Petiole length and biomass investment in support modify light interception efficiency in dense poplar plantations

Page 1

Summary Leaf architecture, stand leaf area index and can-opy light interception were studied in 13 poplar clones growingin a second rotation of a coppice plantation, to determine therole of leaf architectural attributes on canopy light-harvestingefficiency and to assess biomass investment in leaf support tis-sue. Stand leaf area index (L) varied from 2.89 to 6.99, but Lwas only weakly associated with canopy transmittance (TC).The weak relationship between TC and L was a result of a higherdegree of foliage aggregation at larger values of L, leading tolower light-interception efficiency in stands with greater totalleaf area. We observed a strong increase in leaf aggregation anda decrease in light-harvesting efficiency with decreasing meanleaf petiole length (PL) but not with leaf size, possibly because,in cordate or deltoid poplar leaves, most of the leaf area is lo-cated close to the petiole attachment to the lamina. Although PL

was the key leaf characteristic of light-harvesting efficiency,clones with longer petioles had larger biomass investments inpetioles, and there was a negative relationship between PL andL, demonstrating that enhanced light harvesting may lead to anoverall decline in photosynthesizing leaf surface. Upper-can-opy leaves were generally larger and had greater dry mass (MA)and nitrogen per unit area (NA) than lower-canopy leaves. Can-opy plasticity in MA and NA was higher in clones with higher fo-liar biomass investment in midrib, and lower in clones withrelatively longer petioles. These relationships suggest thatthere is a trade-off between photosynthetic plasticity and bio-mass investment in support, and also that high light-harvestingefficiency may be associated with lower photosynthetic plas-ticity. Our results demonstrate important clonal differences inleaf aggregation that are linked to leaf structure and biomass al-location patterns within the leaf.

Keywords: interclonal variability, leaf area index, leaf clump-ing, morphological plasticity, support costs.

Introduction

Forest stands, especially conifer stands, may support very highleaf area indices (leaf area per unit ground area, L) on the orderof 10–20 (Jarvis and Leverenz 1983, Heilman and Fu-Guang1994, Chen et al. 1997, Lüttge 1997). Even such dense standsgenerally have canopy transmittances of 1–2% or more (Jarvisand Leverenz 1983, Chen et al. 1997). Although the dispersionof foliage elements was considered random in earlier canopylight-interception models (Ross 1981), such transmittancesare theoretically impossible for a spatially random dispersionof foliar elements for L values above about 5. In fact, in mostplant canopies, phytoelements are often strongly aggregated(Chen et al. 1997, Cescatti 1998, Kucharik et al. 1999, Lacazeet al. 2002). In aggregated canopies, canopy transmittance is afunction of effective leaf area index, Le, which is equivalent toa hypothetical leaf area index for a random dispersion of foliarelements (Chen and Cihlar 1995b, Chen et al. 1997):

LLe��0

(1)

where �0 is the clumping index that decreases with increasingleaf aggregation.

Clumping in stands of broad-leaved species is generallyconsidered to be less important than in conifer stands, and isthought to be primarily a function of crown shape and branch-ing architecture (Chen and Cihlar 1995a, 1995b, Asner et al.1998, Kucharik et al. 1999). However, the spatial pattern offoliage may strongly affect the radiative regime and, conse-quently, canopy photosynthesis in broadleaf species (Bal-docchi and Hutchison 1986). Simulation studies indicate thatleaf characteristics, such as leaf size and petiole length, alterleaf light-interception efficiency and clumping at the shootscale (Niklas 1988, 1992, Takenaka 1994, Pearcy and Yang1998). Apart from leaf structure effects on shoot morphology,numerous experimental studies have demonstrated that leaf

Tree Physiology 24, 141–154© 2004 Heron Publishing—Victoria, Canada

Petiole length and biomass investment in support modify light-interception efficiency in dense poplar plantations

ÜLO NIINEMETS,1–3 NAJWA AL AFAS,4 ALESSANDRO CESCATTI,2 AN PELLIS4 andREINHART CEULEMANS4

1 Department of Plant Physiology, Institute of Molecular and Cell Biology, University of Tartu, Riia 23, EE 51010 Tartu, Estonia

2 Centro di Ecologia Alpina, I-38040 Viote del Monte Bondone, Trento, Italy

3 Author to whom correspondence should be addressed ([email protected])

4 University of Antwerp, Department of Biology, Universiteitsplein 1, B-2610 Wilrijk, Belgium

Received April 8, 2003; accepted June 15, 2003; published online December 15, 2003

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 2

shape and angle are also important determinants of light-inter-ception efficiency (e.g., Valladares and Pearcy 1999, Fleck etal. 2003). Leaf architecture plays a major role in the light-har-vesting efficiency of herb canopies (Pearcy and Yang 1998,Valladares et al. 2002b), but leaf-scale structural measure-ments have rarely been scaled up to dense broadleaf tree cano-pies (Farque et al. 2001), and the relative significance ofspecific foliar characteristics for stand light-use efficiency hasnot been characterized.

Difficulties in assessing the impact of foliar structure onstand light-harvesting efficiency partly result from the circum-stance that even monotypic stands generally consist of geneti-cally different individuals with important variations in crownshape and leaf characteristics. Such genetic differences in nat-ural stands may be effectively reduced by studying clonalplants, in which all individuals in the stand possess essentiallythe same crown architectures. In particular, crown architecturehas been intensively investigated in clonal poplar stands(Isebrands and Nelson 1982, Weber et al. 1985, Ceulemans etal. 1990, Dickmann et al. 2001). Different poplar clones pos-sess leaves with widely varying shape and size on petioles ofcontrasting length, and may form extremely dense canopieswith L values on the order of 10–17 (Zavitkovski 1982,Heilman and Fu-Guang 1994, Heilman et al. 1996, Dickmannet al. 2001). Thus, clonal poplar plantations provide a uniquesystem to investigate the role of foliar structural attributeswithout interfering genetic variations. According to previouswork, there are large clonal differences in foliar light-intercep-tion efficiency after the stand-to-stand variations in L havebeen accounted for. These canopy differences have been re-lated to clonal variations (Weber et al. 1985) in leaf angulardistributions (e.g., Isebrands and Michael 1986, Ceulemansand Isebrands 1996, Heilman et al. 1996). However, even at acommon leaf angular distribution, there may be major differ-ences in stand light interception as a result of modifications infoliage spatial aggregation.

Our main study objectives were to determine total and effec-tive L and leaf clumping, and to evaluate the significance ofvariations in petiole length, leaf size and shape on standlight-harvesting efficiency in dense clonal poplar plantations.Specifically, we hypothesize that, in stands with similar L,clones with longer petioles and larger leaves intercept lightmore efficiently than clones with shorter petioles and smallerleaves.

Because increased light-interception efficiency may bringabout enhanced foliar support investments in petioles andmidrib that may lead to an overall reduction in total leaf area(Givnish 1986), we also characterized clonal variations in bio-mass investments in petioles and midribs. We hypothesize thatthere is a trade-off between biomass investment in support andtotal stand leaf area formed.

