Inter- and intra-specific variation in phyllode size and growth form among closely related...

Inter- and intra-specific variation in phyllode size and growthform among closely related Mimosaceae Acacia species acrossa semiarid landscape gradient

Gerald F. M. PageA,C, Louise E. CullenA, Stephen van LeeuwenB and Pauline F. GriersonA

AEcosystems Research Group, School of Plant Biology M090, The University of Western Australia,Crawley, WA 6009, Australia.

BWestern Australian Conservation Science Centre, Department of Environment and Conservation,Locked Bag 104, Bentley Delivery Centre, WA 6983, Australia.

CCorresponding author. Email: page@graduate.uwa.edu.au

Abstract. The mulga complex (Acacia aneura F. Muell ex Benth and closely related species) consists of woody trees andshrubs, and is distributed across 20%of theAustralian continent.A. aneura is renowned for awide variety of phyllode shapesandgrowth forms,whichmay co-occur at anyone site.Weexamined the intra- and inter-specific variation in growth formandphyllode shape in four species of the mulga complex, including A. aneura, across topographic gradients in semiarid north-westAustralia.Wemeasured792 trees across 28 sites stratified into six discrete landscapepositions; upper slope, lower slope,lowopenwoodland, bandedwoodland, lowwoodland, anddrainage line.Dominanceof phyllode shapeswas strongly relatedto landscape position. A. aneurawith terete phyllodes were dominant on the hill slopes, whereas broad phyllodes were mostcommononA.aneura in all valleywoodlands.Trends ingrowth formwere less distinct, although single-stemmed formsweremore common on hills, whereas the valleys hadmoremulti-stemmed forms. The quantification of growth form and phyllodeshape variability within the mulga complex provides a basis for the quantitative determination of functional links betweenmorphology and environmental conditions at both the site and landscape level.

Introduction

Leaf size and growth form are commonly used to identify plantfunctional types owing to the strong association between plantform and function, particularly carbon fixation and water use(Noble and Slatyer 1980; Díaz and Cabido 1997; Reich et al.1997; Weiher et al. 1999). Growth form has also long been a partof structural vegetation classification (Kent and Coker 1992).Differences in leaf size and growth form across communities andecosystems have been extensively correlated with environmentalgradients of resource availability including nutrients and water(McDonald et al. 2003; Wright et al. 2004b; Schenk et al. 2008;Ordoñez et al. 2009).However, only a few studies have examinedthese changes within a single species or species complex(Kalapos and Csontos 2003; Prior et al. 2005; Beaumont andBurns 2009; Gouveia and Freitas 2009) even though intra-specific studies facilitate a more direct examination of the linkbetween plant form and function because they are less likely tobe confounded by phylogenetic adaptation (Prior et al. 2005).

The integrated nature of plant traits makes it very difficultto attribute them to only one, or any particular combination ofenvironmental drivers (Shipley 1999; Prior et al. 2005; Westobyand Wright 2006). Much of our understanding of plant trait-resource relationships has also been developed fromMediterranean, temperate and tropical ecosystems (Falster and

Westoby 2003; McDonald et al. 2003; Pickup et al. 2005;Niinemets et al. 2007b; Meier and Leuschner 2008). In suchenvironments, interactions among numerous environmentalvariables including light, soil nutrients, water and frequentdisturbance, result in complex conditions for growth thatconfound analysis of plant traits. In contrast, ecosystems insemiarid landscapes are principally driven by wateravailability (Noy-Meir 1973; Williams and Calaby 1985;Stafford Smith and Morton 1990). While other factors areobviously of importance (e.g. nutrients, soil depth, fireregime), semiarid landscapes provide a relatively simplifiedsystem in which to study the relationship between form andfunction in plant traits.

Mulga woodlands occur across a diverse range of soil typesand topographic positions in semiarid Australia, and theirdistribution occupies close to 20% of the Australian continent(Everist 1949). Even at a local scale, mulga woodlands can growacross a range of positions in a toposequence, including upperand lower southern hill slopes, valley floors and drainage lines.The dominant canopy species of mulga woodlands are membersof the ‘mulga complex’ composed of nine closely related species(Johnson and Burrows 1994; Miller et al. 2002). While Acaciaaneura F. Muell ex Benth (mulga) is the dominant species in thecomplex, it also includes Acacia catenulata C.T. White subsp.

CSIRO PUBLISHING

www.publish.csiro.au/journals/ajb Australian Journal of Botany, 2011, 59, 426–439

� CSIRO 2011 10.1071/BT11057 0067-1924/11/050426

occidentalis Maslin (hereafter referred to as A. catenulata) andAcaciaayersianaMaconochie (Miller et al. 2002).A. aneura, andto a lesser extent A. catenulata and A. ayersiana, are renownedfor exhibiting a wide range of growth forms and phyllodeshapes that often co-occur (Johnson and Burrows 1994).Consequently 10 varieties of A. aneura are currently described(Pedley 2001), and the mulga complex has been the focus ofseveral taxonomic revisions (Pedley 1973, 2001; Randell 1992;Miller et al. 2002).Whilemuch of the variation inmorphology ofA. aneuramay be genetic (Cody 1989; Andrew et al. 2003), it isalso likely that a component of the variation is attributable tophenotypic plasticity in response to differences in growingconditions (Fox 1986; Beard 1990). However, little is knownabout the relationship between morphological diversity withinA. aneura and closely related species, or habitat attributes ofmulga that are likely related to or determined by topographicposition.

The range of both growth forms and phyllode shapes in themulga complex and their landscape distribution must bequantified in order to assess the relationship between formand function, including the potential role of morphologicalplasticity, in contributing to the variation. There have beenonly a few anecdotal descriptions of habitat preferencesby formal varieties of A. aneura (Fox 1986; Pedley 2001).A. aneura commonly occurs as a range of growth formsbetween a shrub of less than 2m in height and a tree up to15m tall and can be single- or multi-stemmed. Phyllode shapesrange from small cylindrical needles less than 1mm wide tobroader leaf-like shapes that are 10mm wide and relatively flat(Pedley 2001). Varieties of A. aneura are most readily identifiedby phyllode and seedpod characteristics, but branchingarchitecture is not often considered. Given that that oursystematic understanding of mulga is also still developing(Miller et al. 2002), it is unclear to what extent branching

architecture can change within varieties, or if phyllode traitsand branching architecture are even linked.

The objectives of this study were: (i) to describe the variationin branching architecture and phyllode shape for members ofthe mulga complex (A. aneura and closely related species)around West Angelas in the Pilbara region of WesternAustralia; and (ii) quantify the landscape distribution ofbranching architectures and phyllode shapes. We expected thatphyllode size and shape are not independent of growth form(branching architecture and tree size), and that mulga variants donot occur homogeneously across different landscape positions.For example, we expected that communities on valley floorswould be dominated by trees with single stems and broaderphyllodes while communities on upper slopes would bedominated by multi-stem trees with more terete phyllodes.

