Characterizing sclerophylly: the mechanical properties of a diverse range of leaf types


Author for correspondence: Jennifer Read Tel: +61 3 9905 5622 Fax: +61 3 9905 5613 Email:


  • • Although sclerophylly is defined by textural properties, its adaptive significance has been debated without a strong base of mechanical data. We measured a wide range of mechanical properties across a diverse range of species and leaf forms, including highly scleromorphic leaves, and compared these with sclerophylly indices to determine the mechanical properties of sclerophylls.
  • • Fracture and flexure tests were used to determine leaf strength, toughness (work to fracture) and flexural stiffness (‘structural’ properties), and specific strength, specific toughness and Young's modulus of elasticity (‘material’ properties, i.e. normalized per unit leaf thickness).
  • • Leaves varied considerably in all properties tested, and in the way they combined various ‘structural’ and ‘material’ properties. However, on average, highly scleromorphic leaves were stronger, tougher and stiffer than soft leaves. ‘Structural’ properties correlated more strongly with sclerophylly than ‘material’ properties, and the ratio of stiffness to strength and toughness increased in sclerophyllous species.
  • • Of the structural properties, strength, toughness and flexural stiffness each made substantial independent contributions to the variation in sclerophylly indices, but the best individual explanators were flexural stiffness and strength, with the best predictive model being a combination of these two properties. This model should now be tested on leaves from contrasting environments.


Leaf texture can vary considerably among species, and can be characteristic of vegetation types and vegetation strata. For example, heath may have ‘harder’ leaves than rainforest, and overstorey plants may have ‘harder’ leaves than understorey plants. ‘Sclerophylly’ means ‘hard-leaved’, though biologists also use terms such as ‘tough’, ‘stiff’ and ‘leathery’ to describe such leaves. The term was coined by Schimper (1903) to distinguish xeromorphic plants with leathery leaves from those exhibiting succulence or leaflessness. However, sclerophylly can be found in a range of environments and its functional significance is uncertain. Hypothesized roles include resistance to water deficits (Schimper, 1903; Oertli et al., 1990), resistance to, or a consequence of low-nutrient soils (Loveless, 1961; Beadle, 1966), enhancement of leaf longevity by leaf protection (Chabot & Hicks, 1982; Turner, 1994a) and as a nonspecific evolved response to, or consequence of, a wide range of environmental stresses (Salleo & Nardini, 2000). Leaf structure, including biomechanical properties such as toughness, may influence herbivory (Coley, 1983; Choong, 1996; Gutschick, 1999) and the community composition of herbivores, including their diversity and density (Peeters et al., 2000; Peeters, 2002), and potentially affects other trophic levels within the community (Press, 1999). Hence, leaf biomechanical properties may play a substantial role in structuring the organization of ecosystems.

Despite the apparent ecological significance of sclerophylly, there is still relatively little known about the mechanical properties that characterize scleromorphic leaves. Much of what is known about leaf mechanics is based on petioles rather than leaf laminae, because the latter are difficult to model analytically (Niklas, 1999). A better understanding of the mechanical components of leaf texture should assist in teasing apart their function, and hence improve our understanding of questions related to the adaptive significance of sclerophylly (Choong et al., 1992; Edwards et al., 2000). In particular, does sclerophylly (i.e. having hard, tough, leathery leaves) have a single function (and what is it?) or does it confer benefits across a range of environmental stresses? Is this function achieved by a single mechanical property or by a group of mechanical properties, each of which may have independent or interacting effects? Furthermore, is sclerophylly even of adaptive significance per se, or is it no more than the consequence of adaptive or nonadaptive anatomical and physiological features (e.g. Salleo & Nardini, 2000)?

In Fig. 1 we show how leaves may vary in three properties that may be related to sclerophylly: strength, toughness and stiffness. Several studies have examined the mechanical properties of a wide range of species in relation to sclerophylly or related issues (Choong et al., 1992; Turner et al., 1993; Edwards et al., 2000; Wright & Cannon, 2001; Wright & Westoby, 2002). These authors found that scleromorphic leaves (sclerophylls), as judged by leaf mass area (syn. specific leaf weight, specific leaf mass: leaf mass per unit area), or the Loveless sclerophylly index (percentage fibre per unit protein; Loveless, 1961), or a subjective categorization/index of sclerophylly based on leaf texture (Edwards et al., 2000), had high toughness (work to fracture) (Edwards et al., 2000; Wright & Westoby, 2002), specific toughness (toughness expressed per unit leaf thickness) (Choong et al., 1992; Turner et al., 1993; Edwards et al., 2000; Wright & Cannon, 2001) and high strength and/or specific strength (strength per unit leaf thickness) (Choong et al., 1992[measured as resistance to a penetrometer]; Edwards et al., 2000). In addition, Reich et al. (1991) showed that leaves with long lifespans, and low specific leaf area (inverse of leaf mass area) have high force to fracture (termed ‘toughness’ in that paper). These conclusions were drawn from fracture tests of tearing (Edwards et al., 2000), punching (Reich et al., 1991; Choong et al., 1992; Turner et al., 1993; Edwards et al., 2000) and shearing (Choong et al., 1992; Turner et al., 1993; Edwards et al., 2000; Wright & Cannon, 2001). All these studies included both soft and scleromorphic leaves (to varying degrees), but only Edwards et al. (2000) investigated a broad range of test types, and none included flexural tests. Choong et al. (1992) viewed toughness (named specific toughness by us to differentiate it from whole-leaf toughness which is influenced by leaf thickness, see Table 1) as the core attribute of a sclerophyll. However, in describing sclerophylls, botanists have used the terms ‘hard’, ‘tough’, ‘stiff’ and ‘leathery’ (Schimper, 1903; Grubb, 1986). These terms suggest that a range of properties may be important in producing the scleromorphic leaf, though each of these terms has been used in a general sense, without strict definition that allows measurement. Botanists have most often used indirect indices such as leaf mass area or the Loveless index to measure sclerophylly, rather than textural properties. We have argued that measurement of textural properties, as inherent in the term ‘sclerophylly’, will be more useful than indirect indices, and that defined and quantifiable properties as used by materials engineers will be most useful (Edwards et al., 2000). However, rather than assume that only toughness has any functional significance in sclerophylly, we address a broad range of properties that relate to terms used by botanists over the last 100 years.

Figure 1.

Representations of force-displacement curves of four leaves with contrasting mechanical properties. Strength is defined as the maximum force to fracture per unit area over which the force is applied. Toughness is defined as the work to fracture, measured as the area under the force-displacement curve. Stiffness is related to the initial slope of the curve, or stress per unit strain. In this diagram, Plant A has the same toughness as Plant B, but is stiffer and stronger. Plant C is stiffer than Plants A and B, stronger than Plant B, but is not as tough as Plants A and B. Plant D is stiffer, stronger and tougher than all other plants. This diagram is simplified and does not show any alterations of slope associated with changes from elastic to plastic deformation.

Table 1.  Measured and derived mechanical properties used in this study (modified after Jackson, 1992; Edwards et al., 2000; Sanson et al., 2001)
Mechanical propertiesCalculationProperty being measured
  1. # More accurately measured by using the cross-sectional area of the cut/torn edge (C) instead of dividing by W and by T. A, area of punch (m2); C, cross-sectional area of tear or shear (m2); D, displacement of moving head of test machine (m); F, force (N); S, span length between supports in a bending test (m); T, thickness of leaf at position of test (m); W, width of leaf in plane of shear, tear or bend (m).

