenetic groups, testing the hypothesis that short-lived leaves have higher silicon content to maximize carbon allocation to growth. We considered leaf longevity a continuous trait (months, n = 155 species) and a binary trait (annual vs. perennial, n = 602 species).
3. A significant negative correlation was found across all species between leaf longevity (months) and relative silicon concentration, indicating that across functional types and plant families, leaves with shorter life spans contain higher concentrations of silicon. A similar significant relationship was also found within deciduous angiosperm leaves. Silicon concentration was significantly higher in annual vs. perennial species across most plant functional types. Within order and family, no significant difference in silicon concentration was found between annual and perennial leaves except in the Poales and the Poaceae, where annual leaves had higher silicon concentration. Relative leaf/shoot silicon concentration varied much more than leaf longevity across the phylogeny, and there was a stronger phylogenetic signal from silicon concentration than from leaf longevity. Evolutionary divergence analysis showed that divergences between relative silicon concentration and leaf life span (months) were not significantly correlated; however, using a larger data set, leaf life span (as an annual/perennial binary trait) and relative silicon concentration were significantly correlated, implying that across a larger data set, shifts to higher relative leaf/shoot silicon concentration are consistently associated with shifts to annual foliage.
4. We suggest that in shorter-lived leaves, silicon could be a metabolically cheaper alternative to carbon, allowing a more favourable leaf carbon balance over short periods. Silicon could be a less versatile, but more disposable resource, with the potential to function in structural, stress alleviation and defensive roles by substituting for carbon that could be allocated to further growth and reproduction.
The concentration of silicon in a plant is influenced by the amount of available silicic acid in the substrate (Liang et al. 2007) and is related to phylogeny (Takahashi, Tanaka & Miyake 1981; Hodson et al. 2005). Plant silica concentration is regulated by silicic acid uptake, with silicon transporters to assist silicic acid uptake from the soil solution and xylem loading recently identified and mapped in rice roots (Ma et al. 2006, 2007; Ma & Yamaji 2008). Ma & Takahashi (2002a), referring to agricultural species, separated plant groups into active accumulators (>1% Si), passive accumulators (0·5–1% Si) and excluders of silicon (<0·5% Si). However, these groups are somewhat arbitrary, and leaf silicon concentration may be better considered a continuous spectrum. Hodson et al. (2005) in an analysis of published leaf silicon concentration data showed that in general, non-vascular plants, clubmosses and horsetails (Equisetum sp.) accumulated more silicon than gymnosperms, angiosperms and ferns. Leaf silicon concentration variation within the latter two groups was consistent with higher-level groupings, with Poaceae and Arecales being examples of high accumulating groups among angiosperms.
It has previously been suggested that silicon may be substitutable for carbon for some plant functions. Raven (1983) proposed that silica can be a compression-resistant alternative to lignin and cellulose. McNaughton et al. (1985) suggested that silica can act as mineral-generated structural support as opposed to support produced by carbon, which could allow a more favourable leaf carbon balance. O’Reagain & Mentis (1989) extended this further by suggesting that in grasses where leaves are relatively short lived, silica may be used as an energetically cheaper but disposable structural material, in comparison to dicotyledonous plants that have relatively longer-lived leaves, in which it may be more beneficial to invest in better quality but more expensive compounds such as lignin. Given that silicon is ubiquitous in plants, silicon use in short-lived leaves may not be limited to grasses, but may be an important factor in leaf function and ecological strategies throughout the plant kingdom.