We also studied canopy variation in leaf dry mass per unitarea (MA) that affects both photosynthetic capacity and theamount of leaf area constructed per unit biomass investment inleaves. In addition to canopy light-harvesting efficiency, previ-ous work has highlighted important interclonal differences in

plastic adjustments in MA and photosynthetic capacity to can-opy light gradient (Casella and Ceulemans 2002). Given thatleaves exposed to higher irradiance require greater petiolarand midrib capacities for water transport and assimilateretranslocation (Vogelmann et al. 1982, Zwieniecki et al.2000), as well as for mechanical support of heavier laminas(Niinemets and Fleck 2002b), we also hypothesize that inter-clonal plasticity in adjustments in MA to irradiance gradient ispositively related to foliar biomass investments in support bio-mass. Thus, clonal differences in support investments maydrive important interactions between canopy variation in MA

and total stand L.

Materials and methods

Layout of the poplar plantation

The study was conducted in 2002, which was the second yearin the second rotation of a coppiced poplar plantation. The0.56-ha study site was located in Boom near the river Rupel (atributary of the river Schelde), province of Antwerpen, Bel-gium (51°05� N, 04°22� E; elevation 5 m). Hardwood cuttingswith a length of 25 cm were planted in April 1996 in a dou-ble-row design with alternating inter-row distances of 0.75 and1.5 m, and a spacing of 0.9 m within rows, resulting in an over-all density of 10,000 trees ha–1. A randomized block designwith 17 clones × 3 replicates was used according to the proto-col suggested by the British Forestry Commission (Armstrong1997). Individual monoclonal plots were 9 × 11.5 m in size,containing 10 rows of 10 stools each. In every plot, 36 (i.e., 6 ×6) assessment stools (individual plant with all its shootsformed after coppicing) were marked in the plot center, result-ing in sub-plots of 5.4 × 6.75 m. The remaining 64 stools madeup two border rows surrounding the assessment trees (Zavit-kovski 1981).

The cuttings that failed to establish in the year of planting,1996, were replaced in the spring of 1997 with new 25-cm-long hardwood cuttings for most clones, but with 40 cm cut-tings for the clones with a mortality rate higher than 10%. InDecember 1996 at the end of the establishment year, as well asin January 2001 after the first rotation cycle of 4 years, all treeswere cut back to a height of 5 cm to create a coppice systemwith a mean number of two to 10 resprouting shoots per stool.The plantation was not fertilized or irrigated after starting theexperiment. Herbaceous weeds were regularly cut back with atrimmer during the 1996 and 1997 growing seasons. In June1996, June 1997 and May 2001, herbicides (a mixture of3.2 kg ha–1 glyphosate and 9.0 kg ha–1 oxidiazon) were appliedwith a hood-covered spray nozzle that minimized the impacton trees. Further details of the site and management history areprovided in Deraedt and Ceulemans (1998), Casella andCeulemans (2002), and Laureysens et al. (2000a, 2000b,2003a).

Poplar clones and study plots

The clonal plant materials were supplied by the Institute forForestry and Game Management (Geraardsbergen, Belgium)

142 NIINEMETS ET AL.

TREE PHYSIOLOGY VOLUME 24, 2004

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 3

and the Forest Research Agency, Forestry Commission (AliceHolt, U.K.), and had all been selected for superior biomassproduction and disease resistance. As some clones did not per-form well in the second rotation, partly because of heavy fun-gal rust (Melampsora) infection (Laureysens et al. 2003a,2003b), and had very open and untypical canopies, we selecteda sub-set of 13 clones for the current study. The studied clonesbelong to the following species and interspecific hybrids:Populus nigra L. (‘Wolterson’), P. trichocarpa Torr. & Gray(‘Columbia River’, ‘Fritzi Pauley’, ‘Trichobel’), P. tricho-carpa × P. deltoides Bart. ex Marsh. (‘Beaupré’, ‘Hazendans’,‘Raspalje’, ‘Unal’), P. deltoides × P. trichocarpa (‘IBW2’),P. deltoides × P. nigra (‘Gaver’, ‘Gibecq’, ‘Primo’) and P. tri-chocarpa × P. balsamifera (‘Balsam Spire’). A detailed de-scription of the clones is provided by Barnéoud et al. (1982)and Laureysens et al. (2003a, 2003b).

The primary selection criterion for plots/clones for our anal-ysis was a requirement to achieve a range of canopy leaf areaindices, different foliage aggregations, and leaf sizes and toavoid large gaps in the canopy caused by nonuniform shootmortality. Most clones satisfying these criteria were sampledfrom one randomly selected plot. ‘Wolterson’, ‘Fritzi Pauley’,and ‘Balsam Spire’ were studied in two plots, and ‘Gaver’ inthree plots. At the time of sampling, all stands essentially con-sisted of unbranched vertical shoots allowing us to study theinfluence of differences in leaf structural characteristics onstand-level light interception independently of branching.

Characterization of leaf morphology and nitrogen content

In June 2002, 10–15 fully developed leaves were randomlycollected from different shoots within the center of the assess-ment plots from the lower (0.5–1 m from the ground) andupper canopy (top meter of the canopy). The leaves were en-closed in plastic bags with moist filter paper, brought to thelaboratory within 2 h of collection, and kept at 2 °C until anal-ysis (at most for 7 days following sampling). Only leaves ofthe same age were sampled from the upper and lower canopy,and therefore, the light availability in the lower canopy de-creased somewhat from bud burst until the full expansion ofthe upper canopy.

Lamina length and largest width, and petiole length weremeasured for each leaf. Leaf area (AL) was measured with anLI-3000 leaf area meter (Li-Cor, Lincoln, NE). For five leavesfrom both canopy levels, we separated the leaf lamina from themidrib with a razor. Fresh mass of all leaf fractions (lamina,midrib, petiole) was measured immediately. Leaf fractionswere dried at 75 °C for at least 48 h, and their dry mass was de-termined and used to calculate lamina (without midrib) drymass per area (MA), and fractions of petiole and midrib in totalleaf dry mass.

Mass-based leaf lamina nitrogen (NM) concentrations weremeasured by gas chromatography after flash-combustion ofthe samples in oxygen in an NC 2100 elemental analyzer(Carlo Erba, Milan, Italy).

Direct measurements of canopy leaf area and stem areaindex

For each plot, five shoots of varying size were harvested fordestructive leaf area and stem area determinations. For eachshoot, stem diameter at 0.22 m from the ground (DB) and atstem tip were determined with digital precision calipers(Model CD-15DC, Mitutoyo, Hampshire, U.K.), stem lengthwith a ruler and total leaf area (AT) with a portable laser areameter (Model CI-203, CID, Seller, USA). Cross-calibrationsof Li-Cor and CID area meters demonstrated that both instru-ments provided similar and not systematically different leafarea values within 3% of average. Stem area (AS) was deter-mined from measured stem diameters and length approximat-ing the stem shape by a frustum of a cone.