Punching tests
Punch strengthFmax/A (N m−2)Absolute punch strength of the whole leaf at the point of testing (influenced by leaf thickness).
Specific punch strength(Fmax/A)/T (N m−2 m−1)Punch strength per unit leaf thickness at the point of testing.
Work to punch(F/A) × D (J m−2)The absolute amount of work done to force the punch through the leaf (influenced by leaf thickness)
Specific work to punch((F/A) × D)/T (J m−2 m−1)Work to punch the leaf per unit leaf thickness.
Shearing tests
Work to shear(F × D)/W (J m−1)The absolute amount of work done to shear the leaf per unit leaf width (influenced by leaf thickness).
Specific work to shear((F × D)/W)/T# (J m−2)Work to shear the leaf per unit leaf thickness.
Tearing tests
Force to tearFmax/WThe force to tear the leaf per unit leaf width
(N m−1)(influenced by leaf thickness).
Tensile strength(Fmax/W)/T# (N m−2)Strength to tear the leaf.
Work to tear(F × D)/W (J m−1)The absolute amount of work done to tear the leaf per unit leaf width (influenced by leaf thickness).
Specific work to tear((F × D)/W)/T (J m−2)Work to tear the leaf per unit leaf thickness.
Bending tests
Young modulus (E)(F/D) × S3/(48 I) (N m−2)Stiffness or elastic resistance to bending, corrected for effects of the dimensions of leaf material (where the second moment of area, I, = W T3/12).
Flexural stiffness (EIW)(F/D) × S3/(48 W) (N m2 m−1)Stiffness or elastic resistance to bending, corrected only for strip width.

In this paper we describe properties from both fracture and flexure tests of ‘dicot’ species (eudicots and magnoliids) that comprise a wide range of leaf textures including highly scleromorphic leaves. In a previous study we investigated native Australian plants growing in close proximity (Edwards et al., 2000). Correlations among leaf characters may therefore have been strongly influenced by a common evolutionary history (both environment and phylogeny) of the study plants. In the present study we use 32 species, each from a different family to reduce phylogenetic bias, from across the world but grown in the same location (Royal Botanic Gardens, Melbourne, Australia). The major aim was to determine the mechanical properties of sclerophylls, including the best explanators of sclerophylly. That is, are sclerophylls well-characterized by a single mechanical property, or group of properties, or are there different ways of being perceived as sclerophyllous, i.e. can different combinations of mechanical properties produce sclerophylls? Mechanical properties were tested against one direct and two indirect indices of sclerophylly, and the associations between sclerophylly and leaf nitrogen, phosphorus and fibre concentrations were also measured.


Leaves were collected in October–November 2000 from the Royal Botanic Gardens, Melbourne, Australia (37.8° S, 145.0° E). The climate of Melbourne is relatively mild, with a mean daily maximum and minimum temperature of 19.8°C and 10.1°C, respectively, and annual rainfall of c. 660 mm (Commonwealth Bureau of Meteorology). The rainfall is more uniformly distributed through the year in Melbourne than in much of southern Australia.

Sampling procedure

Thirty-two dicot families containing evergreen species were selected randomly (by random number tables) from those represented in the Gardens. Within each family, the list of species was reduced to those with at least five plants present. A single species was chosen randomly (using random number tables) from each family (Table 2). Although the plants are native to a variety of regions, Australia is strongly represented (Table 2). Banksia marginata was added for comparison because it was among the toughest and strongest species recorded by Edwards et al. (2000), but was not used in any statistical analyses. It was not growing at the Gardens and was instead collected from the grounds of Monash University (16 km south-east).

Table 2.  The study species and their habit and region of origin
SpeciesFamilyHabitNative distribution
  1. H, herb; l, liane; s, shrub, or shrub-like; ss, subshrub; sc, semiclimber; T, tree; t, small tree. Names and authorities were taken from the International Organization for Plant Information, where possible. 1Phyllodinous leaves.

Acacia melanoxylon R.Br.1MimosaceaeTAustralia
Acmena smithii (Poiret) Merr. & PerryMyrtaceaeTAustralia
Agapetes meiniana F.Muell.Ericaceaes/scAustralia
Ajuga reptans L.LamiaceaehEurope
Anopterus glandulosus Labill.GrossulariaceaesAustralia
Austrobaileya scandens C. WhiteAustrobaileyaceaelAustralia
Banksia marginata Cav.Proteaceaet/sAustralia
Banksia serrata L.f.Proteaceaet/sAustralia
Begonia luxurians Scheidw.BegoniaceaesSouth America
Bergenia crassifolia (L.) FritschSaxifragaceaehSiberia, Mongolia
Buddleja davidii Franch.Buddlejaceaet/sChina
Choisya ternata Humb., Bonpl. & KunthRutaceaesMexico
Diospyros australis (R.Br.) HiernEbenaceaeTAustralia
Echium candicans L.f.BoraginaceaesMadeira
Goodenia ovata SmithGoodeniaceaesAustralia
Guilfoylia monostylis (Benth.) F.Muell.SurianaceaeTAustralia
Heritiera trifoliolata (F.Muell.) Kosterm.SterculiaceaeTAustralia
Hernandia bivalvis Benth.HernandiaceaeTAustralia
Ilex aquifolium L.AquifoliaceaeTEurope & Mediterranean
Justicia adhatoda L.AcanthaceaesIndia, Sri Lanka
Laurus nobilis L.LauraceaeTMediterranean
Leucopogon parviflorus (Andrews) Lindl.EpacridaceaesAustralia
Ligustrum japonicum Thunb.OleaceaesKorea, Japan
Luculia gratissima (Wallich) SweetRubiaceaet/sHimalayas
Magnolia grandiflora L.MagnoliaceaeTSE United States
Mahonia lomariifolia Tak.BerberidaceaesBurma, western China
Melianthus major L.MelianthaceaesSouth Africa
Nothofagus moorei (F.Muell.) KrasserNothofagaceaeTAustralia
Pancheria elegans Brongn.CunoniaceaesNew Caledonia
Pelargonium tomentosum Jacq.GeraniaceaessSouth Africa
Quercus coccifera L.Fagaceaes/TMediterranean
Tasmannia lanceolata (Poiret) A.C. SmithWinteraceaesAustralia
Viola hederacea Labill.ViolaceaehAustralia

Where possible, five plants of each species were chosen haphazardly from locations across the Gardens, but in a small number of species a lesser number of plants was sampled, due to prior death. Leaves were collected from sunlit branches and within 1–2 m of ground level where possible. However, the growth form of the plants ranged from herbs to tall trees (Table 2), and plants were located in various parts of the Gardens, and hence were undoubtedly exposed to some variation in microclimate and soil properties. Only leaves that expanded during the previous growing season (therefore 6- to 12-months-old) were collected.

Leaves were investigated over an 8-week period. Leaves from one plant of each of c. 10 species were harvested per collection batch, and mechanical and morphological measurements were undertaken over a 48-h period. Species were haphazardly allocated to a collection batch, and batches were harvested twice per week until there were five replicates for each species. Batches were systematically alternated across the 8-week period to minimize any potential seasonal effect. Collected leaves were sprayed with water and sealed in a plastic bag with moist tissue in an insulated container. From each collection, leaves were haphazardly taken for measurement of biomechanics, leaf morphology, and for later anatomical study.

Indices of sclerophylly and leaf morphology

Midway through the test period, leaves were collected from each replicate plant of each species to obtain sclerophylly indices, and for chemical analyses. Eleven botanists independently ranked the test species in order of increasing sclerophylly (by feel), to provide a direct assessment of sclerophylly, i.e. an index based on leaf texture (Edwards et al., 2000). The leaves were stored overnight in moist tissue in plastic bags and the following day one leaf of all species was presented to each botanist (leaves from replicate plants were randomly assigned to botanists) to rank in order of sclerophylly. No direction was given to the botanists about the judgements they should employ, other than that they should ignore leaf size. However, they were asked to record the criteria used to make judgements. The average rank sums were used as an index of sclerophylly (BSI: Botanists’ sclerophylly index) to examine the relationship between mechanical properties and sclerophylly as judged by botanists.