Plant functional groups and traits, though not independent of phylogeny, are an alternative framework in which to consider plant functional ecology (Westoby & Wright 2006). The global leaf economics spectrum demonstrates correlations and trade-offs between a range of chemical, physical and physiological leaf traits across biomes and functional groups, based on a leaf dry mass economics framework of carbon and nutrients nitrogen and phosphorus (Wright et al. 2004). In this spectrum, specific leaf area (SLA, leaf area divided by leaf dry mass), mass-based leaf nitrogen concentration and mass-based maximum photosynthetic rate are all positively correlated with each other (Reich, Walters & Ellsworth 1997; Wright et al. 2005). Leaf longevity is an important trait in this spectrum and is negatively correlated with specific leaf area, mass-based leaf nitrogen concentration and mass-based maximum photosynthetic rate (Wright et al. 2005). Leaf life span is also linked to herbivore defence with large amounts of plant defences with high initial constructions costs, such as lignin and tannins, associated with long-lived leaves, while smaller amounts of chemical compounds that can be reabsorbed, such as alkaloids, are generally found in shorter-lived leaves (Coley, Bryant & Chapin 1985; Coley 1988). Leaf life span is considered a trait that optimizes leaf carbon and nutrient use where longer-lived leaves typically have high construction costs but recover this cost through long photosynthetically active life spans (Kikuzawa 1995). If silicon is correlated with leaf longevity, then it is likely to be an active player in the leaf dry mass economic spectrum.
To identify whether leaf silicon is related to leaf longevity, and therefore a potential component of the leaf dry mass spectrum, the relationship between silicon concentration and leaf longevity can be compared across many species divided into functional and phylogenetic groups. Following O’Reagain & Mentis (1989), we predicted that silicon concentration is negatively correlated with leaf longevity, with higher concentrations of silicon in shorter-lived leaves. We tested this prediction across species, within plant groups and using evolutionary divergence (or phylogenetically independent contrast) analysis.
Materials and methods
Hodson et al. (2005) compiled a data base of the silicon concentration of leaves or non-woody shoots from 125 studies that had at least one species in common with another study. They used a residual maximum likelihood procedure to determine the relative silicon concentration of 735 species and provided this data set as a supplementary appendix (Hodson et al. 2005). Use of the relative silicon concentration rather than absolute is important as silicon uptake is affected by silicon availability and therefore can vary between studies. Hence, Hodson et al. (2005) used studies containing at least two species that shared at least one species in common with another study, allowing a relative silicon concentration to be determined as a treatment (species) mean.
We matched the species in the data set from Hodson et al. (2005) with leaf longevity data from the GLOPNET data set (Wright et al. 2004), resulting in paired leaf longevity and silicon concentration for 75 species. For the remaining species, a search was conducted in Google (‘Genus species’ AND ‘leaf longevity’ OR ‘leaf lifespan’ OR ‘leaf life span’ OR ‘leaf lifetime’) resulting in leaf longevity data for a further 80 species being obtained (Appendix S1 in Supporting Information). This resulted in paired leaf longevity (months) and relative silicon concentration for 155 species.
Online floras were used to allocate each species in the Hodson et al. (2005) data set to one of seven growth form categories: fern, grass, herb, rush or sedge, shrub, tree and vine. Tree, shrub and vine species were classified as evergreen or deciduous (including drought deciduous and semi-deciduous). Herb, rush and grass species were identified as annual or perennial. We then condensed foliage longevity (available for 672 species) to annual or perennial, where deciduous species were considered to have annual foliage and evergreen species to have perennial foliage, to allow comparisons across growth forms. This approach, though appropriate for accumulation of an element such as silicon that is not remobilized or reabsorbed, may not be appropriate for all leaf longevity comparisons.
To test for correlation between leaf longevity (months) and relative silicon concentration, the relationship between these two traits was described using a combination of Pearson’s correlation and standardized major axis (SMA) slope fitting across the 155 species and by group (Angiosperm, Gymnosperm and Polypodiophyta), order and leaf longevity type (annual/perennial/evergreen/deciduous). Small sample sizes did not allow analysis within family groups. Both variables were log transformed prior to analysis. SMA analyses were conducted using smatr (Warton, Wright & Falster 2006).