Materials and methodsSampling sitesA total of 28 transects were measured across six landscapepositions in 60 000 ha at West Angelas in the HamersleyRanges of the Pilbara region of Western Australia (–23.175,118.790). West Angelas has been recognised as one of therichest sites of intra-specific diversity within A. aneura,because of its position in a highly dissected landscape that ison the edge of the species’ distribution (Pedley 2001; McLeishet al. 2007). Valleys at West Angelas are ~700m a.s.l., andrelative elevation difference of the hills and ridges is ~250m(Fig. 1), although the highest local peak is Mount Robinson at1144m. We stratified the landscape by terrain analysis andclassification using fuzzy k-means (Vriend et al. 1988;Burrough et al. 2000). Landscape classification used fiveterrain attributes that are in part responsible for theredistribution and availability of light and water across the

Lower southern slope

Valley woodlands: low, banded, low

open and drainage line woodlands

Upper southern slope

Fig. 1. Typical landscape profile at West Angelas. Relative elevation from valley floor to upper slope is ~250m.

Morphological variation in closely related Acacia spp. Australian Journal of Botany 427

landscape. The followingfive terrain attributeswere derived froma30-mdigital elevationmodel; elevation, slope angle, cumulativedownhill slope length, soil wetness index (Moore et al. 1993;Burrough et al. 2000) and annual incident solar radiationcalculated with the program POTRAD5 for PCRaster (vanDam 2000). Seven terrain units were classified across the site,four of which contained mulga woodlands (upper and lowersouthern hill slopes, valley floors and drainage lines). Meanvalues of modelled terrain attributes for the four units thatcontained mulga woodlands are listed in Appendix 1. As~90% of the mulga woodlands occurred on valley floors, thisterrain unitwas further stratified into threemulgawoodland types:banded mulga woodland characterised by distinct linear grovesseparated bybareground, lowmulgawoodland characterisedbyauniform distribution of mulga, and low open mulga woodlandcharacterised by a sparse distribution of mulga separated by bareground and Triodia spp. Consequently, the landscape wasstratified into six positions containing mulga woodland goingfromupper to lower slopes: (i) upper southern hill slope, (ii) lowersouthern hill slope, (iii) low mulga woodland, (iv) banded orlinear mulga woodland, (v) low open mulga woodland, and (vi)drainage line mulga woodland (tall woodland) (Fig. 1). Therewere no woodlands on north-facing slopes (most exposedaspects). We established five transects within each landscapeposition across the site, except for banded mulga woodland anddrainage line mulga woodland. For the banded and drainage linemulga woodlands there were only four sites as a large bushfireburnt most of the study area in October 2006 and limited furthersampling. Belt transects were 4m wide and a minimum of 100mlong, containing a minimum of 20 individuals of A. aneura orclosely related members of the mulga complex.

Field classification of mulga growth formsWhile growth form has already been included in the classificationof somevarieties ofA.aneura (Pedley1973, 2001;Randell 1992),the descriptions have been both qualitative and confusing. Here,we created a subjective growth form classification, whichdifferentiated between single- and multi-stemmed growthhabit, and then by tree height, height to stem division andcanopy area. We measured a total of 792 mulga trees andshrubs along 28 belt transects and visually classified themaccording to a field key of growth forms in the study area(Fig. 2). Phyllode shapes were classified in a similar way.Phyllodes were subjectively classified as one of six shapes that

were prominent in the study area (Fig. 3). The six phyllode shapesbroadly distinguished between three phyllode shape variants ofA. aneura (terete, narrow and broad) and three other closelyrelated species (A. ayersiana, A. catenulata, A. minyura). Acaciaparaneurawas also present in the study area but was not capturedin the field sampling.

Branching architecture, and phyllode size and shapeMeasurements of branching architecture and phyllode size weremade to quantitatively define growth form, and determinewhether the subjective classifications could be substantiated.Six measurements that characterised branching architecturewere recorded for all trees in each belt transect: trunk basaldiameter, diameter of every stem at 50 cm from the base,height of the first stem division, height to crown break, canopyarea (measured at the maximum width and then perpendicular)and total height. Length, width and thickness to the nearest

Shrubs (<3 m)

T1 T2 T3A T3B T4A T4B T4C

Trees (>3 m)

Fig. 2. Field classification key to growth forms of Acacia aneura and closerelatives. T1 and T2 are low multi-stemmed trees, where T1 has only two orthree main stems and T2 has >three main stems of similar diameters. T3Aand T3B are low single-stemmed trees, which differ only in the spread of thecrown. T4A and T4B are tall single-stemmed trees, which differ only in thespread of the crown. T4C is a tall multi-stemmed tree, with a main stemdivision <50 cm from the ground. T5 is a single-stemmed semiconiferouslooking tree characterised by horizontal branching. S1 and S2 are shrubs<3mtall, and differ in the spread of the canopy.

(a) (c) (d ) (e) (f )

Fig. 3. Four species of Acacia including three phyllode variants of Acacia aneura that were common at West Angelas. (a) A. aneura (terete), (b) A. aneura(narrow), (c) A. aneura (broad), (d) A. catenulata, (e) A. ayersiana, ( f ) A. minyura. Hand-drawn sketches of three typical phyllodes from each type are drawnabove stylised transverse sections of each.

428 Australian Journal of Botany G. F. M. Page et al.

0.1mm were measured on 20 phyllodes collected 1–2m fromthe ground from four positions spaced equally around thecanopy from each of 285 trees and shrubs that spanned all sixlandscape positions. Approximately five individuals of eachphyllode shape were included in the subset of trees and shrubsselected at each transect. Differences in branching architecturewill be referred to hereon as differences in ‘growth form’, andphyllodedimensions as ‘phyllode shape’. ‘Phyllodevariants’willbe used only when specifically referring to phyllode shapevariants A. aneura.

Data analysisThe experimental design had four factors: landscape position(six levels, fixed), transect (four or five levels nested within eachlandscape position for a total of 28), growth form (10 levels,random) and phyllode shape (six levels, random). We tested ifmulga variants could be identified from their growth form andphyllode size, and if phyllode size was independent of growthform with a multivariate null hypothesis of no differencesamong a priori groups using permutation multivariate analysisof variance (PERMANOVA, Anderson 2001) and canonicalanalysis of principal coordinates (CAP, Anderson and Willis2003). The F-ratio constructed in PERMANOVA is analogousto Fisher’s F-ratio and is constructed from the sums ofsquared distances within and between groups (Anderson2001). All PERMANOVA tests used 9999 permutations ofthe raw data from residuals under a reduced model. CAPdetermines the ‘distinctness’ of multivariate groups byproducing a misclassification error using the ‘leave one outallocation of the groups’ approach. A higher percentage ofcorrect classifications indicated a more ‘distinct’ group on thecanonical axes (Anderson and Willis 2003). Phyllode datawere square-root transformed, and growth form attributeswere normalised. Bray–Curtis similarity matrices were used inall multivariate analysis. The assumption of homogeneity ofdispersions among groups was tested using PERMDISP. Allmultivariate analysis was performed using PRIMER version 6with the PERMANOVA+ add-on (Clarke and Gorley 2006).