The Loveless sclerophylly index (SI: 100× crude fibre d. wt/crude protein d. wt) was originally devised as an estimate of the ratio of cell wall to cell content (Loveless, 1961, 1962). In this study, cell wall was determined as neutral detergent fibre (NDF) rather than crude fibre. The unequal recovery of lignin, cellulose and hemicellulose in crude fibre leads to a variable relationship between crude fibre and plant cell wall, and it is difficult to predict true fibre content from crude fibre (Van Soest, 1994). Hence, values of SI in our study are not exactly comparable with those of Loveless (1961, 1962). NDF was measured following the method of Van Soest (1994). Foliar nitrogen was measured using a Leco® CHN-2000 analyzer (LECO Corporation, St Joseph, MI, USA). SI was calculated as the ratio of NDF d. wt to protein d. wt (N × 6.25). It was also recalculated as the ratio of NDF d. wt to protein d. wt as a fraction of nonfibre d. wt (NDF/(6.25 (N/(1 − NDF))) to provide an estimate of fibre per unit protoplasmic protein (SIadjusted). Foliar phosphorus was measured using the molybdenum-blue colorimetric method (Grimshaw et al., 1989) following digestion by the sulfuric-peroxide procedure (Grimshaw, 1987) to allow testing of the association between sclerophylly and phosphorus concentration. Leaf mass area (LMA) was measured on one leaf (petiole removed) per replicate plant, chosen haphazardly from the leaves harvested for mechanical analyses. Leaf area was measured by image analysis (Bioscan™ Image Analyser, Monash University, Victoria, Australia) and leaves were then dried to constant weight at 40°C and weighed. Succulence (weight of water per unit leaf area: Cowling & Campbell, 1983) was measured by recording the f. wt of each leaf (assumed to be turgid) before measuring leaf area.

Leaf biomechanical properties

Tests were undertaken within 48 h of collection on haphazardly chosen leaves from each replicate plant. A Universal Testing Machine (Chatillon Universal Tension and Compression Tester, model UTSE-2, AMETEK Inc., Paoli, PA, USA) was modified to produce a data output of 800 points per second of the instantaneous force and displacement to a personal computer. Fracture tests (punching, tearing and shearing) and bending tests were undertaken and the mechanical properties (Table 1) were derived from force-displacement curves using Leaf2k ver. 3.7 (M. Logan, Monash University). We used work to fracture as a measure of toughness, where fracture toughness is defined as the resistance of a given material to the propagation of a crack (Wainwright et al., 1976). Work to fracture, fundamentally an energetic process (Atkins & Mai, 1985), can be estimated from the area under the force-displacement curve, provided the curve returns to zero stress in a series of relatively controlled steps, rather than in a single massive drop (Vincent, 1992). We use work to punch and work to shear as estimates of leaf toughness, even though the former may overestimate the true toughness of the specimen (Sanson et al., 2001). The concept of strength is complex: strength has been considered variously as the yield stress (the change from elastic to plastic flow in a stressed material), the stress at fracture, or the maximum resistance to an applied force (Wainwright et al., 1976). We have been unable to consistently detect the yield stress or stress at fracture, and therefore measured strength in this study as the maximum force recorded divided by the area over which the force was applied. In materials engineering, where the usual interest is in the property of the material in test situations, it is conventional to normalize strength and toughness by dividing by appropriate dimensions (including thickness) of the test piece. Such properties may be seen as ‘material’ properties. However, properties normalized to thickness, ignore the contribution of thickness to the mechanical properties of the leaf. Consequently we present both ‘structural’ properties (normalized to width of the test piece, but not thickness) and ‘material’ properties (i.e. normalized to both width and thickness and which we term ‘specific’, following the term ‘specific work of fracture’ used by Atkins & Mai, 1985), where ‘material’ (normalized) properties reflect an average of a very heterogeneous structure rather than material properties as more often measured by engineers in isotropic materials. Further aspects of these tests have been discussed by Aranwela et al. (1999) and Sanson et al. (2001).

Punch tests

In this test a hole is punched through the leaf lamina, with the resistance to penetration suggested to be a combination of shear and compressive strength and resistance to crack propagation (Vincent, 1992). A die was mounted onto the moving head of the test machine so that it engaged a hardened steel, flat-ended, sharp-edged cylindrical punch of 0.5-mm diameter with a clearance of 0.05 mm and a displacement speed of 0.3 mm s−1. Holes were punched at 10 random positions across the leaf, noting the tissue-type punched. Extra holes were punched as necessary to achieve five randomly positioned punches of each of intercostal lamina (lamina between secondary veins), secondary veins and midrib. For Leucopogon parviflorus, with closely spaced secondary veins, the ‘intercostal’ punches would probably include secondary veins for at least some punches. The punch was wider than many secondary veins across species and in those cases underestimates the mechanical properties of the veins. Lamina thickness was measured using a digital micrometer. Leaf strength and toughness were derived from the force-displacement curve (Table 1).

Shearing tests

Two K100 knife-steel guillotine blades, hardened and ground, were mounted onto the Universal Testing Machine. The cutting edge was horizontal in the lower blade and 20° in the upper blade, providing a constant approach angle of 20°. The bottom horizontal blade moved at a displacement speed set at 0.3 mm s−1, shearing the test leaf into two parts. The upper blade was raked at 45° and the bottom blade was not raked but was inclined to provide a relief angle of 4°. A blank run of the blades was undertaken before every test to record background friction and was subtracted from the test force-displacement curve. A transverse cut was made across the leaf, at a random distance along the leaf (distance selected by random number tables) ignoring the first and last 10% of the leaf (base and tip). The width and surface area of the leaf at the shearing plane was measured by image analysis, allowing work to be expressed per unit leaf width and per cross-sectional area (Table 1). Calculations of mechanical properties were made both over the full width of the leaf, and from one side of the leaf, excluding the midrib and the margin.

Tearing tests

A longitudinal strip of lamina 41 × 4.5 mm was cut from the middle of the left-hand-side of each leaf, where large enough, such that the length was greater than eight times the width to counter the effects of necking (Vincent, 1990). Test strips were secured in the force-tester by gluing each end of the strip with cyanoacrylate glue into the slot of a brass cheesehead screw. The strips were notched on the left-hand side (0.5 mm length) to direct the position of fracture so the test strip did not break at the secured ends. Any effect of notch sensitivity (Vincent, 1990; Lucas et al., 1991) was minimized by standardizing the relative notch length. The fracture length was measured by calipers and lamina thickness was measured as for the punch test, to estimate the cross-sectional area of the fracture surface, corrected for the area of the initial notch. It was impossible to obtain the recommended aspect ratio from lamina tissue alone in two small- and narrow-leaved species (Leucopogon parviflorus and Pancheria elegans) and therefore the test strips of these species comprised the full width of the leaf.

Bending tests

The flexural Young modulus (E) and flexural stiffness (EI, where I is the second moment of area) were measured in a three-point bending test (Jackson, 1992) (Table 1). The tests were undertaken on lamina strips that excluded the midrib and leaf margins, and also on whole leaves. A 5-mm wide strip of leaf was cut from one side of the leaf, parallel to the midrib. A pilot study was first undertaken to estimate the optimal span to thickness ratio (S/T) to minimize significant shear deformation, following Jackson (1992). In retrospect, four-point bending would have been preferable, since shear deformation is excluded (Jackson, 1992). A ratio of 50–70 times the strip thickness appeared suitable for most species. A ratio of 50 was the maximum possible for Ligustrum japonicum and Agapetes meiniana, and it is not certain whether E was maximum and constant at that ratio in these species. Therefore, it is possible that E and EI are underestimated in these two species.

The test protocol followed Jackson (1992), except that machine stiffness was not measured (we assumed that the forces encountered were negligible compared with machine stiffness). Thickness was measured in the centre of the test strip using a digital micrometer and the optimal span was calculated. The test strip was placed centrally over the two lower supports that were positioned to create the optimal span, standardized at 70× strip thickness where possible. The cross-head was positioned so that the upper support was almost touching the test strip. The cross-head was then activated to push the upper support onto the test strip at 0.3 mm s−1 until a sufficiently long trace was obtained. The initial slope of the force displacement curve (apparent stiffness, F/Dapp) was measured, ignoring any toe region. For a small number of narrow-leaved species, the strip width was less than 5 mm, and so EI was expressed per unit strip width (EIW) for all species.