Relative silicon concentration values were log transformed, and then anova was used to determine whether silicon concentration varied with categorical leaf longevity by growth form and family for those comparisons containing more than 10 species and more than three species within each group of any comparable pair of leaf types. A two-tailed t-test was used to determine significant differences between annual and perennial foliage within each growth form. Vines, ferns and rushes were excluded from the analysis because of very small sample sizes. These analyses were carried out with spss Ver 15.0 (SPSS 2006).
A phylogeny was generated using Phylomatic (online, but see Webb & Donoghue 2005) by selecting the maximally resolved tree, which included 148 of the 155 species with continuous leaf longevity. The Poaceae and gymnosperms were further resolved following Barker et al. (2001) and The Tree of Life (http://tolweb.org/tree/, accessed 30 May 2009), respectively, and the Polypodiophyta were manually added following Smith et al. (2006), allowing the inclusion of all 155 species (82% resolution, with 252 of a potential 308 branches, see nexus file, Appendix S2 in Supporting Information). This process was repeated for the species for which binary leaf longevity data were available (annual/perennial), resulting in a phylogeny including 602 species (76% resolution with 925 of potential 1207 potential branches, see nexus file, Appendix S3 in Supporting Information). The Analysis of Traits (AOT) module of Phylocom (Webb, Ackerly & Kembel 2008) was then used to test whether divergences in relative leaf/shoot silicon content were correlated with divergences in leaf longevity throughout the phylogenetic tree. Pseudo branch lengths were not estimated, but were assigned a value of one as branch lengths are not required for this analysis of traits. Analyses were performed on log10-transformed data.
A significant negative correlation was found across all species between leaf longevity (months) and relative silicon concentration (r2 = 0·163, P < 0·001, n = 155), indicating that across growth forms and plant families, leaves with shorter life spans contain higher concentrations of silicon (Fig. 1a). This negative relationship was maintained within the angiosperms, but not within the gymnosperms and ferns, which had much smaller sample sizes (Table 1). Within the angiosperms, the correlation was maintained for all dicotyledons and annual dicotyledons (which included annual and deciduous species grouped) and deciduous dicotyledons (Table 1, Fig. 1a). Again, these were the groups with the larger sample sizes.
Table 1. Results of standardized major axis regression between relative silicon concentration and leaf longevity (months) across all species, within major phylogenetic groups, and within growth form and leaf longevity classes
Significant results are in bold.
Within order and family, no significant difference in silicon concentration was found between annual and perennial leaves except in the Poales and the Poaceae (Table 2), where annual leaves had higher silicon concentration.
Table 2. Summary of t-test results comparing relative silicon concentration between leaf longevity groups (annual vs. perennial), by orders and families with >10 species in total and >3 species within each leaf type
Total no. species
Significant results are in bold.
Across all growth forms, silicon concentration was found to be significantly higher in annual vs. perennial foliage (F2,732 = 9·67, P < 0·001, Fig. 2). T-tests indicated that there was a significant difference between annual and perennial foliage within grasses (t = 2·37, P = 0·024, n = 149), herbs (t = 3·04, P = 0·003, n = 235) and trees (t = 3·38, P < 0·001, n = 165), but not within shrubs (t = 1·45, P = 0·15, n = 65).
The evolutionary divergence analysis showed that divergences between log10 relative silicon concentration and log10 leaf life span (months) were not significantly correlated (n = 108, r = 0·134, ns, Fig. 1b). Relative leaf/shoot silicon concentration varied much more than leaf longevity across the phylogeny, and there was a stronger phylogenetic signal from silicon concentration (K = 0·17) than from leaf longevity (K = 0·06), suggesting that more closely related species are more likely to have similar leaf silicon concentrations than similar leaf longevity. However, both signal values were low, indicating a large amount of variation between closely related taxa in both traits. In contrast, divergences between log10 relative silicon concentration and leaf life span as a binary trait (annual vs. perennial) were significantly correlated (t = −2·44, P < 0·01, n = 83; P = 0·014, n = 83, sign test). This suggests that across a larger data set representing a larger phylogenetic tree, shifts to higher relative leaf/shoot silicon concentration are consistently associated with shifts to annual foliage.