Measurements of DB, AT and AS were used to developallometric regressions in the form of AT = aDB

b and AS = cDBd,

where a–d are the regression coefficients. Because the plot-to-plot differences in allometric relations were small (Pellis2002), all measurements for a specific clone were pooled.Thus, the number of destructively harvested shoots for eachclone was 15 (Figure 1).

All shoots in the assessment plot were measured for stem di-ameter at a height of 0.22 m from the ground with digital preci-sion calipers in August 2001 and January 2003 (correspondingto the August–September 2002 leaf area index). These mea-surements were used to calculate stand leaf area index (L) andstem area index (S, stem area per unit ground area) based onallometric regressions on two occasions. The two estimateswere further employed to determine L and S at the time of sam-pling of the foliage for morphological measurements and forthe assessment of canopy light environment based on the dataof canopy phenological development and stand growth mea-sured at the same site (Casella and Ceulemans 2002, Pellis2002). Overall, these data suggested that a linear average ofthe estimates obtained at various dates was appropriate in most

TREE PHYSIOLOGY ONLINE at http://heronpublishing.com

LEAF AGGREGATION AND LIGHT INTERCEPTION IN POPLAR 143

Figure 1. Sample allometric relationships between shoot diameter andtotal shoot leaf area for two poplar clones of contrasting branching ar-chitecture, leaf number per unit shoot length and leaf size (Table 1).Filled symbols, solid line = P. nigra ‘Wolterson’; open symbols, bro-ken line = P. trichocarpa × P. balsamifera ‘Balsam Spire’.

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 4

cases. Given that estimates of leaf and stem area index werestrongly correlated on different dates (r2 = 0.73, P < 0.001 forL, and r2 = 0.79, P < 0.001 for S), it is unlikely that interpola-tion of L and S measurements to sampling dates biased ourconclusions with respect to the statistical significance of therelationships.

Determination of canopy transmittance, effective leaf areaindex and leaf clumping

On completely overcast days, four to five hemispherical pho-tographs were taken at 0.4–0.5 m above the ground in the cen-ter of the assessment plots with a Nikon Coolpix 990 digitalcamera equipped with a Nikon Fisheye Converter FC-E80.21X. From the hemispherical images, canopy gap fractions,T(�), for different zenith angles, �, were determined with anangular resolution of 1° using WinPhot 5.0 software (terSteege 1996). These values allowed us to estimate canopytransmittance (TC), and mean irradiance on the leaf surface(IM) as (1 – TC)/L, where L is the allometrically determinedstand leaf area index.

The gap fraction data were further used to calculate canopytransmittance in angular bands of 15°, and effective leaf areaindex was estimated from these values according to the inver-sion procedure developed by Miller (1967) and Campbell andNorman (1989), and as implemented in the Li-Cor plant can-opy analyzer (Li-Cor 1992):

L T de � ��20

2

ln( ( )) cos sin/

� � � ��

(2)

To avoid edge effects, gap fractions in the outermost 15° bandwere not employed in the inversion, as suggested by Ceule-mans et al. (1993), and the gap fraction distribution for the out-ermost band was assumed to correspond to that in the 15–30°band. Comparative studies have demonstrated that Le estima-tion based on hemispherical photographs provides essentiallythe same values as the Li-Cor plant canopy analyzer (Chen etal. 1997).

Effective leaf area index characterizes the fraction of lightintercepted for a hypothetical random dispersion of foliage el-ements. The ratio between effective leaf area index (Le) andactual leaf area index obtained from allometric relations (L) isa measure of spatial aggregation of phytoelements in the can-opy space (�0). From the canopy gap fraction measurementsand for each canopy photograph, we determined Le, and calcu-lated the clumping index, �0 (Equation 1). Mean values of Le,�0 and IM were determined for each plot.

Effective leaf area index calculated by Equation 2 also in-cludes the contribution of woody biomass in light interception.Chen et al. (1997) suggested a correction to account for stemarea index (S) as:

LL

S

S L

e ��

�

�0

1(3)

The basic assumptions behind this correction are that the an-gular distributions of stem area and leaf area are identical, thatthe clumping index is the same for stems and leaves, and thatthe spatial distributions of stems and leaves are independent,i.e., the stem area is not preferentially shaded by leaves. How-ever, in our study, S was primarily determined by erectunbranched shoots of poplar coppice that were also heavilyshaded by leaves such that all these assumptions were violated.Therefore, we did not correct Le values for light interceptionby woody biomass in our study. Furthermore, the estimates ofstem area index and leaf area index obtained with the biomet-ric models were strongly correlated (r2 = 0.71, P < 0.001), andthe correction factor, 1 – S/(S + L), was relatively small (range0.89–0.92 for different clones, mean ± SE = 0.915 ± 0.004).Given this limited variability, we conclude that the presence ofstem area in our estimations of effective leaf area does not biasour conclusions qualitatively.

Indirect estimation of leaf angular distribution

The inversion algorithm for Le determination is unaffected bythe angular distribution of leaf inclination angles (Campbelland Norman 1998), but canopy light climate depends heavilyon leaf inclination as well:

T e eG L G L( ) ( ) / ( ) /� � � � � �cos cose� �� �0 (4)

where G(�) is the canopy extinction coefficient that is deter-mined by angular distribution of leaf surface (Nilson 1971). Tocharacterize G(�), we calculated the mean leaf tilt angle (M)according to the algorithm of Lang (1986, Li-Cor 1992). Thisalgorithm uses gap fraction data to find an average angle thatbest matches the observations. Implicit in this algorithm is thatall leaves have a fixed angle M, i.e., the angular distribution ofleaf surface is conical. Another assumption in this model isthat �0 is independent of �.

To simulate canopy light interception capacity morerealistically and to compare the effect of various leaf inclina-tion angle distributions on canopy transmittance, we also useda hypothetical M value of 0° (horizontal inclination), M = 90°(vertical inclination) as well as spherical and ellipsoidal leafangle distributions (Campbell 1986, Campbell and Norman1989). The leaves have no preferable orientation if the distri-bution of leaf areas is spherical. The ellipsoidal distribution as-sumes that leaf surfaces are distributed parallel to the surfaceof an oblate or prolate spheroid:

Gc

Bc( )

cos sin�

� ��

�2 2 2

(5)

where the parameter c is the ratio of ellipsoid horizontal to ver-tical semiaxis (c > 1 for an oblate spheroid, and c < 1 for aprolate spheroid), and B is dependent on c as:

144 NIINEMETS ET AL.

TREE PHYSIOLOGY VOLUME 24, 2004

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 5

Bc

c

cc

�

�� �

� ��

11 1

21

1 1

1 1

12

12

2

ln(( ) / ( )),

sin,

� �

�

�

�

���

���

(6)

where �121� � �c and �2

21� � c . We determined a value ofc that resulted in a distribution having a mean angle equal to M.