Statistical analyses

The Botanists’ ranks of sclerophylly were analysed using Friedman two-way anova to test the hypothesis that there is no significant difference among species in sclerophylly ranks (Friedman's Test). Agreement among botanists in their rankings was measured by Kendall's Coefficient of Concordance (ranges from 0 to 1), and reflects both agreement by botanists in their concept of sclerophylly and variation in sclerophylly among replicate leaves.

anova was used to test for differences among species in each mechanical property. Pearson correlation analysis was used to investigate associations among mechanical properties, with Spearman correlation used to test associations with the sclerophylly indices, due to BSI being derived from ranks. No adjustment was made for multiple comparisons, following Stewart-Oaten (1995). For parametric statistics, assumptions were checked and loge transformations used to improve normality and reduce heteroscedasticity of variances where necessary. Where outliers remained, data were re-analysed without outliers to see if conclusions were altered. In each case the conclusion was not altered and outliers were retained. Principal components analysis (PCA) was used to reduce a set of 12 mechanical variables to major components to summarize variation among species. ‘Average’ punch values were excluded to avoid over-weighting of punch tests; the intercostal lamina values were retained since this was the only test that exclusively measured intercostal lamina properties. Given the high degree of correlation among mechanical properties, hierarchical partitioning (Chevan & Sutherland, 1991; MacNally, 2000; Quinn & Keough, 2002) was used to assess the independent contribution of each of three major ‘structural’ mechanical properties (strength, toughness and stiffness) to these indices. The Schwarz Bayesian Information Criterion (BIC) was used to determine the combination of these properties that provides the best fit to the sclerophylly indices (Quinn & Keough, 2002). Multiple regression was then used to determine the proportion of variation in sclerophylly explained by these combined properties. A critical value of α= 0.05 was used in hypothesis testing. All analyses other than hierarchical partitioning were undertaken using systat™ v. 10.


Sclerophylly indices

Species differed significantly in sclerophylly ranks (Friedman's test statistic = 297.3, P < 0.001). Of the 33 species ranked by botanists, Justicia adhatoda was ranked least sclerophyllous and the Banksia species most sclerophyllous, with Laurus nobilis, a Mediterranean species commonly regarded as sclerophyllous, occupying the median position with Acmena smithii (Table 3). Kendall's Coefficient of Concordance was high (0.87) reflecting little variation in sclerophylly among replicate leaves, and/or high agreement among botanists in their concept of sclerophylly. For example, the range of ranks given to J. adhatoda was 1–5, and to B. serrata was 28–33. However, the ranks of a few species were quite varied, the extreme being 15–33 in Agapetes meiniana.

Table 3.  Three indices of sclerophylly: BSI, the Botanists’ Sclerophylly Index (average rank); LMA (leaf mass area); SI, the Loveless sclerophylly index (NDF per unit protein)
SpeciesBSILMA (g m−2)SI (g g−1)Succulence (g m−2)
  1. An index of succulence is included, mass of water per unit leaf area. Values given are means ± se. The BSI values in brackets were altered after removal of B. marginata (see Methods section).

Acacia melanoxylon21.4 ± 1.5117 ± 53.43 ± 0.08168 ± 2
Acmena smithii19.0 ± 1.1122 ± 85.35 ± 0.72161 ± 5
Agapetes meiniana26.7 ± 1.9176 ± 125.28 ± 0.98603 ± 63
Ajuga reptans8.8 ± 0.5 45 ± 41.45 ± 0.15254 ± 12
Anopterus glandulosus19.5 ± 1.1107 ± 73.73 ± 0.25268 ± 13
Austrobaileya scandens16.5 ± 1.1106 ± 74.05 ± 0.32327 ± 20
Banksia marginata31.5 ± 0.6185 ± 178.20 ± 0.77197 ± 10
Banksia serrata32.0 ± 0.4
(31.5 ± 0.4)
190 ± 99.55 ± 0.51217 ± 9
Begonia luxurians10.8 ± 0.5 46 ± 51.65 ± 0.29310 ± 14
Bergenia crassifolia8.5 ± 1.0 94 ± 63.06 ± 0.42401 ± 30
Buddleja davidii4.3 ± 0.8 38 ± 71.57 ± 0.17173 ± 9
Choisya ternata12.5 ± 0.9104 ± 102.54 + 0.23189 ± 15
Diospyros australis23.9 ± 1.0162 ± 133.45 ± 0.05195 ± 6
Echium candicans7.9 ± 0.7 73 ± 61.93 ± 0.12354 ± 25
Goodenia ovata8.5 ± 1.0 46 ± 41.72 ± 0.08184 ± 12
Guilfoylia monostylis14.0 ± 1.1 86 ± 32.25 ± 0.06249 ± 22
Heritiera trifoliolata19.7 ± 1.2123 ± 285.51 ± 0.45116 ± 8
Hernandia bivalvis14.2 ± 0.9 82 ± 72.12 ± 0.16187 ± 11
Ilex aquifolium28.4 ± 1.1
(28.1 ± 1.0)
179 ± 195.36 ± 0.74237 ± 13
Justicia adhatoda2.9 ± 0.4 83 ± 132.88 ± 0.48179 ± 6
Laurus nobilis17.7 ± 1.3101 ± 95.15 ± 0.45123 ± 4
Leucopogon parviflorus22.1 ± 1.7107 ± 137.24 ± 0.99135 ± 13
Ligustrum japonicum23.5 ± 1.6168 ± 73.71 ± 0.06371 ± 20
Luculia gratissima4.5 ± 0.7 46 ± 71.54 ± 0.18169 ± 15
Magnolia grandiflora27.9 ± 1.3
(27.7 ± 1.2)
226 ± 228.30 ± 0.28266 ± 22
Mahonia lomariifolia27.5 ± 0.8
(27.4 ± 0.7)
139 ± 115.25 ± 0.34228 ± 18
Melianthus major4.7 ± 1.0 60 ± 51.30 ± 0.04208 ± 11
Nothofagus moorei22.5 ± 0.9130 ± 66.61 ± 0.21153 ± 7
Pancheria elegans23.5 ± 1.4140 ± 154.09 ± 0.12211 ± 8
Pelargonium tomentosum3.6 ± 0.8 63 ± 22.69 ± 0.06262 ± 6
Quercus coccifera28.2 ± 0.7
(28.1 ± 0.7)
Tasmannia lanceolata20.3 ± 0.8107 ± 52.96 ± 0.20270 ± 19
Viola hederacea3.9 ± 0.7 29 ± 22.13 ± 0.18204 ± 9

LMA varied c. 8-fold among species, from 29 g m−2 in Viola odorata to 226 g m−2 in Magnolia grandiflora (Table 3). SI varied c. 7-fold from 1.30 g g−1 in Melianthus major to 9.55 g g−1 in B. serrata. SI was highly correlated with SIadjusted (Rp = 0.97, P < 0.001) so SIadjusted was not used in further analyses. BSI was highly correlated with both LMA and SI, most strongly with LMA (Fig. 2). Succulence varied 5-fold from 116 g m−2 in Heritiera trifoliolata to 603 g m−2 in A. meiniana. There was no correlation between succulence and any sclerophylly index (P > 0.05), and therefore succulence was not reviewed any further. B. serrata, one of the most sclerophyllous species as judged by LMA and SI, recorded similar values in each index to B. marginata, one of the most sclerophyllous species (as judged by these indices) in our previous study (Edwards et al., 2000).

Figure 2.

Relationships between BSI and each of LMA and SI. The data points are species’ means.