The silicon concentration of leaves and non-woody shoots was found to be higher in shorter-lived leaves across all species and within most growth forms (Figs 1a and 2). The pattern across all species was driven predominantly by groupings of long-lived gymnosperm leaves with low silicon concentration and shorter-lived leaves with higher silicon concentrations. However, deciduous leaves of dicotyledons showed the strongest correlation between silicon concentration and leaf longevity (Table 1). The (negative) relationship that was found across growth forms, across deciduous dicotyledons and upheld in one family (Poaceae) in annual vs. perennial comparisons, supports the hypothesis that high silicon concentration occurs in shorter-lived leaves but that silicon is restricted in longer-lived leaves. This hypothesis was originally proposed by O’Reagain & Mentis (1989) for grasses: our data show that the hypothesis is more generally applicable across plant types. In the grasses, the significant difference in silica concentration between annual vs. perennial grasses (n = 148), but lack of relationship between silica concentration and leaf longevity as a continuous trait (n = 57), is probably an artefact of sample sizes.
We found that across evolutionary divergences, silicon concentration was not correlated with leaf longevity measured as a continuous variable (months, Fig. 1b), but was significantly negatively correlated with leaf longevity measured as a binary trait (annual vs. perennial) using a much larger data set (672 vs. 155 species). Together, this suggests that shifts to higher foliage silicon content have been associated with shifts to shorter leaf longevity throughout evolutionary history. This is suggested by the deciduous dicotyledon leaves from a range of families (Fig. 1a) that show a negative correlation between leaf silicon and leaf life span. Similarly, there is a significantly higher concentration of silicon in annual compared with perennial leaves within both trees and shrubs that encompass an even broader range of taxa (Fig. 2, 750 species).
Plants with shorter-lived leaves are generally thought to be capable of maximizing growth during optimum growing conditions, using leaf (or plant) senescence to avoid suboptimal periods (Chabot & Hicks 1982). Leaf longevity is negatively correlated with SLA, mass-based maximum photosynthetic rate (Amass), relative growth rate (RGR) and mass-based leaf nitrogen concentration (Nmass), such that shorter-lived leaves have high SLA, Amass, RGR and Nmass (Reich, Walters & Ellsworth 1992). In areas with lower soil nutrient availability, leaves have longer leaf life span, which is considered a strategy to increase nutrient use efficiency and carbon gain (Wright & Westoby 2002). At one end of the spectrum, species have short-lived leaves, high SLA, Amass, RGR and Nmass trait values. Our data suggest that the leaves of these species also have high levels of silicon, suggesting silicon could be an active component of the leaf dry mass economics framework.
We suggest that plant silicon may be a useful resource for shorter-lived leaves that contributes towards a more favourable leaf carbon strategy. The metabolic cost of silicon incorporation into plant tissue is likely to be much lower than for carbon. Raven (1983) calculated that 27 times less glucose was required to incorporate a mole of silicon into cell walls than a mole of carbon. Similarly, Villar & Merino (2001) found a negative correlation between leaf construction cost (g glucose g−1) and leaf ash concentration (g g−1), with silicon potentially constituting a major component of ash. At a lower metabolic cost, silicon may be considered a cheaper and therefore more disposable resource than carbon (O’Reagain & Mentis 1989).