We calculated the relative dominance (%) of each growthform and phyllode shape in every landscape position, and testedfor differences among landscape positions for every measuredattribute with one-way ANOVA models. Pairwise comparisonswere made with Tukey’s Honestly Significant Difference post-hoc tests. ANOVA models and post-hoc tests were calculatedusing R 2.11.0 (R Development Core Team 2010).

Results

General classification of mulga growth formsand phyllode shapes

Nearly 40% of the trees atWest Angelas were tall multi-stemmedtrees (T4C), while most others were single-stemmed trees(T3A, T3B, T4A, T4B) and multi-stemmed low trees (T1)(Appendix 2, also see Fig. 2 for tree types). The multi-stemmed shrub forms, S1 and S2, and the horizontallybranched tree form, T5, were only recorded occasionally andconstituted less than 4% of the total plants measured (Appendix2). Plant height was highly variable (1.7–16.4m), while thenumber of stems 50 cm from the ground ranged between 1 and14. Quantitative measurements of branching architectureconfirmed the validity of growth form types identifiedqualitatively (Appendix 2), as all forms were statisticallydistinct with the exception of T1, T3C and T5(PERMANOVA P-values >0.05, data not shown).

Phyllode shape was also highly variable across all 28 sites forthe four co-occurring Acacia species, particularly A. aneura(Table 1). Inter-specific differences in phyllode shape wereconsistent with taxonomic descriptions of the species (Pedley2001), and were the key feature for field identification. Phyllodewidths were broadest in A. ayersiana and decreased in the orderA. catenulata, A. minyura, broad A. aneura, narrow A. aneurawith terete A. aneura having the narrowest width. A. minyura iseasily distinguished from the other three species by its shortphyllodes (Table 1, Fig. 3). Phyllode width was highly variablefor A. aneura and varied between 0.46 and 10.3mm, which isalso the total variation described for the species (Pedley 2001).At the two extremes, terete phyllodes are nearly cylindricalwhereas the broadest phyllodes were on average nearly seventimes as wide as they were thick. A. ayersiana and A. catenulataalso showed a large variation in phyllode length and width.Phyllode width varied between 7 and 17mm in A. ayersianaandbetween1and14mminA. catenulata.Phyllode lengthvariedfrom 45 to 90mm in A. ayersiana and 30 to 130mm inA. catenulata (Table 1).

Phyllodes of fourAcacia species, including three subgroups ofA. aneura, were classified into six shape categories during fielddata collection based on a visual assessment (Fig. 3). BroadA. aneura were the most common phyllode type (38.5%),followed by terete A. aneura (22%), narrow A. aneura(19.2%) and then A. catenulata (17.4%). Both A. ayersianaand A. minyura were rarely present (<3%). Of the plants forwhich phyllode dimensions were measured, 81 were ‘terete’, 78

Table 1. Phyllode size and shape of four Acacia species, and three phyllode variants of A. aneuraMean values bolded, and standard errors given in parentheses. Data is untransformed. Lowercase letters denote significant groupings from Tukey’s Honestly

Significant Difference post-hoc tests when one-way ANOVA P< 0.05

Length (mm) Width (mm) Thickness (mm)n Min. Mean Max. Min. mean Max. Min. Mean Max.

A. ayersiana 140 45.5 65.4ab (0.66) 90.5 7.60 12.3a (0.16) 17.8 0.31 0.48a (0.006) 0.72A. catenulata 840 29.7 80.2c (0.55) 133.2 0.97 6.30b (0.07) 14 0.15 0.41b (0.003) 0.87A. minyura 220 8.39 26.7d (1.08) 45.3 2.59 5.86c (0.13) 10.8 0.31 0.46ab (0.006) 0.72A. aneura (terete) 1617 7.40 62.6a (0.34) 115.4 0.46 1.01d (0.01) 2.23 0.25 0.64c (0.004) 1.50A. aneura (narrow) 1560 23.9 66.3b (0.44) 148.9 0.48 1.62e (0.02) 4.99 0.20 0.56d (0.004) 1.18A. aneura (broad) 1440 19.2 63.5a (0.36) 114.6 0.65 3.18f (0.03) 10.3 0.17 0.47a (0.003) 0.99

were ‘narrow’, 70 were ‘broad’, 40 were ‘A. catenulata’, 7 were‘A. ayersiana’ and 9were ‘A.minyura’. All phyllode shapesweresignificantly different (PERMANOVA P < 0.001); however,the assumption of homogeneity of dispersions was notfulfilled (P = 0.004 PERMDISP). The standard error of groupdispersions were consistent among phyllode shapes, except for‘A.ayersiana’,whichhadonly seven samples and a standard errorless than 50% of all other groups. There was significant variationamong phyllode shapes at the transect level (PERMANOVAP < 0.001), but not among landscape positions (PERMANOVAP > 0.05). Estimates of the components of variation indicatethat most variation was among phyllode shapes, and thattransects, and phyllode shape by transect interactions hadestimates comparable with the residuals, which were 20% ofthe variation among phyllode categories.

There was no clear association between growth forms andphyllode shapes for the 792 trees and shrubs measured in thestudy, particularly within the population of A. aneura. Teretephyllodes were found on seven different growth forms, includingboth multi- and single-stemmed shrubs and trees, and a similarpattern was observed for the ‘narrow’ phyllode shape. Sevendifferent growth forms also had broad phyllodes; nevertheless,just under half of individuals with broad phyllodes occurred onmulti-stemmed trees (T4C). Similarly 50% of A. catenulataweremulti-stemmed trees (T4C). A. minyura had five distinct growthforms, but mostly occurred (70%) as a multi-stemmed shrub inthe category ‘S2’ (Fig. 4).

Distribution of growth forms in relation to landscapeposition

Growth form varied with landscape position (PERMANOVAP = 0.06). Drainage lines, banded woodlands and lower hillslopes had the most distinct assemblages of growth forms

(40–50% allocation success in CAP cross validation), whilethe three other landscape positions were indistinguishable fromeach other (0% success rate). Accordingly, drainage line andbanded woodlands are clearly separated along the first principlecoordinate axis, which is driven by the abundance of the treegrowth forms T4C, T4B and T4A (Fig. 5).

Tall trees with two or three main stems that divide close to theground (T4C) are the most abundant growth form at WestAngelas, and dominate all landscape positions except for theupper hill slopes, which are instead dominated by low single-stemmed trees with narrow crowns (T3B). On the lower hillslopes low trees with both spreading crowns (T3A) andnarrow crowns (T3B) together are equal in abundance toT4C. Taller trees are more dominant in the lower landscapepositions, particularly in banded woodlands and the drainagelines. Open woodlands contained a high diversity of tree types:while tall multi-stemmed trees account for half of the growthforms present in open woodlands, the other half is evenlycomprised of all other growth forms found at the WestAngelas study site. Similarly, ~30% of the woodlands aredominated by a range of forms including trees and shrubs withmore than three main stems branching at ground level (S1, S2,T1 and T2, Appendix 3).