Punch tests There were significant differences among species for each of the punch tests, with 9- to 18-fold variation for tests of intercostal lamina and average punches across the lamina (Table 4). Highest values of the intercostal punch variables were usually obtained in one or both of the Banksia species, the exception being specific strength which was highest in Acacia melanoxylon, H. trifoliolata and Quercus coccifera (Table 4). For average punches, high strength was recorded in Q. coccifera (only one plant sampled) and A. melanoxylon, and high specific strength in Q. coccifera, A. melanoxylon, L. nobilis and H. trifoliolata (Table 4). Work was highest in Austrobaileya scandens, Magnolia grandiflora and B. marginata with high specific work in Q. coccifera, B. marginata, A. melanoxylon, L. nobilis, A. scandens and B. serrata. Each of the punch variables correlated significantly with BSI and the other sclerophylly indices, the correlations generally strongest in the variables unadjusted for leaf thickness (Table 5).

Table 4.  The mechanical properties of each species
SpeciesIntercostal punch testsAverage punch testsShearing testsTearing testsBending tests
  1. The values given are means with SE in brackets. The results of anova are also given for each property (data were loge-transformed). Punch tests: SP, strength (MN m−2); SSP, specific strength (GN m−2 m−1); WP, work (kJ m−2); SWP, specific work (MJ m−2 m−1); Shearing tests: WS, work (J m−1); SWS, specific work (kJ m−2); Tearing tests: FT, force per unit width (N m−1), ST, strength (MN m−2); WT, work (J m−1); SWT (kJ m−2); Bending tests: E (MN m−2); EIW(mN m2 m−1).1No strips were successfully torn in Acacia melanoxylon and Quercus coccifera: the values given are minimum values, i.e. derived from the force at which the strip pulled from the screws.

Acacia melanoxylon13.4154.63.1212.7219.3676.44.1816.490.6891.95136616.2810.74813.4418111.12
Acmena smithii7.5326.81.585.6312.1740.42.468.150.1720.584221.490.2020.712660.68
Agapetes meiniana5.247.12.823.829.2610.84.925.740.4450.495600.970.4220.73833.40
Ajuga reptans2.9311.00.572.143.5913.00.973.530.0560.171650.910.2531.401010.21
Anopterus glandulosus7.1418.41.403.627.5218.41.904.650.2270.555481.620.5401.591340.69
Austrobaileya scandens9.6626.22.947.9817.8747.75.8915.710.3680.953131.040.3401.131740.81
Banksia marginata12.5442.54.2114.2413.4043.25.3017.100.8212.008783.800.5571.924001.07
Banksia serrata14.0647.64.0113.5715.5149.54.6914.970.4861.3810413.570.7042.423641.78
Begonia luxurians1.626.30.401.562.247.80.772.670.0570.161270.610.1890.90550.09
Bergenia crassifolia4.829.61.412.807.0214.
Buddleja davidii1.5111.10.241.762.3613.10.683.780.0820.39900.860.1571.49730.03
Choisya ternata4.3013.80.872.795.2316.01.293.930.1020.313031.290.4131.761880.65
Diospyros australis11.8942.02.418.5212.1841.72.458.400.2450.746462.180.4891.652390.69
Echium candicans2.127.20.441.502.808.10.952.750.1040.371880.870.3921.81820.25
Goodenia ovata2.9214.80.472.373.4416.40.653.100.1040.441790.990.3041.683280.20
Guilfoylia monostylis4.0514.71.083.926.1720.71.745.780.1370.402700.930.2570.892030.51
Heritiera trifoliolata10.6854.21.015.1512.5362.21.427.060.2741.296782.900.4241.813970.32
Hernandia bivalvis3.1317.40.502.786.2126.91.606.960.2361.122521.280.3311.681910.14
Ilex aquifolium10.5928.02.737.2116.0437.64.4710.480.6461.5710262.401.0652.492432.17
Justicia adhatoda2.7819.20.432.954.4323.01.055.500.0950.451671.210.1651.201350.07
Laurus nobilis7.5743.41.106.3314.1276.22.9716.030.2161.086394.170.6404.1710670.72
Leucopogon parviflorus6.4737.31.287.367.3636.71.507.490.4691.04176710.580.4412.6410541.01
Ligustrum japonicum8.3314.92.674.778.6514.
Luculia gratissima3.0530.90.505.124.5323.31.517.770.1180.591611.460.2622.371580.05
Magnolia grandiflora8.9225.83.389.7915.9841.25.5114.230.5231.209312.600.5841.634262.15
Mahonia lomariifolia10.5828.02.416.3817.9145.24.5911.590.5211.3412964.750.7012.576503.37
Melianthus major2.3212.50.372.013.5816.60.773.560.0710.321150.690.1811.09970.08
Nothofagus moorei8.7732.72.147.9613.7748.43.4011.970.3791.347173.310.4872.253540.70
Pancheria elegans5.1916.90.953.095.2416.51.364.300.1220.503151.080.1930.662480.96
Pelargonium tomentosum1.568.80.392.183.6711.91.795.810.1190.341370.740.1650.89950.24
Quercus coccifera10.2353.11.9510.1123.98104.34.7820.800.3991.411387111.2010.40013.2316110.76
Tasmannia lanceolata5.6313.
Viola hederacea2.119.50.391.762.3510.00.522.220.0550.231310.800.1651.011450.16
P< 0.001< 0.001< 0.001< 0.001< 0.001< 0.001< 0.001< 0.001< 0.001< 0.001< 0.001< 0.001< 0.0010.004< 0.001< 0.001
Table 5.  Spearman correlations (Rs) of mechanical properties with the Botanists’ sclerophylly index (BSI), leaf mass area (LMA) and the Loveless index (SI)
Mechanical propertiesSclerophylly indices
  1. Punch test results are given for the intercostal lamina, and for the averages of random punches across the leaf (excluding the midrib). The ratios of mechanical properties were calculated using strength and work from punch tests of the intercostal lamina. *, 0.01 ≤ P < 0.05; **, 0.001 ≤ P < 0.01; ***, P < 0.001.

Punch tests
intercostal lamina0.85***0.85***0.79***
Specific strength:
intercostal lamina0.55**0.55**0.62***
intercostal lamina0.84***0.85***0.77***
Specific work:
intercostal lamina0.77***0.79***0.80***
Shearing tests (strip)
Specific work0.66***0.65***0.71***
Tearing tests (strip)
Force per width0.87***0.85***0.87***
Specific work0.290.200.30
Bending tests (strip)
PCA components

The properties of the 2° veins and midribs are not presented here, but these also differed significantly among species (P < 0.05), with 8- to 15-fold variation among species for 2° vein properties, and 12- to 48-fold variation for midrib properties. The ratios of mechanical properties (punch tests) of intercostal tissue : main veins are not presented, but significant differences occurred among species in all ratios (P < 0.05). There was no pattern in relation to sclerophylly in ratios of strength, but for specific strength the ratios of midrib : intercostal lamina and midrib : 2° vein correlated positively with all indices of sclerophylly (Table 6). There was also a positive correlation of midrib : 2° veins with all sclerophylly indices for specific work to punch (Table 6). However, there was a negative correlation of 2° vein : intercostal lamina with BSI for work to punch and specific work (Table 6). Since the punch was sometimes larger than the 2° vein, these patterns should be interpreted cautiously.

Table 6.  Correlations (Rs) of the ratios of punch test properties for main veins : intercostal lamina with the sclerophylly indices
Mechanical propertiesSclerophylly indices
  1. Note that the punch diameter (0.5 mm) is greater than the diameter of some main veins, so the ratio reflects both mechanical properties and width of the main veins. Leucopogon parviflorus is excluded from the statistical analysis since its closely spaced parallel secondary veins may have prevented intercostal lamina punches avoiding main veins. *, 0.01 ≤ P < 0.05; **, 0.001 ≤ P < 0.01; ***, P < 0.001.