The trade-off found here between silicon and leaf longevity does, however, provide evidence that there are costs involved in silica accumulation, although the costs of silicon for plants are not fully understood (Cooke & Leishman 2011). Once deposited in plants, silicon is not retranslocated and, unlike carbon, its capacity to form complex molecules in natural conditions is limited (Epstein 1999). Silicon in plants can bind to aluminium and other metals, but predominantly exists as amorphorous silica (Epstein 1999). Carbon, in contrast, forms a multitude of large, highly complex molecules. In addition, effective biochemical functions of silicon rely on a concurrent supply of silicic acid as plants that are only pre-treated with silicic acid before a stress is applied fail to respond in the same way as plants that have a concurrent supply of available silicon (Iwasaki et al. 2002) Silicon may therefore be considered a less versatile resource than carbon, which may limit its use to shorter-lived leaves (O’Reagain & Mentis 1989). In addition, silica is denser than lignin and cellulose (O’Reagain & Mentis 1989) and therefore incurs higher mass costs. Silicon concentration has been demonstrated to increase over time in the leaves or needles of Picea glauca (Hodson & Sangster 1998), Picea albies (Wyttenbach, Tobler & Bajo 1991) and Sasa veitchii (Motomura, Fujii & Suzuki 2004), so it is likely that plants with longer-lived leaves have developed a mechanism to exclude silicon to prevent excessive build-up. Some species are capable of excluding almost all silicon (Ma & Takahashi 2002a, Ma et al. 2002b).
Most research regards silicon as independent of the carbon cycle of plants, and although carbon and silicon are not coupled (unlike C and N for example; Claquin & Martin-Jezequel 2002), McNaughton et al. (1985) hypothesized that silicon deposition in plants may contribute to a more favourable carbon balance by substituting mineral-generated support for carbon-generated support. Increased silicon has been demonstrated to reduce lodging in Poaceae species such as rice (Hossain, Horiuchi & Miyagawa 1999) and wheat (Gartner, Charlot & Paris-Pireyre 1984), and clearly both plant silicon and carbon could contribute to leaf structural strength. Silicon in cell walls decreases extensibility and toughens walls of root stele cells and has also been shown to increase cell wall extensibility in the basal zone of isolated stele tissue in Sorghum bicolor (Hattori et al. 2003).
Silicic acid uptake, and hence the amount of silicon that can potentially be deposited as silica, is determined in part by silicon-specific transporters in the roots (Ma & Yamaji 2008). The mechanisms by which silicon concentration in leaf tissue is controlled are only partially understood (Richmond & Sussman 2003). Using a small subset of species for which we had data, we tested whether the between-species differences in leaf silicon could be attributed to leaf lifetime transpiration and found no relationship (data not shown).
Leaf silicon also plays a role in herbivore defence, with abrasive and indigestible silica reducing palatability and digestibility (McNaughton & Tarrants 1983, Massey & Hartley 2009; Reynolds, Keeping & Meyer 2009), and particularly the absorption of nitrogen (Massey & Hartley 2006b). It is possible that leaf silicon could be used as an alternative to carbon-based phenolics and other herbivore defences, allowing for a more favourable leaf carbon balance. Massey, Ennos & Hartley (2007a) found a strong negative relationship between growth rate and overall investment in defence, but there was but there was strong inter-specific variation in grass defence strategies. Coley, Bryant & Chapin (1985) predict that large amounts of plant defences with high initial constructions costs should occur in long-lived leaves, while smaller amounts of chemical compounds that can be reabsorbed will be found in shorter-lived leaves. Silica, with low construction costs, is predicted to be found in shorter-lived leaves (as found in this study), but differing from this hypothesis is not remobilized as is predicted for defence compounds in short-lived leaves (Coley, Bryant & Chapin 1985). Increased silicon accumulation can be induced by herbivore damage (Massey, Ennos & Hartley 2007b), with induction being an effective strategy that can reduce defence costs until herbivore damage occurs. This is a particularly useful strategy when herbivory is intermittent or predictable (Karban et al. 1999). Hence, the inducible response potential of silicon could allow further savings of carbon allocated towards defence by only allocating resource to defence when required.