Each of the measured growth form attributes also varied withlandscape position (Table 2). Basal diameters were smallest(generally <120mm) on the upper hill slopes and largest(>170mm) in the open and banded woodlands. Woodlandsand open woodlands have the most number of stems per tree(generally two or more), while trees and shrubs on drainage linesand upper hill slopes generally had only one stem. The height tocrown breakwas uniform across the landscapewith the exceptionof the tall trees in the drainage lines where crown break occurred2m or higher. Nevertheless, there was a trend in average treeheight from highest in the valleys (drainage lines, woodlands,

A. aneura (terete)

A. aneura (narrow)

A. aneura (broad)

A. catenulata A. minyura A. ayersiana

Fig. 4. The relative abundance of growth forms present in four species of the mulga complex at West Angelas, including three phyllode variants of Acaciaaneura. Horizontal and vertical stripes represent multi-stemmed growth forms, diagonal stripes represent single-stemmed tree forms, and the white dotted fillrepresent the dominant multi-stemmed tree ‘T4C’. While A. ayersiana was only present as T4C, only four trees were sampled.

open woodlands and banded woodlands) to lowest on hill slopes.Canopy areas were smallest in drainage lines, woodlands andupper hill slopes and largest in the banded woodlands.

Distribution of phyllode shapes in relation to landscapeposition

Each landscape position also contained unique assemblages ofphyllode shapes (ANOSIM, P= 0.04), particularly whencomparing hill slopes to the valley woodlands. Upper hill

slopes had the most distinct assemblages (60% crossvalidation success in CAP), while lower hill slopes, openwoodlands (40%) and banded woodland (25%) were slightlyless distinct. Woodlands and drainage lines were difficult toclassify based on phyllode composition and had a 0%allocation success, meaning that they were indistinguishablefrom other landscape positions. The low allocation success ofdrainage lines occurred because of the similarity in phyllodecomposition to banded woodlands. In contrast, open woodlandand the two hill slope woodlands were well separated on the firsttwo principle coordinate axes (Fig. 6). Banded woodlands anddrainage lines were scattered with no clear separation, and eventhough woodlands were not well distinguished by crossvalidation (0% allocation success) they were well separated bythe first two principle coordinate axes (Fig. 6).

Broad phyllodes of A. aneura dominated all four of thevalley woodland communities, and were most dominant in thedrainage lines (Fig. 7). Indeed, two of the drainage sites weremonocultures of broad phyllode mulga (data not shown). Incontrast, the two hill slope woodlands had very few broadphyllodes, but a much higher representation of teretephyllodes. The abundance of narrow phyllodes was consistentamong all landscape positions except drainage lines. A. minyurawas rare, and A. catenulata was not present in large numbersexcept on upper hill slopes. Of all the landscape positions,the open woodlands had the most even distribution of all thephyllode shapes.

There was no evidence that A. catenulata or any of the threephyllode variants ofA. aneura changed growth form according tolandscapeposition (Appendix4).Close to90%of tereteA.aneuragrew as single-stemmed trees in both upper hill slopes anddrainage lines, but the valley woodlands and lower hill slopeshad an even composition of single- and multi-stemmed forms.Similarly, 100% of broad A. aneura was multi-stemmed on theupper hill slopes, but elsewhere was evenly split between multi-and single-stemmed growth forms. Narrow A. aneura andA. catenulata were both evenly present as single-stemmed andmulti-stemmed growth forms in all landscape positions(Appendix 4).

–0.4 –0.2 0 0.2 0.4 0.6

–0.2

Low woodland

Low open woodland

Banded woodland

Lower hill slope

Upper hill slope

Drainage line

Fig. 5. Canonical analysis of principal coordinates (CAP) of growth formdensity at each site labelled by landscape position. Hill slopes were separatedfromdrainage lines and bandedwoodlands on thefirst axis (a),whichwere themost distinct groups by cross validation. Contribution vectors indicate thatmulti-stemmed and single-stemmed tall trees weremore abundant in drainagelines and banded woodlands, and multi-stemmed shrubs and low trees weremore abundant on hill slopes and the remaining valley woodlands (b). CAPwas generated from a Bray–Curtis similarity matrix derived from square-roottransformations of the density of growth forms at each site.

Table 2. Branching attributes changed according to landscape positionMean values are bolded and standard errors are in parentheses. P-values indicate significance level for one-way ANOVA for each attribute. Lowercase letters

denote significant groupings for each attribute according to Tukey’s Honestly Significant Difference post-hoc tests

Landscape position

Drainagelin

Bandedwoo

ersouthern

hillslop

southern

hillslop

P-value

Basal diameter (mm) 148a (5.59) 185b (7.60) 175b (6.30) 147a (5.83) 138a (6.42) 112c (4.85) <0.001Number of stems 1.41a (0.05) 1.75ac (0.13) 2.24bd (0.14) 2.35b (0.12) 1.76cd (0.09) 1.33a (0.05) <0.001Diameter of stem 50 cm from ground 113a (4.55) 138bc (6.02) 115ac (4.11) 100ad (4.01) 111a (4.78) 91d (4.50) <0.001Height to first stem division (m) 1.71a (0.12) 1.28ac (0.13) 0.86b (0.09) 0.68b (0.08) 0.78bc (0.07) 1.14ac (0.08) <0.001Height to crown break (m) 5.53a (0.15) 3.60b (0.25) 2.53b (0.13) 2.49b (0.11) 2.04b (0.10) 2.24b (0.09) <0.001Canopy area (m2) 12.4a (0.96) 18.8b (1.34) 15.8bc (0.89) 12.5a (0.82) 15.3bc (0.97) 12.6ac (1.03) <0.001Tree height (m) 7.97a (0.19) 7.41a (0.21) 5.95bc (0.13) 6.12b (0.10) 5.71bc (0.12) 5.62c (0.17) <0.001

Discussion

Variation in phyllode shape in A. aneura

Phyllode shape of A. aneura was strongly influenced bylandscape position (Figs 6, 7) although the relationshipbetween growth form and landscape position was less clear(Figs 4, 5). Shifts in average leaf size across topographicgradients have previously been reported at the communitylevel. For example, average leaf size in Californian chaparraldeclines with increasing incident solar radiation, which in turn isdetermined by topographic position (Ackerly et al. 2002). Largerleaved species were present across the landscape, while smallerleaved species were absent from the shadier sites (Ackerly et al.

2002). We also observed species with larger phyllodes (i.e.A. catenulata) across all landscape positions, but the smallestphyllodes (terete A. aneura) were most prominent on theshaded upper southern hill slopes. Smaller leaf sizes andlower specific leaf area are traits often associated with aridity,including high heat and light typical of exposed habitats.Increasing leaf thickness and/or density can be a photo-protective adaptation to high light (Niinemets and Tenhunen1997; Niinemets et al. 1999). Decreasing leaf width increasesleaf cooling through higher boundary layer conductance (Givnish1987; Niinemets et al. 1999). Our finding that the smallestphyllodes were in the most shaded sites (i.e. south-facing) isunusual given that other semiarid studies have shown plants withthe smallest leaf sizes (or photosynthetic lamina) are on the mostexposed sites (Ackerly et al. 2002; Cornwell and Ackerly 2009).Our results suggest that high light or temperature may not bethe most significant factors controlling the distribution of mulgaat West Angelas.