Secondary vein: lamina−0.17−0.08 0.07
Midrib: lamina−0.03 0.10 0.25
Midrib: secondary vein 0.27 0.27 0.31
Specific strength:
Secondary vein: lamina 0.18 0.19 0.33
Midrib: lamina 0.55** 0.54** 0.59**
Midrib: secondary vein 0.59** 0.54** 0.46**
Secondary vein: lamina−0.46**−0.34−0.21
Midrib: lamina−0.35−0.26−0.13
Midrib: secondary vein−0.04−0.01 0.06
Specific work:
Secondary vein: lamina−0.43*−0.34−0.15
Midrib: lamina 0.15 0.19 0.37
Midrib: secondary vein 0.58** 0.56** 0.58**

Shearing tests Properties in shear were highly correlated for strips vs whole leaves (Rp = 0.92–0.95, P < 0.001), and we report only the results of leaf strips here. There were significant differences among species for both shear properties, with 12- to 15-fold variation (Table 4). Highest values for work to shear were recorded in B. marginata followed by A. melanoxylon and I. aquifolium (Table 4). Specific work was highest in B. marginata and A. melanoxylon, followed by I. aquifolium (Table 4). Both properties were positively correlated with all sclerophylly indices, most strongly in work to shear (Table 5).

Tearing tests There were significant differences among species for each of the tearing tests, with 6- to 20-fold variation (Table 4). A. melanoxylon and Q. coccifera could not be successfully torn at the notch, and the values listed were the forces generated when the strip pulled from the screw, i.e. they represent minimum values. Highest values of force per width were recorded in Leucopogon parviflorus (only 1 successful tear) and A. melanoxylon, both of which have longitudinally parallel 2° veins, and Q. coccifera and M. lomariifolia (Table 4). Highest tensile strength was recorded in Q. coccifera and L. parviflorus (Table 4). Work to tear was highest in I. aquifolium, and highest specific work to tear was recorded in L. nobilis, A. melanoxylon and Q. coccifera (Table 4). Tearing variables other than specific work were positively correlated with all sclerophylly indices, the strongest correlation being recorded in force per strip width (Table 5).

Bending tests There were significant differences among species for both bending tests (Table 4). There was 19-fold variation in the Young modulus (E), with highest values recorded in L. nobilis and L. parviflorus, followed by A. melanoxylon. Flexural stiffness (EIW) varied 115-fold, highest in A. meiniana and M. lomariifolia. Both variables correlated positively with each sclerophylly index, the strongest correlation being with EIW (Table 5).

Patterns among mechanical properties

Most of the mechanical properties were strongly intercorrelated (Table 7). The main patterns were: intercostal lamina properties were highly correlated with the same properties averaged from random punches across the lamina (excluding the midrib) (Rp = 0.90–0.95); punch strength was highly correlated with work (toughness) from punching (Rp = 0.90–0.92) and shearing tests (Rp = 0.84–0.89), and specific punch strength was highly correlated with specific work from punching (Rp = 0.88–0.90) and shearing tests (Rp = 0.82–0.87), i.e. strong leaves were on average tough, both at the level of whole leaf and per unit leaf thickness; tearing test properties generally correlated highly with punch and shearing properties, except for specific work to tear which generally showed weaker or no correlation; of the bending properties, E, a ‘material’ property, generally showed strongest correlations with other ‘material’ properties (specific strength and specific work), and EIW, a ‘structural property’, tended to correlate most strongly with other structural properties (strength and work); and for all test types, structural properties were significantly correlated with material properties, that is, strong leaves were on average strong per unit leaf thickness, and tough leaves were tough per unit thickness. The latter correlations were strongest in punching and shearing tests.

Table 7.  Correlations (RP) among mechanical properties (all data were loge-transformed for analysis; *, 0.01 ≤ P < 0.05; **, 0.001 ≤ P < 0.01; ***, P < 0.001)
 Punch tests: laminaPunch tests: averageShear testsTearing testsBending tests
Punch tests: lamina
Specific strength (SS)0.79***              
Work (W)0.92***0.56***             
Specific work (SW)0.91***0.88***0.85***            
Punch tests: average
Strength (S)0.94***0.78***0.89***0.92***           
Specific strength0.80***0.95***0.60***0.88***0.87***          
Specific work0.78***0.79***0.74***0.90***0.91***0.90***0.86***        
Shear tests
Specific work0.78***0.82***0.66***0.83***0.85***0.87***0.71***0.87***0.89***      
Tearing tests
Force per width (F)0.90***0.73***0.85***0.86***0.88***0.74***0.76***0.75***0.90***0.80***     
Specific work0.43*0.65***0.250.51**0.46**0.63***0.300.57**0.49**0.67***0.55**0.76***0.64***  
Bending tests

However, there is a danger in ignoring the detail of the correlations, i.e. in concentrating on average patterns, in that species do vary considerably in the way they combine mechanical properties. For example, A. meiniana showed high toughness (work to punch) relative to strength compared with other species, and H. trifoliolata was relatively strong per unit toughness (Fig. 3a). There was wide variation in EIW among tough species, with particularly high EIW in M. lomariifolia and A. meiniana (Fig. 3b), though M. lomariifolia had higher strength than A. meiniana (Fig. 3c). By contrast, H. trifoliolata combined high strength with low EIW (Fig. 3c). When ratios of these structural properties are compared with the sclerophylly indices some further patterns emerge (Fig. 4, Table 5). First, toughness (work to punch) : strength, stiffness (EIW) : toughness and stiffness : strength were generally positively correlated with each of the sclerophylly indices (Table 5). Second, there was wide variation in character combinations among the most sclerophyllous species (Fig. 4).

Figure 3.

Relationships between the structural properties punch strength, work to punch (punch tests of intercostal lamina) and flexural stiffness, EIW. a, Agapetes meiniana; h, Heritiera trifoliolata; m, Mahonia lomariifolia. The data are untransformed to show the full range of variation, with a fitted line derived by linear regression. In statistical analyses (see Results section) the data were loge-transformed.

Figure 4.

The relationship between ratios of structural properties, strength, toughness and stiffness, and BSI. (a) work to punch (toughness) per unit punch strength (punch tests of intercostal lamina); (b) flexural stiffness, EIW, per unit work to punch; (c) flexural stiffness, EIW, per unit punch strength. Some of the tougher species are indicated: a, Agapetes meiniana; b, Banksia serrata; m, Mahonia lomariifolia; q, Quercus coccifera. Spearman correlations are given in Table 5.

Relative contributions of mechanical properties to sclerophylly

The pattern of variation among species in mechanical properties is summarized in the PCA configuration plot (Fig. 5). Of the total variation among species, 73% was explained by Component 1, consistent with the high degree of correlation among mechanical properties. All properties had loadings above 0.6 on Component 1, with five properties ≥ 0.9 (punch strength and specific work, work to shear and specific work to shear and force to tear). Specific work to tear and EIW had the highest loadings (−0.57 and 0.57, respectively) on Component 2 which contributed 13% of the total variation. Component 1 did not correlate so highly with the sclerophylly indices as some individual mechanical properties (Table 5), and therefore did not provide a ‘combination variable’ that better summarized the mechanics of sclerophylly. Again, the wide variation in character combinations among the most sclerophyllous species can be seen (Fig. 5).

Figure 5.

PCA of leaf biomechanical properties and the relationship of the main component to sclerophylly. (a) PCA configuration plot (Components 1 and 2); (b) BSI vs Component 1. In (a) the triangles indicate the nine species with the highest BSI scores.

Given the high degree of correlation among mechanical properties, we used hierarchical partitioning to assess the independent contribution of the major properties to sclerophylly. We narrowed the data set to the three main structural properties of strength, toughness (work to fracture) (both properties from punch tests of random positions) and flexural stiffness (EIW), given that material properties did not correlate highly with sclerophylly. The results indicated that each of these properties made a substantial independent contribution to the variance in sclerophylly indices, with total contributions of 0.74–0.82 (Table 8). The strongest contributors were flexural stiffness and strength, particularly in the case of BSI, where the contribution of flexural stiffness was 74% higher than that of work to punch (Table 8). The BIC indicated that the best predictive model was a combination of EIW and strength to punch (Table 8). The same conclusions were reached when work to shear was substituted for work to punch. Multiple regression indicated that 79% of the variation in BSI and LMA, and 72% in SI, was explained by EIW plus strength to punch (all loge-transformed except BSI).