Canopy gradients in foliar characteristics

We calculated the canopy gradient (�V) in leaf dry mass perunit area (MA), nitrogen content per area (NA) and mean leafarea (AL) per unit intercepted relative irradiance (IA) to de-scribe the light-related plasticity in these foliar variables:

�VV V

I�

�U L

A

(7)

where VU and VL are values of the foliar characteristic for theupper and lower canopy, respectively. Provided the specificleaf variable varies linearly with irradiance, �V is equivalent tothe slope of the relationship between this characteristic andirradiance. Even if MA, NA and AL vary curvilinearly withirradiance on some occasions, �V provides a useful estimate oftotal light-dependent variability in these leaf characteristics.

Statistical analyses

Relationships between stand- and leaf-level attributes as wellas between foliar characteristics were analyzed by linear re-gressions. Because leaf area index and biomass production dif-fered greatly, up to 80% between some plots of the same clone(Pellis 2002, Laureysens et al. 2003a, 2003b), we investigatedthe relationships between canopy transmittance, effective andtotal leaf area index, clumping index and foliar characteristicsusing each plot as an experimental unit. In interclonal compar-isons of foliar architecture, data for the same clone werepooled, and subjected to one-way ANOVA separately for up-per- and lower-canopy leaves followed by the Bonferroni testto separate the means between clones. Although the lack ofreplication may mean that we did not obtain a true statisticalsignificance for the differences between the clones, the com-parisons provide plot-specific estimates that correspond torespective whole-canopy attributes. All relationships wereconsidered significant at P < 0.05 (Wilkinson 1990).

Results

Differences in foliar architecture between clones and withinthe canopy of different clones

For both upper- and lower-canopy leaves, means of foliage lin-ear measures—lamina length and width, and petiole length—varied 2- to 2.5-fold between different clones (Table 1). Meanleaf area varied 6-fold for both upper- and lower-canopyleaves and lamina dry mass per unit area (MA) varied 1.3- and1.7-fold for the lower- and upper-canopy leaves, respectively

TREE PHYSIOLOGY ONLINE at http://heronpublishing.com

LEAF AGGREGATION AND LIGHT INTERCEPTION IN POPLAR 145

Tabl

e1.

Mea

n±

SEva

lues

ofup

per-

and

low

er-c

anop

yfo

liar

mor

phol

ogic

alch

arac

teri

stic

sin

diff

eren

tPop

ulus

clon

es.1

Clo

ne2

Lam

ina

leng

th(c

m)

Lam

ina

wid

th(c

m)

Lam

ina

area

(cm

2 )L

amin

adr

ym

ass

per

area

(gm

–2 )

Petio

lele

ngth

(cm

)

Low

erU

pper

Low

erU

pper

Low

erU

pper

Low

erU

pper

Low

erU

pper

Wol

ters

on4.

54±

0.14

a7.

07±

0.24

a3.

24±

0.17

a6.

28±

0.21

a9.

15±

0.51

a27

.9±

2.2

a51

.1±

1.8

a81

.3±

2.6

abc

2.32

±0.

17a

5.11

±0.

12ab

cC

olum

bia

Riv

er7.

35±

0.25

bc13

.92

±0.

54b

3.99

±0.

21ab

9.12

±0.

53ab

19.0

±1.

6ab

c85

.8±

9.6

b43

.4±

1.4

ab79

.2±

2.4

abcd

1.85

±0.

18a

4.07

±0.

25a

Fritz

iPau

ley

9.86

±0.

3d

18.9

±1.

2c

4.92

±0.

17bc

11.3

3±

0.82

b35

.2±

1.9

cd15

4±

23c

44.5

±1.

9ab

103.

6±

3.7

e3.

57±

0.21

bc5.

07±

0.31

abT

rich

obel

8.45

±0.

38bd

e18

.9±

1.4

c5.

05±

0.28

bd11

.5±

1.1

b25

.9±

1.7

cde

166

±29

c49

.4±

6.2

ab91

.0±

5.4

bef

3.81

±0.

18bc

5.17

±0.

26ab

Bea

upré

7.97

±0.

45ef

12.3

0±

0.83

bd5.

76±

0.50

ce9.

16±

0.44

ab31

.3±

4.6

ce69

.6±

9.2

ab41

.2±

3.6

ab64

.7±

2.2

dgh

3.52

±0.

17bc

d6.

24±

0.36

bde

Haz

enda

ns11

.18

±0.

59d

7.31

±0.

50ae

7.22

±0.

54e

5.85

±0.

39a

54.0

±7.

0f

27.6

±2.

8a

48.3

±2.

1ab

85.8

±2.

6af

3.23

±0.

13bd

3.91

±0.

27a

Ras

palje

7.51

±0.

31bc

e12

.66

±0.

71bd

6.18

±0.

33de

9.98

±0.

73b

30.8

±2.

6bd

e89

±13

a45

.8±

0.9

ab84

.0±

2.2

af3.

69±

0.27

be5.

44±

0.20

bcU

nal

6.11

±0.

36bf

12.0

8±

0.72

bd3.

83±

0.19

acdf

8.53

±0.

48ab

16.1

±1.

6bd

66.1

±6.

7a

48.0

±3.

1ab

84.9

±3.

1cf

h3.

08±

0.21

bf6.

12±

0.27

cfIB

W2

7.70

±0.

56ce

12.9

4±

0.47

bd5.

22±

0.41

bce

10.2

0±

0.59

b24

.9±

3.6

cd82

.7±

9.5

a51

.2±

0.7

ab67

.2±

2.9

agh

3.78

±0.

25b

7.01

±0.

10e

Gav

er6.

16±

0.18

b8.

18±

0.21

a5.

25±

0.12

cde

7.28

±0.

19a

20.8

±1.

0bd

e37

.8±

2.1

a39

.7±

2.0

b62

.7±

2.2

gh2.

72±

0.08

df4.

74±

0.12

aG

ibec

q6.

02±

0.25

acg

8.92

±0.

40ad

f6.

02±

0.25

cd8.

55±

0.43

ab22

.1±

1.4

bde

48.9

±5.

1ad

48.6

±1.

2ab

62.4

±1.

5dg

h4.

36±

0.24

c5.

81±

0.21

cde

Prim

o8.

30±

0.44

ce15

.06

±0.

82bc

4.83

±0.

24be

11.3

1±

0.48

b25

.3±

2.5

bcd

114.

4±

9.5

bcd

38.9

±1.

6b

65.7

±0.

9ag

3.02

±0.

15bf

4.98

±0.

27ad

Bal

sam

Spir

e6.

03±

0.16

bg10

.93

±0.

68be

f3.

67±

0.11

af10

.90

±0.

76b

14.7

±0.

8ab

94±

12bd

44.7

±2.

4ab

76.5

±2.

7af

g1.

97±

0.12

ae4.

24±

0.16

af

1M

eans

foru

pper

-and

low

er-c

anop

yle

aves

wer

eco

mpa

red

with

one-

way

AN

OV

Aan

alys

esfo

llow

edby

Bon

ferr

onit

est.

Mea

nsin

each

colu

mn

follo

wed

byth

esa

me

lette

rare

nots

igni

fica

ntly

dif-

fere

nt(P

>0.

05).

For

each

clon

e,up

per-

and

low

er-c

anop

yfo

liar

char

acte

rist

ics

wer

eal

way

sdi

ffer

enta

tP<

0.01

,exc

eptf

orla

min

aw

idth

inth

ecl

one

Haz

enda

ns(P

>0.