A gradient in water availability controlled by landscapeposition may be driving the differences in phyllode size ofA. aneura. Similar shifts in leaf size-based traits of perennialvegetationhavepreviously been reported alonggradients ofwateravailability at a continental (Wright et al. 2004a), community(Prior et al. 2003; Cornwell and Ackerly 2009), and specieslevel (Farrell and Ashton 1978; Prior et al. 2003, 2005). While itis often hard to separate the influence of heat, light and wateravailability in hot dry environments, the prominence of thesmallest phyllode sizes in the most shaded environments thatoccur at high elevations indicates that low water availability maybe a significant factor. Smaller leaves (often with higher specificleaf area)may contain a greater investment in leaf structural tissuethan photosynthetic tissue (Niinemets et al. 2007a), making themmore capable of withstanding severe water deficit. Phyllodeshave, in themselves, been considered an adaptation to droughtowing to their anatomical differences with leaves (Boughton1986; Warwick and Thukten 2006). Phyllodes of Acaciamelanoxylon are better able to recover from more negativewater potentials, and have higher water-use efficiency thanleaves from the same species (Brodribb and Hill 1993).Seedlings grown from seeds collected at drier sites switchedmore quickly from leaves to phyllodes than those from wettersites when grown under uniform conditions (Farrell and Ashton1978). Given that most features of phyllode anatomy areconsidered to be an adaptation to drought and that we haveobserved the opposite trend in phyllode size than would beexpected under reduced light and temperature conditions,we hypothesise that a gradient in water availability is drivinga shift in phyllode size among landscape positions at WestAngelas.

Co-occurrence of species with different phyllode shapes

Differences in the distribution and relative abundance ofA. aneura and A. catenulata may also reflect differentstrategies of resource acquisition and use. For example, thedominance of larger phyllode sizes in the lowest parts of thelandscape may reflect a competitive advantage over smallerphyllode sizes in conditions of higher resource availability.Competitive exclusion of smaller leaf sizes due to light

–0.3 –0.2 –0.1 0 0.1 0.2

–0.2

–0.1

A. aneura (terete)

A. aneura (narrow)

A. aneura (broad)

A. catenulata

A. ayersiana

A. minyura

–0.2

–0.1

Low woodland

Low open woodland

Banded woodland

Lower hill slope

Upper hill slope

Drainage line

Fig. 6. Canonical analysis of principal coordinates (CAP) of phyllodeshape and species density at each site labelled by landscape position.(a) Hill slopes and valleys were well separated by the first two axes.Upper hill slopes were the most distinct by cross validation (60%allocation success), followed by lower hill slopes, open woodlands (40%)and banded woodlands (25%). Woodlands and drainage lines wereindistinguishable by cross validation (0%). Correlations of phyllode shapeswith CAP axes indicated that terete and broad Acacia aneura contributedstrongly to the first CAP axis, and A. catenulata and A. minyura to thesecond (b).

limitation may also control community assembly of leaf sizes inthe Californian chaparral (Ackerly et al. 2002). Global trendsamong plant traits indicate that smaller leaves have a long life-span at the cost of lower rates of mass-based photosynthesis andlower rates of dark respiration, all of which result in slowergrowth rates (Wright et al. 2004b; Poorter et al. 2009). Trees alsogrow taller when there is competition for light (King 1990; Kochet al. 2004). Average tree height was greatest in landscapepositions that had the lowest relative abundance of tereteA. aneura (Table 2, Fig. 7). If larger phyllode sizes inA. aneura translates into faster growth rates, broad A. aneuramay competitively exclude terete A. aneurawhenwater and lightavailability are high (i.e. drainage lines, mulga groves and valleywoodlands). Conversely, when light is low and water runoffpotential high (i.e. south-facing hill slopes) there is greaterdiversification in water-use strategies and less competition forlight resulting in a greater range of phyllode sizes and loweraverage tree height.

Contrasting patterns in phyllode shape and growth form

In contrast to the trend observed in A. aneura, the closely relatedA. catenulata exhibited a constant phyllode size and shape acrosslandscape positions and so does not appear to be subject to thesame selection pressures. Heterogeneous soil and light resourceswithin sites can also influence the distribution of these closelyrelated Acacia species below the landscape level. Hill slopeshave a mixture of rock outcrops, rocky colluvium and smallreservoirs of soil. Similarly, shaded slopes produce a variety ofexposures to direct light, with trees growing directly beneathvertical rock faces receiving no direct sunlight at any time during

the day, while trees farther away intercept varying amounts ofsunlight depending on season. Similarly, global (Wright et al.2004b; Ordoñez et al. 2009) and local leaf trait models(Ackerly et al. 2002; Beaumont and Burns 2009) include veryhigh variability and so we would not expect consistency at thecommunity level particularly among species. If heterogeneouslight or soil conditions contribute to species co-occurrence, thenthe distribution of A. aneura and A. catenulata within sites mayprovide insight into the disconnection between landscape andphyllode shape for A. catenulata.

The systematic variation of phyllode size across the landscapesequence at West Angelas strongly supports the link betweenform and function (particularly water use) in morphologicallydiverse mulga woodlands. It is surprising, however, that therelationship between growth form and landscape was lessclear. Arid plants with multiple stems have been associatedwith increased drought tolerance due to their hydraulicallymodular design, where segmentation of the xylem pathwayensures that drought damage is not propagated throughout therest of the plant (Schenk et al. 2008; Espino and Schenk 2009). Ifmultiple stems were an adaptation to drought in A. aneura wewould have expected a relationship between growth form andphyllode shape, or a similar relationship with landscape as forterete A. aneura. However, hydraulic modularity may also beevident at the anatomic level and not simply expressed by thenumber of stems (Schenk et al. 2008). So while the number ofstems does not infer drought tolerance, hydraulic modularitymay still play a role the physiological function of mulga.Similarly, the capacity for A. catenulata to inhabit multiplelandscape positions without changing either abundance orphenotype may also indicate plasticity in traits other than

Draina

line Lo

woodla

A. minyura

A. catenulata

A. aneura (broad)

A. aneura (narrow)

A. aneura (terete)

Fig. 7. Relative dominance (%) of phyllode variants and species by landscape position. Terete and narrow phyllode variants of Acacia aneura wereprominent on hill slopes, and broad phyllodes were most common in the open valleys. A. catenulata was present in all woodlands.

phyllode size, i.e. hydraulic architecture including vessel sizedistributions (Tyree and Ewers 1991; Sperry 2000; Cochardet al. 2002; Mokany et al. 2003; Cavender-Bares et al. 2004)or leaf level physiological adaptations, including nitrogenpartitioning (Evans 1989; Reich et al. 1998), contributions bystable osmotica (Hsiao et al. 1976; Merchant et al. 2006, 2007),and root architecture (Lynch 1995; North and Nobel 1998).