Table 8.  Contributions of the "structural" mechanical properties – strength, toughness (work to punch) and flexural stiffness – to sclerophylly
IndexMechanical propertyIZ-scoreBIC
  1. Strength and work to punch are taken from punch tests at random positions across the lamina, excluding the midrib. The data include results from hierarchical partitioning analysis: I is the total independent contribution of the independent variable (mechanical property) to the variance in the sclerophylly index; the Z-score is derived by randomization procedures and also provides a measure of the independent contribution of each independent variable. BIC indicates the combination of mechanical properties providing the best fitting model. (L), data were loge-transformed.

Average punch strength(L)0.2827.73include
Average work to punch(L)0.1975.15
Average punch strength(L)0.2766.66include
Average work to punch(L)0.2435.77
Average punch strength(L)0.2846.35include
Average work to punch(L)0.1984.20

Fibre, nitrogen and phosphorus

Percent NDF correlated positively, and N and P correlated negatively with each of the sclerophylly indices (Fig. 6). The correlation of N expressed as a percentage of nonfibre d. wt (an estimate of protoplasmic N) with the sclerophylly indices was significant (Rs = −0.51 to −0.59; P = 0.003–< 0.001) but weaker than that of N per unit mass (Fig. 6), indicating that much of the strength of the latter relationship was due to dilution by cell wall in sclerophyllous species. Similarly, the correlation of P expressed as a percentage of nonfibre d. wt with the sclerophylly indices (Rs = −0.42 to −0.54, P = 0.016–0.002) was weaker than that of P per unit mass (Fig. 6).

Figure 6.

Correlations of fibre (NDF), nitrogen and phosphorus with the sclerophylly indices.

NDF concentration correlated positively with all mechanical variables, most strongly with lamina and average punch strength, lamina and average specific punch strength, lamina specific punch strength, specific work to shear, force to tear and E (R = 0.70–0.75, P < 0.001). However, NDF concentration explained only 18–52% of the variation in mechanical properties.


Six major conclusions regarding the biomechanics of sclerophyllous leaves can be drawn from this study:

(1) Sclerophyllous leaves have high strength to punch, work to punch, work to shear, force to tear and EIW (all ‘structural’ properties), i.e. sclerophyllous leaves are generally stronger, tougher and stiffer than softer leaves. The same conclusion is reached irrespective of the sclerophylly index used (BSI, LMA and SI). High strength and high toughness are usually mutually exclusive in isotropic materials. However, there are ways of combining materials to give high strength and promote high toughness (Atkins, 1974; Atkins & Mai, 1985). For example, high toughness can be achieved in structures comprising strong materials by mechanisms that promote fibre pull-out (the use of strong fibres glued weakly to the adjacent matrix, allowing dissipation of the fracture energy as the fibre shears through the matrix), together with other features that arrest cracks once they have been initiated (Atkins & Mai, 1985). The known structure of leaves can achieve such combinations of properties (high strength, toughness and stiffness) by incorporating the high strength of cellulose in structures that act at various scales (cellulose microfibril to whole-leaf anatomy) to enhance toughness (e.g. cellulose chains or vascular strands provide fibre–pull-out and air spaces arrest cracks) and stiffness (e.g. positioning S1 and S2 layers at different angles in the cell wall).

(2) Sclerophyllous leaves on average increase more in toughness than strength, and in flexural stiffness more than strength or toughness. This stiffness could be conferred, for example, by thickening the cuticle or the outer wall of the epidermis. This thickening could provide a direct effect as well as an indirect effect of thickening the most distant tissues from the central axis of the leaf (increasing EI). Thickening the cuticle need not substantially increase the strength or toughness of the leaf.

(3) Sclerophyllous leaves, on average, have higher specific strength, specific toughness and E (‘material’ properties, i.e. standardized for leaf thickness). However, sclerophylly is better characterized by whole leaf (structural) properties than by material properties. In this study we have ignored some aspects of leaf structure that may contribute to whole leaf mechanics, because of our assumption that sclerophylly is best characterized by lamina properties rather than midrib and margin properties. However, whole-leaf properties can be influenced by both of the latter. For example, the midrib of some species, such as Banksia marginata, is a greater proportion of the width of the leaf, which would strongly influence whole-leaf flexural properties. The revolute margins of M. grandiflora, and undulating and reinforced margins of I. aquifolium, must also influence flexural properties in particular. However, estimation of whole leaf properties is particularly difficult in such leaves because of their extreme anisotropic nature and tissue arrangement.

(4) Punch test properties of intercostal lamina correlate similarly with the sclerophylly indices to the random punches across the leaf lamina. That is, sclerophyllous leaves generally have strong and tough intercostal tissue, irrespective of the properties and densities of secondary veins. Leaf anatomy will be discussed in a later paper.

(5) Hierarchical partitioning indicated that strength, toughness and flexural stiffness each made a substantial independent contribution to the variance in sclerophylly indices, but the best explanators were flexural stiffness and strength, with the best predictive model being a combination of these two properties. In the case of BSI, flexural stiffness contributed 1.7× that of toughness, and 1.2× that of strength. BSI is derived from ranks, rather than from values. Hence, the shape of the relationship of BSI with the mechanical variables may change if botanists had assigned relative values rather than only ranks. However, analyses in relation to the other two indices, LMA and SI, yielded the same conclusion as for BSI. Choong et al. (1992) considered that sclerophylly was a function of fracture toughness (syn. specific work to shear) rather than whole lamina properties that are influenced by leaf thickness. However, from our study there is no evidence that toughness per unit thickness is a superior explanator of sclerophylly; instead it correlated less well with sclerophylly indices than toughness uncorrected for leaf thickness. Furthermore, from the definitions of sclerophylly used in the past, it is not clear that specific work should be a more central property to leaves described as hard, stiff, tough, leathery (e.g. Schimper, 1903) than stiffness and strength. ‘Hardness’ as defined by materials engineers is a surface property that is appropriate to measure in isotropic materials, but which is less meaningful in relation to whole leaf properties, and when used by Schimper and others ‘hardness’ was probably not used in this sense (Edwards et al., 2000) and this usage might relate more to flexural stiffness. Lucas et al. (2000) note that hardness may be influenced by features such as local deposits of silica, and this feature probably warrants further investigation, even if not a major contributing factor to sclerophylly.

The apparent importance of flexural stiffness is consistent with the methods used by botanists to rank the leaves. When asked after ranking the leaves what properties influenced their decision, all botanists cited bending or flexibility as a major character, i.e. that sclerophyllous leaves were more difficult to bend. No one cited breaking or fracture as a major character. In the bending tests used, the bending is undertaken over a short displacement, and shearing stress is minimized, i.e. the test probably approximated quite closely what the botanists were attempting by hand. Cowling & Campbell (1983) used a subjective classification of sclerophylls as ‘leaves hard, coriaceous and thick, breaking when folded’. In the folding, they bent the leaves over a sufficient range to ensure their fracture, and therefore may have a slightly different sense of sclerophylly from the botanists ranking sclerophylly in this study, i.e. they emphasized the property of brittleness. Atkins & Mai (1985) distinguish between elastic, elastoplastic and plastic fracture essentially by the size of the region adjacent to the crack surface that is deformed plastically and hence irreversibly during the fracture process. In elastic materials this region is very small and so the two surfaces can be put back together almost perfectly. Such materials are defined as brittle, and elastic fracture is often called brittle fracture for this reason. Brittleness should not be confused with stiffness because some materials such as jelly can fracture in a brittle fashion. In our experience some stiff sclerophyllous leaves, such as those of Eucalyptus baxteri which have closely packed layers of turgid cells throughout the mesophyll, snap (often audibly), in a brittle fashion when folded beyond their yield strength. However, in others, such as Banksia marginata, the crack develops slowly with almost imperceptible noise as individual veins break. In these leaves the region around the crack is more damaged and the two surfaces do not fit perfectly. This is more characteristic of elastoplastic fracture (Atkins & Mai, 1985). The eucalypt cells have very high connectivity, and coupled with high turgidity there is little opportunity for them to plastically deform or slide over each other, with consequent brittle fracture. There is little opportunity for fibre pull-out in the cell walls promoting high irreversible work to fracture, and plastic work of deformation of surrounding tissues. Conversely, B. marginata fractures with extensive fibre pull-out and consequent high work to fracture and achieves high stiffness in different structural ways (Read et al., 2000). Therefore, the coriaceous leaves identified by Cowling & Campbell (1983), may be a subset of a larger group of scleromorphic leaves. Since sclerophylly has been so loosely defined, botanists may vary in their concept of the textural properties of sclerophylly and we do not claim that the botanists in our study were more ‘correct’ in their concept of sclerophylly than any others. Even so, hierarchical partitioning does indicate that EIW and strength are also the best explanators of sclerophylly as judged by the objective indices, LMA and SI, though these indices have their own deficiencies (Witkowski & Lamont, 1991; Groom & Lamont, 1999; Edwards et al., 2000).