08).

2T

hest

udie

dcl

ones

belo

ngto

the

follo

win

gsp

ecie

san

din

ters

peci

fic

hybr

ids:

Pop

ulus

nigr

a(c

lone

‘Wol

ters

on’)

,P.t

rich

ocar

pa(‘

Col

umbi

aR

iver

’,‘F

ritz

iPau

ley’

,‘T

rich

obel

’),P

.tri

choc

arpa

×P

.del

-to

ides

(‘B

eaup

ré’,

‘Haz

enda

ns’,

‘Ras

palje

’,‘U

nal’

),P

.del

toid

es×

P.t

rich

ocar

pa(‘

IBW

2’),

P.d

elto

ides

×P

.nig

ra(‘

Gav

er’,

‘Gib

ecq’

,‘Pr

imo’

)and

P.t

rich

ocar

pa×

P.b

alsa

mife

ra(‘

Bal

sam

Spir

e’).

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 6

(Table 1). Overall, variation between clones of the same parentspecies was smaller than among clones of different parentage(e.g., P. trichocarpa clones ‘Columbia River’, ‘Fritzi Pauley’and ‘Trichobel’ versus P. nigra clone ‘Wolterson’ in Table 1).

Upper-canopy leaves of most clones had longer and widerlaminas and longer petioles than lower-canopy leaves (Ta-ble 1). On average, laminas of upper-canopy leaves were 1.70± 0.16 (± SE, range 0.96–2.50) times longer and 1.96 ± 0.20(0.95–3.37) times wider, resulting in 3.9 ± 0.8 (0.9–10.3)times larger lamina surface area than in lower-canopy leaves.Petiole length increased on average 1.81 ± 0.16 (1.06–2.77),and MA 1.74 ± 0.08 (1.29–2.21) times from bottom to the topof the canopy (Table 1). The most responsive clone for leafarea changes was P. trichocarpa × P. balsamifera ‘BalsamSpire’, whereas P. trichocarpa × P. deltoides ‘Hazendans’ and‘Beaupré’ were the least responsive (Table 1). Petiole lengthchanged the most in canopies of P. nigra ‘Wolterson’and ‘Bal-sam Spire’, whereas the smallest response was observed for‘Hazendans’ and ‘Columbia River’ (Table 1).

Total and effective leaf area index, canopy transmittance andleaf clumping

Leaf area index (L) varied from 2.87 in P. deltoides × P. tri-chocarpa ‘IBW2’ to 6.99 in ‘Wolterson’, averaging (± SE) 5.44± 0.29 across all studied stands. Differences in stand L wereaccompanied by significant variations in canopy transmit-tance. The fraction of penetrating irradiance varied from 0.11(‘Fritzi Pauley’) to 0.32 (‘Beaupré’), averaging 0.176 ± 0.014.Effective leaf area index (Le, Equations 1 and 2) ranged from2.019 ± 0.027 (‘IBW2’) to 2.96 ± 0.13 (P. deltoides × P. nigra‘Gaver’), and averaged 2.49 ± 0.07.

Relatively low Le values indicate that the foliage elementswere not arranged randomly, but were strongly aggregated.Importantly, stands with the highest L values did not absorbthe largest fractions of irradiance, indicating that there are sig-nificant clonal differences in stand-level foliage clumping(Equations 1–3, Figure 2). There was only a weak relationshipbetween the fraction of intercepted irradiance and total L (Fig-ure 3A), and between the effective leaf area and total L (Fig-ure 3B). Canopies with the highest leaf area index tended to bethe most aggregated (Figure 2, Figure 3C), providing an expla-nation for the weak relationships of canopy transmittance andLe with L (Figures 3A and 3B).

We observed relatively high canopy transmittances and ex-tensive variation in penetrating irradiance (Figure 4A), butmean M varied from only 47.3° to 66.6° for various clones(mean ± SE = 55.1 ± 1.1°). As demonstrated by the simula-tions with different hypothetical leaf angular distributions, as-suming random spatial arrangement of foliar elements, andwith M (Figure 4A), such high canopy transmittances cannotbe explained without leaf clumping. The same result was alsoobtained using the ellipsoidal distribution of leaf angles de-rived from the mean leaf tilt angle (Equations 5 and 6).

The mean irradiance on leaf surface area decreased with in-creasing L, whereas leaf clumping led to significantly lower

146 NIINEMETS ET AL.

TREE PHYSIOLOGY VOLUME 24, 2004

Figure 2. Representative hemispherical photographs taken at the bot-tom of the canopies of five clones of similar effective leaf area index(Le, Equation 1) but differing total leaf area index (L), degree of leafclumping (coefficient of spatial clumping, �0, Equations 1 and 2) andvarying petiole length (PL) and leaf size (AL). Means (± SE) of PL andAL were calculated from two samples of 10–13 leaves taken from boththe upper and lower canopy. The clones were arranged according toincreasing clumping index: (A) Populus nigra ‘Wolterson’; (B) P. del-toides × P. nigra ‘Gaver’; (C) P. trichocarpa × P. balsamifera ‘BalsamSpire’; (D) P. trichocarpa × P. deltoides ‘Unal’; (E) P. deltoides ×P. nigra ‘Gibecq’.

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 7

mean leaf irradiances than a random leaf arrangement for allpossible leaf angular distributions (Figure 4B).

Overall, we observed a good correspondence between mea-sured canopy transmittance and estimates based on conicalleaf angle distributions obtained from mean leaf angle andmeasured values of leaf clumping (Figure 5). This indicatesthat, although both the inversion of canopy gap fractions andthe derivation of mean leaf tilt angle include several assump-tions, our approach provides a realistic description of canopyradiative transfer.

TREE PHYSIOLOGY ONLINE at http://heronpublishing.com

LEAF AGGREGATION AND LIGHT INTERCEPTION IN POPLAR 147

Figure 3. Correlations of the fraction of intercepted light (A), effectiveleaf area index (B, Equations 1 and 2) and clumping index (C, Equa-tion 1) with stand leaf area index for 16 clonal poplar stands. Datawere fitted with linear regressions. In A and B, regressions with (dot-ted line) and without (dashed line) the stands of P. nigra ‘Wolterson’,which exhibited the highest clumping (encircled data points), are de-picted. Error bars provide ± SE.

Figure 4. Measured (filled symbols) and simulated (lines and opensymbols) canopy transmittance (A) and the fraction of interceptedlight per unit leaf area index (B) in relation to stand leaf area index inthe studied poplar clones. The simulations were conducted for aspherical, horizontal and vertical leaf angular distribution, or usingthe ellipsoidal leaf angle distribution (Equations 5–6) derived frommeasured mean leaf tilt angles (Lang 1986), and assuming that therewas no clumping in the canopy. Fraction of intercepted light is givenas 1 – canopy transmittance. Second-order polynomial regressions fit-ted to the data were significant at P < 0.001 (r2 = 0.51 for A, and r2 =0.98 for B).