If leaf silicon and leaf longevity are significantly correlated, it is possible that leaf silicon is also correlated with a suite of other leaf-level traits such as SLA, Amass and Nmass. We tested for these relationships and found that silicon concentration was poorly correlated with SLA, Amass and Nmass, though all three showed positive trends (data not shown). If a trade-off existed between silicon and carbon, then strong relationships between these traits would not necessarily be expected, as the interaction between carbon and silicon could be independent of these. A more likely correlation would be between leaf construction cost (g glucose g−1 leaf) and Si concentration. Leaf construction cost is frequently calculated following Williams et al. (1987) in which the ash fraction by mass (potentially dominated by silicon) is subtracted before carbon costs are determined, to which the costs associated with nitrogen are then added.
The two plant groups that had the strongest relationship between silicon concentration and leaf longevity were deciduous dicotyledons and grasses, although grasses and deciduous plants may have very different silicon uptake and allocation patterns. Recently, the first silicon transporter in dicotyledons has been identified in pumpkin, which differs to those identified in monocotyledons (rice, barley and maize) in transport substrates and cell-type specificity of localization (Mitani et al. 2011). Silicon is also deposited in these two plant groups in very different ways. In grasses, silica is deposited in comparatively large amorphous bodies, often filling whole cells or forming a layer under the cuticle (Ma 2003), while in contrast, deciduous plants have much smaller amounts of silicon deposition, with silicon often occurring as more fragile deposits in the cell walls (Wilding & Drees 1971). The functions of silicon in grass leaves have been examined in terms of herbivore defence, stress alleviation and structure (Brizuela, Detling & Cid 1986; Massey, Ennos & Hartley 2006a, Massey & Hartley 2006b, Liang et al. 2007, as examples). Comparatively little is known of the role of silicon in deciduous leaves, although the uptake and cycling rates of silicon were found to be considerably higher in a temperate deciduous forest than a temperate coniferous forest (Bartoli 1983). Further research into the respective allocation and functions of silicon in deciduous leaves and grasses may be a useful way to better understand the incorporation of silicon into the dry mass spectrum.
The following working hypotheses summarize possible ways in which silicon may contribute to a more favourable carbon strategy in leaves with shorter life spans, assuming the lack of versatility makes a build-up of silicon in long-lived leaves undesirable:
1 Silicon contributes to cell strength and tissue support in place of lignin or cellulose (Raven 1983).
3 Silicon alleviates a range of abiotic stresses (including water stress, nutrient and heavy metal toxicity and salinity stress), allowing shorter-lived leaves to reach their potential maximum photosynthetic rates during narrow windows of growth opportunities.
If silicon is substituted for carbon in any of these functions, then more of the carbon assimilated could potentially be directed to new leaf production, thus increasing growth rates and the rate of return for invested carbon. The hypotheses are not mutually exclusive; for example, Hunt et al. (2008) have suggested that increase in cell strength can contribute to the anti-herbivore functions of silica by preventing cell crushing by herbivores. In addition, the roles of silicon in defence against herbivores and pathogens are complex and are not yet fully understood; hence, the list of processes above in which silicon might substitute for carbon in defence is likely to be conservative.
To date, research on plant silicon has been dominated by investigations into the alleviation of plant stresses and predominantly relates to agricultural systems where silicon can be added artificially (Cooke & Leishman 2011). Silicon is remarkable in that it is involved in a broad range of unrelated functions (Epstein 2001) and given the broad range of benefits demonstrated to be induced or resulting from silicon addition and its presence in concentrations exceeding that of many other macronutrients, it follows that silicon is likely to have important functions in natural ecosystems. The recognition of the correlation of silicon content with leaf life span, whereby silicon potentially allows shorter-lived leaves to achieve a more favourable leaf carbon strategy, is a step towards considering silicon as an important element in natural environments and to incorporate silicon concentration into our understanding of plant ecological strategies.
The authors thank Ian Wright and Peter Reich for use of the Glopnet data set, Ian and Rachael Gallagher for helpful discussions and comments on the manuscript, three anonymous reviewers for their helpful comments, Amy Zanne for help with phylogenetic analyses, and Hans Cornelissen for providing additional data.