Conclusions

In summary, our study demonstrates that landscape positionstrongly influences phyllode shape in A. aneura, but notA. ayersiana or A. catenulata. We also determined that growthform was less strongly related to landscape position. While itappears that water availability may be responsible for drivingdifferences in phyllode shape further research is required toexamine whether differences in phyllode shape amongvarieties of A. aneura are reflected by or contribute todifferences in water use and drought tolerance.

Acknowledgements

We thank Bruce Maslin and Jordan Reid for identification of mulgaspecimens, and for discussions on mulga ecology. Matthias Boer providedadvice on terrain classification. Thanks also to Justin Fairhead andRebecca McIntyre for field assistance. Figure 3 was kindly drawn byAlison O’Donnell. Funding and field support from Rio Tinto Iron OrePty Ltd, specifically Sally Madden and Neville Havelberg, is gratefullyacknowledged. We also thank two anonymous reviewers who alsoimproved the quality of the manuscript.

References

Ackerly D, Knight C,Weiss S, BartonK, StarmerK (2002) Leaf size, specificleaf area and microhabitat distribution of chaparral woody plants:contrasting patterns in species level and community level analyses.Oecologia 130, 449–457. doi:10.1007/s004420100805

AndersonMJ (2001) A newmethod for non-parametric multivariate analysisof variance. Austral Ecology 26, 32–46.

AndersonMJ,Willis TJ (2003) Canonical analysis of principal coordinates: auseful method of constrained ordination for ecology. Ecology 84,511–525. doi:10.1890/0012-9658(2003)084[0511:CAOPCA]2.0.CO;2

Andrew RL, Miller JT, Peakall R, Crisp MD, Bayer RJ (2003) Genetic,cytogenetic and morphological patterns in a mixed mulga population:evidence for apomixis. Australian Systematic Botany 16, 69–80.doi:10.1071/SB01043

Beard JS (1990) ‘Plant life ofWesternAustralia.’ (KangarooPress:Kenthurst,NSW)

Beaumont S, Burns K (2009) Vertical gradients in leaf trait diversity in aNew Zealand forest. Trees - Structure and Function 23, 339–346.

Boughton VH (1986) Phyllode structure, taxonomy and distribution in someAustralian acacias. Australian Journal of Botany 34, 663–674.doi:10.1071/BT9860663

Brodribb T, Hill RS (1993) A physiological comparison of leaves andphyllodes in Acacia melanoxylon. Australian Journal of Botany 41,293–305. doi:10.1071/BT9930293

Burrough PA, van Gaans PFM, MacMillan RA (2000) High-resolutionlandform classification using fuzzy k-means. Fuzzy Sets and Systems113, 37–52. doi:10.1016/S0165-0114(99)00011-1

Cavender-Bares J,KitajimaK,BazzazFA(2004)Multiple trait associations inrelation to habitat differentiation among 17 Floridian oak species.Ecological Monographs 74, 635–662. doi:10.1890/03-4007

Clarke KR, Gorley RN (2006) ‘PRIMER version 6: user manual/tutorial.’(PRIMER-E Ltd: Plymouth, UK)

Cochard H, Coll L, Le Roux X, Ameglio T (2002) Unraveling the effects ofplant hydraulics on stomatal closure during water stress in walnut. PlantPhysiology 128, 282–290. doi:10.1104/pp.010400

Cody ML (1989) Morphological variation in mulga. I. Variation andcovariation within and among Acacia aneura populations. IsraelJournal of Botany 38, 241–257.

Cornwell WK, Ackerly DD (2009) Community assembly and shifts in planttrait distributions across an environmental gradient in coastal California.Ecological Monographs 79, 109–126. doi:10.1890/07-1134.1

Díaz S, Cabido M (1997) Plant functional types and ecosystem function inrelation to global change. Journal of Vegetation Science 8, 463–474.

Espino S, SchenkHJ (2009)Hydraulically integrated ormodular?Comparingwhole-plant-level hydraulic systems between two desert shrub specieswith different growth forms. New Phytologist 183, 142–152.doi:10.1111/j.1469-8137.2009.02828.x

Evans JR (1989) Photosynthesis and nitrogen relationships in leaves of C3plants. Oecologia 78, 9–19. doi:10.1007/BF00377192

Everist SL (1949) Mulga (Acacia aneura F. Muell.) in Queensland. TheQueensland Journal of Agricultural Science 6, 87–131.

FalsterDS,WestobyM (2003)Leaf size and angle varywidely across species:what consequences for light interception?NewPhytologist 158, 509–525.doi:10.1046/j.1469-8137.2003.00765.x

Farrell TP, Ashton DH (1978) Population studies on Acacia melanoxylonR. Br. I. Variation in seed and vegetative characteristics. AustralianJournal of Botany 26, 365–379. doi:10.1071/BT9780365

Fox JED (1986) Vegetation: diversity of the Mulga species. In ‘The Mulgalands’. (Ed. PS Sattler) pp. 27–32. (Royal Society of Queensland: NorthQuay, Qld)

Givnish TJ (1987) Comparative studies of leaf form: assessing the relativeroles of selective pressures and phylogenetic constraints.NewPhytologist106, 131–160. doi:10.1111/j.1469-8137.1987.tb04687.x

Gouveia A, Freitas H (2009) Modulation of leaf attributes and water useefficiency inQuercus suber along a rainfall gradient.Trees - Structure andFunction 23, 267–275.

HsiaoTC,AcevedoE,FereresE,HendersonDW(1976)Water stress, growth,and osmotic adjustment. Philosophical Transactions of the Royal Societyof London. Series B, Biological Sciences 273, 479–500.doi:10.1098/rstb.1976.0026

Johnson RW, Burrows WH (1994) Acacia open-forests, woodlands andshublands. In ‘Australian vegetation’. (Ed. RH Groves) pp. 257–290.(Cambridge University Press: Cambridge, UK)

Kalapos T, Csontos P (2003) Variation in leaf structure and function of theMediterranean treeFraxinus ornusL. growing in ecologically contrastinghabitats at the margin of its range. Plant Biosystems 137, 73–82.doi:10.1080/11263500312331351351

Kent M, Coker P (1992) ‘Vegetation description and analysis: a practicalapproach.’ (CRC Press: London)

KingDA (1990)The adaptive significance of tree height.AmericanNaturalist135, 809–828. doi:10.1086/285075

KochGW,Sillett SC, JenningsGM,DavisSD (2004)The limits to tree height.Nature 428, 851–854. doi:10.1038/nature02417

Lynch J (1995) Root architecture and plant productivity. Plant Physiology109, 7–13.