Figure 5. Predicted versus observed canopy transmittance for thestudied poplar clones with or without consideration of foliage aggre-gation and for ellipsoidal (Equations 5 and 6) and conical leaf inclina-tion angle distributions. Leaf angular distributions were derived frommean leaf tilt angle in all simulations.

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 8

Variation in canopy light-harvesting efficiency in relation toleaf characteristics

Apart from the strong correlation between �0 and L (Fig-ure 3C), �0 increased with increasing petiole length of both up-per- and lower-canopy leaves (Figure 6A). Thus, increasingthe distance of leaf lamina from the shoot axis led to lower ag-gregation and self-shading within the shoot. We also tested forcorrelations between �0 and individual leaf area (Figure 6B),

lamina length (r 2 = 0.09 for lower-canopy leaves and r2 = 0.07for upper-canopy leaves, P > 0.3 for both) and width (r2 = 0.25,P < 0.05 for lower-canopy leaves, and r2 = 0.17, P > 0.1 for up-per-canopy leaves) as well as between �0 and the ratio of peti-ole to total leaf length (r2 = 0.10, P > 0.2 for lower-canopyleaves, and r2 = 0.01, P > 0.9 for upper-canopy leaves), but theeffect of these characteristics on leaf aggregation was small.

There was a negative correlation between petiole length andstand leaf area index (Figure 7A), but light-interception effi-ciency increased significantly with increasing petiole length(Figure 7B), further underscoring the important role of peti-oles in modifying light interception between clones.

Biomass costs for efficient light harvesting

The negative correlation between petiole length and stand leafarea index (Figure 7A) suggests that formation of a large totalleaf area is constrained in clones with long petioles. In lower-canopy leaves, the percentage of total leaf dry mass in thepetiole varied from about 6% in ‘Hazendans’ to 11.5% inP. deltoides × P. nigra ‘Gibecq’. The corresponding variationfor upper-canopy leaves ranged from about 6.5% in ‘Tricho-bel’ to 11.5% in ‘Gibecq’ (Table 2). Fractional biomass invest-ment in petioles was weakly affected by canopy position(Table 1), but it increased with increasing petiole length (Fig-ure 8A) with differing allometries for upper and lower canopy(Figure 8A). These data demonstrate that increases in light-harvesting efficiency become progressively more expensivewith increasing petiole length.

Different poplar clones invested 4–7% of total leaf biomassin midrib (Table 2; 4.2–8.9% of lamina biomass). This per-centage was strongly correlated with leaf size (Figure 8B),demonstrating that increases in leaf size bring about enhancedinvestments in support. Although the canopy differences inbiomass investment in midrib were small (Table 2), lower-can-opy leaves of the same size invested more biomass in midribthan upper-canopy leaves (Figure 8B).

Biomass investments in midrib and petiole varied about1.7-fold between clones, and total biomass investment in sup-port (fraction of leaf biomass in petiole and midrib) varied1.3-fold for the entire data set, from about 12% in the lowercanopy of ‘Hazendans’ and upper canopy of ‘Balsam Spire’ toabout 16% in the lower canopy of ‘Trichobel’ and upper can-

148 NIINEMETS ET AL.

TREE PHYSIOLOGY VOLUME 24, 2004

Figure 6. Clumping index (�0,Equations 1 and 2) in relation topetiole length (A) and lamina area(B) of lower- (filled symbols) andupper-canopy (open symbols)leaves in the clonal poplar stands.Data are fitted by linear regressions,and only significant regressions areshown. Error bars give ± SE.

Figure 7. Correlations between (A) total stand leaf area index (L) and(B) the fraction of intercepted light per L with mean petiole length inlower- (filled symbols) and upper-canopy (open symbols) leaves ofstudied poplar clones. Fraction of intercepted light was found as 1 –canopy transmittance. Data were fitted by linear regressions. Errorbars give ± SE.

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 9

opy of ‘Unal’, ‘Beaupré’and ‘Gibecq’(Table 2). Total biomassinvestment in support (petiole + midrib) was positively corre-lated with petiole length (r2 = 0.40 for lower-canopy leavesand r2 = 0.47 for upper-canopy leaves, P < 0.01 for both), but itwas not associated with leaf area (r2 = 0.03 for lower canopyand r2 = 0.00 for upper canopy).

Trade-offs between support investments and canopyvariation in leaf dry mass per unit area and nitrogen

Leaf dry mass per unit area and AL were strongly modified bycanopy position (Table 1), suggesting that part of the inter-clonal variability in these characteristics may have resultedfrom differences in canopy transmittance of different clones.We observed positive correlations of MA and area-based leafnitrogen concentration (NA) with canopy transmittance forlower-canopy leaves (Figures 9A–B), partly corroborating thissuggestion. However, mean leaf area of lower-canopy leaveswas unrelated to canopy transmittance (r2 = 0.14, P > 0.1).Canopy differences in MA (Figure 9C), NA (r2 = 0.19, P > 0.08)and AL (r2 = 0.11, P > 0.2) were not associated with canopytransmittance, highlighting a strong clonal effect on the adjust-ment of foliar attributes to the light gradient from the top to thebottom of the canopy.

We observed that the plasticity (Equation 7) in MA, NA andAL was higher in clones with larger leaf biomass investment in

TREE PHYSIOLOGY ONLINE at http://heronpublishing.com

LEAF AGGREGATION AND LIGHT INTERCEPTION IN POPLAR 149

Tabl

e2.

Mea

n±

SEin

vest

men

tof

tota

llea

fdr

ym

ass

insu

ppor

tin

uppe

r-an

dlo

wer

-can

opy

leav

esin

the

stud

ied

Popu

lus

clon

es.1

Clo

ne2

Petio

le(%

)M

idri

b(%

)Pe

tiole

+m

idri

b(%

)

Low

erU

pper

Low

erU

pper

Low

erU

pper

Wol

ters

on8.

39±

0.28

acA

9.71

±0.

41ab

dB4.

81±

0.34

abA

3.87

±0.

16aB

13.2

1±

0.38

aA13

.58

±0.

46ab

cAC

olum

bia

Riv

er7.

28±

0.49

acA

7.46

±0.

33ac

dA6.

61±

0.71

bcA

5.24

±0.

17bc

A13

.9±

1.1

aA12

.70

±0.

22ab

cAFr

itziP

aule

y8.

48±

0.57

aA8.

52±

0.71

abcd

A6.

94±

0.39

cA6.

59±

0.39

bA15

.42

±0.

71aA

15.1

1±

0.43

acA

Tri

chob

el9.

64±

0.65

abA

6.56

±0.

82cB

6.89

±0.

40bc

dA5.

96±

0.56

bdA

16.5

3±

0.80

bA12

.5±

1.3

abB

Bea

upré

8.31

±0.

71ac

A10

.30

±0.

99ab

dA5.

96±

0.35

abcd

A5.

36±

0.38

bdeA

14.2

7±

0.39

abA

15.6

5±

0.66

acA

Haz

enda

ns5.

95±

0.23

cA10

.00

±0.