(6) Leaves varied considerably in all the properties tested, and in the way they combined properties, both structural and material. However, while sclerophylls are generally stiff, strong, and tough, there is some variation in the ways of achieving sclerophylly. This has implications for ways in which studies of leaf form and function are undertaken. There has been a tendency to concentrate on specific toughness, or ‘toughness’ as measured by a penetrometer (usually measuring force to fracture). This has yielded some interesting patterns, such as in studies of herbivory and recent studies by Wright & Westoby (2002) on adaptive aspects of leaf form. There may be a priori reasons for targeting features such as toughness or specific toughness, but in their absence, studies of the adaptive significance of leaf texture may benefit from including more than one biomechanical property. For example, in studies of herbivory, strength is relevant in terms of the force required to bite a leaf, and toughness in terms of the energy of processing the leaf meal.

Sclerophylly correlated positively with fibre (NDF) concentration and negatively with nitrogen and phosphorus, consistent with other studies (e.g. Loveless, 1961; Sobrado & Medina, 1980; Medina et al., 1990; Specht & Rundel, 1990; Reich et al., 1991; Choong et al., 1992; Turner, 1994b; Read et al., 2000). Furthermore, the estimated protoplasmic concentrations of nitrogen and phosphorus correlated negatively with the sclerophylly indices, indicating that low mass-based estimates in sclerophyllous plants are more than just an effect of dilution by cell wall, although the relationships with sclerophylly were weakened when expressed as an estimated protoplasmic concentration. These results are significant in that all plants were growing under similar conditions of climate and soil, and hence these correlations probably represent inherent capacities of the plants rather than direct effects of the contrasting environments to which the plants are naturally restricted.

We again emphasize that species can achieve sclerophylly (in its textural sense) in different ways, both mechanically as seen in this study, and anatomically (Read et al., 2000). Furthermore, while structural properties (fracture strength, fracture toughness and flexural stiffness) are stronger correlates of sclerophylly than material properties, and strength and stiffness contribute most strongly to sclerophylly, this does not necessarily mean that these are the only, or even the most ecologically important, mechanical characteristics. It is likely that sclerophylly is a complex of associated mechanical characteristics with both overlapping and contrasting roles, and that the roles of individual properties will be difficult to tease apart for various reasons. Firstly, textural (mechanical) properties may have multiple functions or effects. Leaves must be mechanically designed to sustain static loads that leaf tissues impose on themselves and be able to twist and bend when subjected to larger forces such as wind (Niklas, 1999). Mechanical properties may influence resistance to water deficits (Schimper, 1903; Oertli et al., 1990) and tolerance of low-nutrient soils (Loveless, 1961), and may enhance leaf longevity by leaf protection (Chabot & Hicks, 1982; Turner, 1994a). Hence, the mechanical design of a leaf is likely to involve a compromise between the requirement to withstand static and dynamic loads (Niklas, 1999) as well as the requirement for design that enhances physiological performance under resource-poor conditions.

Secondly, the specific mechanical properties may act at different scales. For example, if sclerophylly provides an advantage by resistance to cell collapse due to negative turgor pressure (Oertli et al., 1990), then mechanical properties at the level of individual cells may be more relevant than properties conferred by arrangements of tissues. Even herbivores can be influenced by mechanical properties at different scales due to their different body sizes and characteristics of their mouthparts (Bernays & Hamai, 1987; Peeters, 2002). Salleo & Nardini (2000) suggested that sclerophylly may be a nonspecific evolutionary response to a range of environmental stresses, including those mentioned previously, as well as freezing and high light irradiance. This may be a direct evolutionary response, or an indirect response as in the leaf protection hypothesis, i.e. low resource availability favours sclerophylly because resource shortage selects for leaves with long lifespans (Chabot & Hicks, 1982; Turner, 1994a). If the former is true, then different mechanical patterns may be present in contrasting environments. If the latter is true, then similar mechanical patterns may be present under contrasting conditions of environmental stress. In our study we have investigated plants native to a wide range of environments; the results therefore represent ‘average’ trends that may mask different associations of mechanical properties in leaves adapted to contrasting environments. Whether the term ‘sclerophylly’ can be abandoned and replaced with more specific mechanical terms will be influenced by the outcomes of further studies, such as those comparing leaf mechanical properties in contrasting environments. For the moment we regard it as a useful term, so long as the above caveats are recognized.

It is tempting to assume that the mechanical pattern is adaptive. However, it is also possible that the key adaptive features are aspects of the anatomy that confer physiological benefit, such as a thick cuticle reducing water loss and heat gain, and that happen to confer certain mechanical properties, the latter not being adaptive but only by-products. In addition, the mechanical properties may arise from nonadaptive consequences of resource limitation (e.g. Salleo & Nardini, 2000). Hence, assessment of the functional and ecological roles of the mechanical properties of scleromorphic leaves will be difficult because of multiple roles of mechanical properties, the contrasting scales at which the properties exert their effects, and the potential difficulty of separating adaptive from nonadaptive roles.

Finally, why measure mechanical properties in studies of sclerophylly rather than a simpler property such as LMA (or the separate components of density and thickness) or SI? The limitations of SI as an index of sclerophylly have been noted previously (Witkowski & Lamont, 1991; Groom & Lamont, 1999). However, if plant resource allocation is the aspect of interest, then both LMA and SI may have more relevance than mechanical properties, though mechanical properties are relevant if benefits of resource allocation are considered in a broad sense. LMA (or SLA) has particular relevance to studies investigating allocation patterns in relation to photosynthetic performance and growth. SI may be particularly relevant to studies of palatability because it provides a measure of dilution of nutrients by cell wall. However, if the questions are centred on the potential function of the mechanical aspects of the leaf textural form, then LMA (and its components of density and thickness) and SI provide only a coarse estimate, since leaf mechanical properties are a function not just of quantity of material, or density, but also of the type of material and its arrangement. Measures of allocation such as LMA are valuable, together with measures of mechanical aspects of leaf texture, because they record different aspects of leaf form.


We gratefully acknowledge the Royal Botanic Gardens, Melbourne, Australia, for permission to undertake this study, particularly P. Symes and T. Turner for their advice and assistance. We thank F. Clissold, C. Brunt, E. Gras and particularly P. Wevill, for their technical assistance, and C. Ashburner, D. Ashton, M. Burd, M. Clayton, D. O’Dowd, D. Gaff, B. Gott, N. Hallam, J. Hamill, G. Rayner and D. Smyth for ranking leaves, and R. MacNally for running hierarchical partitioning analyses. We are grateful to J. Vincent, P. Grubb and B. Lamont for valuable comments on this paper, J. Vincent, A. Atkins, G. Jeronimidis and N. Aranwela for helpful discussions on mechanics, and B. Lamont for having originally suggested the potential value of investigating bending. This project was funded by the Australian Research Council Small Grant Scheme.