McDonald PG, Fonseca CR, Overton JM, Westoby M (2003) Leaf-sizedivergence along rainfall and soil-nutrient gradients: is the method ofsize reduction common among clades? Functional Ecology 17, 50–57.doi:10.1046/j.1365-2435.2003.00698.x

McLeish M, Chapman T, Schwarz M (2007) Host-driven diversification ofgall-inducing Acacia thrips and the aridification of Australia. BMCBiology 5, 3. doi:10.1186/1741-7007-5-3

Meier I, Leuschner C (2008) Leaf size and leaf area index in Fagus sylvaticaforests: competing effects of precipitation, temperature, and nitrogenavailability. Ecosystems (New York, N.Y.) 11, 655–669.doi:10.1007/s10021-008-9135-2

Merchant A, Richter A, Popp M, Adams M (2006) Targeted metaboliteprofiling provides a functional link among eucalypt taxonomy,physiology and evolution. Phytochemistry 67, 402–408.doi:10.1016/j.phytochem.2005.11.027

Merchant A, Ladiges PY, AdamsMA (2007) Quercitol links the physiology,taxonomy and evolution of 279 eucalypt species. Global Ecology andBiogeography 16, 810–819. doi:10.1111/j.1466-8238.2007.00338.x

Miller JT, Andrew RA, Maslin BR (2002) Towards an understanding ofvariation in the Mulga complex (Acacia aneura and relatives).Conservation Science Western Australia 4, 19–35.

Mokany K, McMurtrie RE, Atwell BJ, Keith H (2003) Interaction betweensapwood and foliage area in alpine ash (Eucalyptus delegatensis) trees ofdifferent heights. Tree Physiology 23, 949–958.

Moore ID, Gessler PE, Nielsen GA, Peterson GA (1993) Soil attributeprediction using terrain analysis. Soil Science Society of AmericaJournal 57, 443–452. doi:10.2136/sssaj1993.03615995005700020026x

Niinemets Ü, Tenhunen JD (1997) A model separating leaf structural andphysiological effects on carbon gain along light gradients for the shade-tolerant speciesAcer saccharum.Plant,Cell&Environment20, 845–866.doi:10.1046/j.1365-3040.1997.d01-133.x

NiinemetsÜ,KullO, Tenhunen JD (1999)Variability in leafmorphology andchemical composition as a function of canopy light environment incoexisting deciduous trees. International Journal of Plant Sciences160, 837–848. doi:10.1086/314180

Niinemets Ü, Portsmuth A, Tena D, Tobias M, Matesanz S, Valladares F(2007a) Do we underestimate the importance of leaf size in planteconomics? Disproportional scaling of support costs within thespectrum of leaf physiognomy. Annals of Botany 100, 283–303.doi:10.1093/aob/mcm107

NiinemetsÜ, PortsmuthA,TobiasM (2007b) Leaf shape andvenationpatternalter the support investments within leaf lamina in temperate species: aneglected source of leaf physiological differentiation? FunctionalEcology 21, 28–40. doi:10.1111/j.1365-2435.2006.01221.x

Noble IR, Slatyer RO (1980) The use of vital attributes to predict successionalchanges in plant communities subject to recurrent disturbances. PlantEcology 43, 5–21. doi:10.1007/BF00121013

North GB,Nobel PS (1998)Water uptake and structural plasticity along rootsof adesert succulentduringprolongeddrought.Plant,Cell&Environment21, 705–713. doi:10.1046/j.1365-3040.1998.00317.x

Noy-Meir I (1973) Desert ecosystems: environment and producers. AnnualReview of Ecology and Systematics 4, 25–51.doi:10.1146/annurev.es.04.110173.000325

Ordoñez JC,BodegomPMv,Witte J-PM,Wright IJ,ReichPB,AertsR (2009)A global study of relationships between leaf traits, climate and soilmeasures of nutrient fertility. Global Ecology and Biogeography 18,137–149. doi:10.1111/j.1466-8238.2008.00441.x

Pedley L (1973) Taxonomy of the Acacia aneura complex. TropicalGrasslands 7, 3–8.

Pedley L (2001) Acacia aneura and relatives. In ‘Flora of Australia 11B,Mimosaceae, Acacia part 1. Vol. 11B’. (Eds AE Orchard, AJG Wilson)pp. 309–328. (CSIROPublishing/AustralianBiological ResourcesStudy:Canberra)

Pickup M, Westoby M, Basden A (2005) Dry mass costs of deploying leafarea in relation to leaf size. Functional Ecology 19, 88–97.doi:10.1111/j.0269-8463.2005.00927.x

Poorter H, Niinemets Ü, Poorter L, Wright IJ, Villar R (2009) Causes andconsequences of variation in leaf mass per area (LMA): a meta-analysis.New Phytologist 182, 565–588. doi:10.1111/j.1469-8137.2009.02830.x

Prior LD, Eamus D, Bowman DMJS (2003) Leaf attributes in the seasonallydry tropics: a comparisonof four habitats innorthernAustralia.FunctionalEcology 17, 504–515. doi:10.1046/j.1365-2435.2003.00761.x

Prior LD, Bowman DMJS, Eamus D (2005) Intra-specific variation in leafattributes of four savanna tree species across a rainfall gradient in tropicalAustralia. Australian Journal of Botany 53, 323–335.doi:10.1071/BT04080

R Development Core Team (2010) ‘R: a language and environment forstatistical computing.’ (R Foundation for Statistical Computing: Vienna)

Randell BR (1992) Key to the varieties of A. aneura. Journal of the AdelaideBotanic Gardens 14, 120–121.

Reich PB, Walters MB, Ellsworth DS (1997) From tropics to tundra: globalconvergence inplant functioning.Proceedingsof theNationalAcademyofSciences of the United States of America 94, 13 730–13 734.doi:10.1073/pnas.94.25.13730

ReichPB,EllsworthDS,WaltersMB (1998)Leaf structure (specific leaf area)modulates photosynthesis-nitrogen relations: evidence from within andacross species and functional groups. Functional Ecology 12, 948–958.doi:10.1046/j.1365-2435.1998.00274.x

SchenkHJ, Espino S, Goedhart CM,NordenstahlM, Cabrera HIM, Jones CS(2008) Hydraulic integration and shrub growth form linked acrosscontinental aridity gradients. Proceedings of the National Academy ofSciences of the United States of America 105, 11 248–11 253.doi:10.1073/pnas.0804294105

Shipley B (1999) Testing causal explanations in organismal biology:causation, correlation and structural equation modelling. Oikos 86,374–382. doi:10.2307/3546455

Sperry JS (2000) Hydraulic constraints on plant gas exchange. Agriculturaland Forest Meteorology 104, 13–23.doi:10.1016/S0168-1923(00)00144-1

Stafford Smith DM, Morton SR (1990) A framework for the ecology of aridAustralia. Journal of Arid Environments 18, 255–278.