44ab

cdB

6.42

±0.

32bc

dA4.

42±

0.20

acdB

12.3

7±

0.53

aA14

.45

±0.

55ab

cBR

aspa

lje9.

62±

0.43

abA

8.62

±0.

51ab

cdA

5.35

±0.

37ab

cdA

4.80

±0.

16ac

dfA

14.9

7±

0.52

abA

13.4

2±

0.47

abcA

Una

l10

.10

±0.

22ab

A10

.90

±0.

57dA

4.69

±0.

20ad

A5.

26±

0.23

abA

14.7

6±

0.17

abA

16.1

6±

0.53

cBIB

W2

9.32

±0.

89ab

A10

.80

±0.

76dA

5.09

±0.

20ab

cdA

4.79

±0.

19ac

dfA

14.4

0±

0.86

abA

15.6

0±

0.94

acA

Gav

er8.

30±

0.29

aA8.

56±

0.25

abcd

A5.

23±

0.23

abcd

A4.

33±

0.14

aceB

13.5

3±

0.44

aA12

.89

±0.

31ab

AG

ibec

q11

.50

±0.

34bA

11.4

0±

1.3

dA3.

95±

0.35

adA

4.34

±0.

20ac

dA15

.46

±0.

65ab

A15

.8±

1.3

acA

Prim

o8.

81±

0.79

abA

7.94

±0.

48ab

cdA

6.95

±0.

18bc

A6.

08±

0.44

bfA

15.7

5±

0.91

abA

14.0

2±

0.79

abcA

Bal

sam

Spir

e7.

54±

0.32

acA

7.51

±0.

39ab

cA5.

73±

0.37

abcd

A4.

79±

0.18

acdf

A13

.27

±0.

33aA

12.3

0±

0.40

bA

1M

eans

ford

iffe

rent

cano

pypo

sitio

nsw

ere

com

pare

dw

ithon

e-w

ayA

NO

VA

follo

wed

byB

onfe

rron

ites

t.M

eans

inea

chco

lum

nw

ithth

esa

me

smal

llet

tera

reno

tsig

nifi

cant

lydi

ffer

ent.

Upp

erca

sele

tter

deno

tes

diff

eren

ces

betw

een

the

cano

pypo

sitio

nsfo

rea

chch

arac

teri

stic

and

clon

e.2

Clo

nepa

rent

ages

asin

Tab

le1.

Figure 8. Fractional leaf biomass investment in petiole versus petiolelength (A) and the fractional investment in leaf midrib versus laminaarea (B) in lower- (filled symbols) and upper-canopy (open symbols)leaves of studied poplar clones. Data were fitted by linear regressions.Error bars provide ± SE. by guest on M

ay 23, 2011treephys.oxfordjournals.org

Dow

nloaded from

Page 10

midrib (Figures 10A, 10C and 10E), demonstrating a trade-offbetween enhanced leaf morphological plasticity and leaf sup-port. Similar relationships were found for midrib to lamina drymass ratio. Furthermore, plasticity in MA, NA and AL was nega-tively related to relative petiole length (Figures 10B, 10D and10F), but not to the fraction of leaf biomass in petiole (data notshown). This suggests that avoidance of leaf-level clumpingby longer petioles may be associated with a constrained capac-ity for MA and NA variation. There was a strong negative corre-lation between relative petiole length and biomass investmentin midrib (r2 = 0.69, P < 0.001) complicating the interpretation

of the observed relationships with leaf-level morphologicalplasticity.

Discussion

Aggregation and light-use efficiency

Previous studies have demonstrated a strong clonal variabilityin light interception versus stand L relationships (Heilman etal. 1996). We also observed an overall poor relationship be-tween canopy transmittance and total leaf area index (Fig-ure 3A), and large differences in total and effective leaf area(Figure 3B), indicating an important effect of leaf aggregationon foliar light-harvesting efficiency (Equation 1). Dependingon foliage inclination angle distributions, leaf clumping in-creased canopy transmittance 1.5–4-fold, and reduced canopylight harvesting by about 10–20% relative to light interceptionfor a random dispersion of foliar elements in various poplarclones (Figures 2–4). This important effect of foliar clumpingon light-interception efficiency is comparable with previousobservations in dense poplar plantations growing at ambientand free-air elevated CO2 concentrations, and exhibiting 1.5–2.5-fold larger canopy transmittance as a result of leaf aggre-gation than a canopy with equivalent leaf area and a randomdispersion of foliage elements (Gielen et al. 2001). Despite thedramatic reduction in canopy light interception, mean irra-diance on the leaf surface was only moderately affected com-pared with random dispersion of foliar elements (Figure 4B).Thus, foliar clumping results in an enhancement in the total fo-liar area (Chen et al. 1997, Cescatti 1998) that is compensatedfor by a reduction in mean irradiance on the leaf surface.

We observed that the degree of leaf clumping increased(lower �0) with increasing stand leaf area index (Figure 3C),suggesting that canopy light-interception capacity does notnecessarily increase linearly with increasing total leaf area in-dex. In fact, the annual productivity of poplar clones is curvi-linearly related to intercepted irradiance, and biomass produc-tion per unit leaf area decreases with increasing stand L(Heilman and Fu-Guang 1994). As our study suggests, lowerstand light-harvesting efficiency with increasing L may bepartly determined by progressively lower mean irradiance onthe leaf surface as a result of the enhanced clumping (Fig-ure 4B).

Leaf structural determinants of canopy light harvesting

According to our study, canopy-level leaf aggregation is sig-nificantly affected by foliar attributes in dense poplar cano-pies. Petiole length was most strongly correlated with foliageclumping (Figure 6A), most likely because increases in petiolelength reduce the shading of basal lamina portions by the stemand adjacent leaves, leading to significantly increased light-harvesting efficiency (Figure 7B).

Although leaf size itself may affect leaf aggregation in thecanopy (Horn 1971, Bunce 1990, Takenaka 1994), we ob-served no relationships between leaf clumping and leaf size orlength (Figure 6B). This may be associated with the specificleaf shapes of poplar species. For cordate, deltoid or ovate leaf

150 NIINEMETS ET AL.

TREE PHYSIOLOGY VOLUME 24, 2004

Figure 9. Lamina dry mass per unit area (MA) of lower-canopy leaves(A), corresponding leaf nitrogen contents per area (B), and the differ-ence in MA between upper and lower canopy (C) in relation to canopytransmittance in the studied poplar clones. Data were fitted by linearregressions. Error bars denote ± SE in A and B, and the sum of stan-dard errors of upper- and lower-canopy leaves in C.

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 11

shapes that are typical in poplar, most of the leaf area is nearthe leaf base, implying that a change in leaf size only moder-ately affects self-shading, whereas modifications in petiolelength may radically change leaf light-interception capacity.