Tyree MT, Ewers FW (1991) Tansley Review No. 34. The hydraulicarchitecture of trees and other woody plants. New Phytologist 119,345–360. doi:10.1111/j.1469-8137.1991.tb00035.x

van DamO (2000) ‘Modelling incoming potential radiation on a land surfacewith PCRaster: POTRAD5.MOD manual.’ (Utrecht Centre forEnvironment and Landscape Dynamics, Utrecht University: Utrecht,The Netherlands)

Vriend SP, van Gaans PFM,Middelburg J, De Nijs A (1988) The applicationof fuzzy c-means cluster analysis and non-linear mapping to geochemicaldatasets: examples from Portugal. Applied Geochemistry 3, 213–224.doi:10.1016/0883-2927(88)90009-1

Warwick NWM, Thukten (2006) Water relations of phyllodinous and non-phyllodinous Acacias, with particular reference to osmotic adjustment.Physiologia Plantarum 127, 393–403.doi:10.1111/j.1399-3054.2006.00663.x

Weiher E, van derWerf A, Thompson K, RoderickM, Garnier E, Eriksson O(1999) Challenging Theophrastus: a common core list of plant traits forfunctional ecology. Journal of Vegetation Science 10, 609–620.doi:10.2307/3237076

Westoby M, Wright IJ (2006) Land-plant ecology on the basis of functionaltraits. Trends in Ecology & Evolution 21, 261–268.doi:10.1016/j.tree.2006.02.004

Williams OB, Calaby JH (1985) The hot deserts of Australia. In ‘Ecosystemsof theworld 12A: hot deserts and arid shrublands’. (EdsMEvenari, I Noy-Meir, DW Goodall) pp. 269–312. (Elsevier: Amsterdam)

Wright IJ, GroomPK, Lamont BB, Poot P, Prior LD, Reich PB, Schulze E-D,Veneklaas EJ, Westoby M (2004a) Leaf trait relationships in Australianplant species. Functional Plant Biology 31, 551–558.doi:10.1071/FP03212

Wright IJ, Reich PB, Diemer M, Flexas J, Garnier E, Groom PK, Gulias J,HikosakaK, Lamont BB, Lee T, LeeW, Lusk C,Midgley JJ, NavasM-L,NiinemetsÜ,Oleksyn J,OsadaN, PoorterH, Poot P, Prior L, PyankovVI,RoumetC,ThomasSC,TjoelkerMG,VeneklaasEJ,VillarR (2004b) Theworldwide leaf economics spectrum. Nature 428, 821–827.doi:10.1038/nature02403

Manuscript received 17 February 2011, accepted 6 June 2011

Appendix 1. Modelled landscape attributes by landscape positionMean values are bolded and standard errors are in parentheses

Landscape position Elevation(m)

Length/slopeFactor

Slope(degrees)

Incident solar radiation(MJm–2 day–1)

Soil wetnessindex

Drainage line 692 (0.36) 27.0 (0.11) 4.62 (0.02) 6.81 e3(1.02) 12.0 (0.008)Valley 747 (0.21) 8.85 (0.07) 4.12 (0.02) 6.86 e3(0.72) 10.9 (0.004)Lower southern slope 822 (0.62) 36.9 (0.10) 19.8 (0.03) 6.14 e3(1.75) 8.81 (0.003)Upper southern slope 921 (0.48) 63.5 (0.10) 30.1 (0.03) 5.52 e3(2.09) 8.61 (0.003)

endix2.

Meanvalues

ofbran

chingattributes

foreach

thform

estAng

Meanvalues

arebolded,and

standard

errorsaregiveninparentheses.P-valuesdenotesignificanceforo

ayANOVAtestsford

ifferences

amongform

sforeachattribute.Pairw

isecomparisons

notshown

Growth

sShrubs

Multi-stem

Single-stem

P-value

individu

Basaldiam

137(21.7)

127(8.76)

116(5.56)

215(46.3)

196(4.19)

126(5.60)

92(2.57)

214(8.63)

137(5.51)

87(15.03

(0.80)

(0.70)

(0.13)

(2.12)

(0.06)

(0.02)

(0.01)

(0.00)

<0.001

Diameter

ofstem

ground

(9.43)

(5.38)

(3.74)

153(32.4)

139(2.96)

102(5.10)

(2.01)

175(7.5)

105(3.22)

(11.09

firststem

division

(0.03)

(0.13)

(0.05)

(0.02)

(0.09)

(0.10)

(0.16)

(0.19)

(0.34)

<0.001

nbreak(m

(0.15)

(0.13)

(0.10)

(0.22)

(0.13)

(0.11)

(0.14)

(0.36)

(0.25)

(0.23)

<0.001

(1.76)

(1.14)

(0.68)

(8.59)

(0.72)

(0.82)

(0.61)

(1.79)

(0.62)

(2.46)

<0.001

Treeheight

(0.34)

(0.18)

(0.16)

(0.56)

(0.11)

(0.13)

(0.34)

(0.24)

(0.68)

<0.001

Appendix 3. Relative dominance (%) of growth forms in each landscape position.The dominant growth formT4C is representedbyblackdots, boldeddiagonal lines represent single-stemmed tall trees, fine diagonal lines represent single-stemmed small trees, and horizontal and vertical lines represent

all other multi-stemmed forms (S1, S2, T1, T2)

Draina

line Lo

woodla

Appendix 4. Relative abundance of each growth form in each landscape position for four species of mulga complex at West Angelas including threephyllode variants of Acacia aneura. Diagonal lines represent single-stemmed growth forms, dots represent the multi-stemmed tall tree ‘T4C’ and

horizontal and vertical lines represent all other multi-stemmed growth forms

A. aneura(terete)

A. aneura(narrow)

A. aneura(broad)

Drainageline

Lowwoodland

Bandedwoodland

Low openwoodland

Lower hillslope

Upper hillslope

A. catenulata

http://www.publish.csiro.au/journals/ajb

Inter- and intra-specific variation in phyllode size and growth form among closely related...

Documents

Transcript of Inter- and intra-specific variation in phyllode size and growth form among closely related...

COLOR DIFFERENCES AMONG CLOSELY RELATED SPECIES OF RED-BREASTED MEADOWLARKS (STURNELLA

Rheological properties and surface tension of Acacia tortuosa gum exudate aqueous dispersions

Evolution in closely adjacent plant populations

Frequency Distribution of Extreme Hydrologic Drought of Southeastern Semiarid Region, Iran

Heat Shock Responses of Closely Related Species of Tropical and Desert Fish

Surface water-groundwater interactions in semiarid drainages ...

Distinct male reproductive strategies in two closely related oak species

Sediment loss in semiarid small watershed due to the land use

A probabilistic approach for predicting rainfall soil erosion losses in semiarid areas

Reproductive strategies in two closely related stony corals (Agaricia, Scleractinia)

Macrophytic, epipelic and epilithic primary production in a semiarid Mediterranean stream

natural dyes from acacia

Five New Species of the Botryosphaeriaceae from Acacia Karroo in South Africa

Evidence of serine-protease activity closely associated with Drosophila alcohol dehydrogenase

Triterpenoids from Acacia ataxacantha DC - Archive ouverte ...

NITROGEN DYNAMICS IN AN AUSTRALIAN SEMIARID GRASSLAND SOIL

Soil clay influences Acacia encroachment in a South African grassland

Divergent MicroRNA targetomes of closely related circulating strains of a polyomavirus

Moisture Controls on Trace Gas Fluxes in Semiarid Riparian Soils

Multi-objective optimization for sustainable groundwater resource management in a semiarid catchment