Petiole length plays a dual role in leaf light harvesting.Apart from the effects of petioles on clumping, changes in pet-iole length may significantly modify lamina angle with respectto the horizontal. According to beam theory, lamina deflectionfrom the horizontal increases with the cube of petiole length(Niklas 1991a, 1991b), implying that modest changes in peti-ole length may have a large impact on leaf angle. However, theexact relationship between petiole length and leaf deflectionunder its own mass depends on petiole geometry (Niklas1991b) and on the shape of the distribution of lamina load(Niinemets and Fleck 2002a, 2002b). Given that most of theleaf lamina mass is located close to the petiole attachment tothe lamina, increases in petiole length in poplar lead to smallerchanges in lamina deflection than in leaves with lanceolate, el-liptical, or obovate shapes. Nevertheless, there are importantclonal differences in leaf angle distributions (Isebrands and

Michael 1986, Ceulemans and Isebrands 1996) that may sig-nificantly depend on interclonal variations in petiole lengthand leaf biomass investments in petioles.

In our study, most coppiced stands consisted primarily ofunbranched vertical shoots, making it possible to test for theeffects of foliar characteristics on canopy light harvesting.Nevertheless, part of the variability in foliar clumping is gen-erally associated with clonal differences in branching andcrown structure (Ceulemans et al. 1990, Chen et al. 1994), e.g.,with differences in sylleptic versus proleptic branch formation(Ceulemans et al. 1990, Ceulemans and Isebrands 1996, Dick-mann et al. 2001), as well as with differences in internodelengths (Gielen et al. 2002). Such variations in branchingarchitecture may modify the relative importance of leaf char-acteristics for whole-canopy light-harvesting efficiency. Al-though the degree of branching was low in our study, some ofthe clonal differences in light harvesting may have been influ-enced by differences in branching. For instance, ‘Wolterson’possesses many short branches with short internodal distances,leading to the most clumped canopy among the studied species

TREE PHYSIOLOGY ONLINE at http://heronpublishing.com

LEAF AGGREGATION AND LIGHT INTERCEPTION IN POPLAR 151

Figure 10. Difference in (A, B) MA

between upper- and lower-canopyleaves per unit relative interceptedirradiance (�MA/IA, Equation 7), in(C, D) nitrogen content per area(NA) between upper- and lower-canopy leaves per IA (�NA/IA), andin (E, F) mean leaf area (AL) be-tween upper- and lower-canopyleaves per IA (�AL/IA) in relation tothe fraction of foliar biomass inmidrib (A, C, E), and the ratio ofpetiole to total leaf length (B, D, F)in the studied poplar clones. Bothexplanatory variables were themeans of the upper- and lower-canopy leaves. Relative interceptedirradiance is equal to 1 – canopytransmittance. Data were fitted bylinear regressions.

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 12

(Figures 2A and 3). In contrast, in ‘Balsam Spire’, a clone thatbranches significantly less than most other clones (Figure 2C),canopy-level leaf clumping is mainly associated with limitedpetiole length. Although branching can be a significant deter-minant of light harvesting, young poplar stands invest onlyabout 13–20% of aboveground woody biomass in branches(Heilman et al. 1994), demonstrating that the contribution ofbranch-level clumping is relatively less in coppiced poplarstands than in broad-leaved natural forest stands.

Biomass costs for light harvesting

The fractional biomass investment in petioles observed in ourstudy (5–12%, Table 2) is low compared with the fraction ofbiomass generally invested in branches. Nevertheless, tissuedensity and cross-sectional dimensions are larger in branchesthan in petioles. This suggests that, for a common fraction ofbiomass in support, total length of support framework is lon-ger for petioles than for branches. Given that the efficiency oflight interception scales positively with the length of support-ing elements (Pearcy and Yang 1998), large canopy variationin light-harvesting efficiency is not necessarily at odds withlow fractions of leaf biomass in petioles.

Increases in petiole length led to enhanced fractional foliarbiomass investments in petioles (Figure 8A, Table 2), suggest-ing that enhanced light harvesting is increasingly expensive interms of non-photosynthetic throw-away biomass. There wasa negative correlation between stand L and petiole length (Fig-ure 7A), providing correlative evidence that the support costsmay limit extensive foliar area development. Nevertheless,there was a significant variation in leaf biomass investments inpetioles at a common petiole length (Table 2), suggesting thatdifferences in petiole cross-sectional dimensions may affectthe costs for light interception.

Many studies have observed a positive correlation betweenmean leaf size and L in clonal poplar stands (Ridge et al. 1986,Barigah et al. 1994, Ferris et al. 2001), but the mechanistic ex-planations for this relationship are essentially lacking. Appar-ently, larger leaves have greater biomass investments in midrib(Figure 8B), suggesting that in canopies with the same L, bio-mass investment in support within the lamina is larger in a can-opy with larger leaves. However, larger leaves require lowerbiomass investment in small branches for efficient lamina ex-position than small leaves (Givnish 1978), implying that lowerrequirements for biomass investments in woody structuresmay outweigh the larger requirements for support within thelamina. From a different perspective, the size of poplar leavesis influenced by modifications in carbohydrate availability.For instance, there are often positive correlations between leafsize and foliage photosynthetic capacity (Barigah et al. 1994).Thus, overall carbohydrate availability for foliage construc-tion may also explain positive correlations between leaf sizeand canopy L.

Although we cannot infer cause and effect from correlativerelationships, our results suggest that there may be importanttrade-offs between biomass investments in support and totalstand leaf area formed. This suggestion is supported by experi-mental observations that clones with the largest leaves and

total leaf area index are not always the most productive (Dick-mann et al. 2001).

Morphological constraints on the vertical variation in leafdry mass per area and nitrogen

Because variation in MA and NA along the plant canopies is theprimary determinant of light-dependent modifications in leafphotosynthetic capacity (Ellsworth and Reich 1993, Niine-mets et al. 1998), canopy variation in MA and NA provides anindirect estimate of foliage photosynthetic plasticity. Accord-ing to optimization studies, plants with the largest light-de-pendent plasticity in MA and nitrogen have the highest canopyphotosynthesis rates for given plant biomass and nitrogen inleaves (Gutschick and Wiegel 1988, Farquhar 1989). How-ever, plant acclimation is inherently limited, and there is alarge species (Meir et al. 2002) and clonal (Casella and Ceule-mans 2002) variation in plasticity. We observed a clear effectof the fraction of penetrating irradiance on the minimum MA

(Figure 9A) and NA (Figure 9B) at the bottom of the canopy.The gradient in MA (Figure 9C) and nitrogen between canopytop and bottom was apparently unrelated to penetrating ir-radiance. Given that MA and NA acclimated to low irradiance inall clones in a similar manner (Figures 9A and 9B), it seemsthat limited plasticity was primarily related to limitations inthe adjustment of upper-canopy leaves. This result is in con-trast with other studies comparing different species, where agreater plasticity in MA was observed in lower than in higherirradiance (Strauss-Debenedetti and Bazzaz 1991, Valladareset al. 2002a).

Overall, changes in MA, NA and mean leaf area per inter-cepted irradiance were positively related to the fraction of leafdry mass in midrib (Figures 10A, 10C and 10E), suggestingthat enhanced plasticity is bound to elevated support costs. Theclones inherently investing more biomass in support within thelamina also had more efficient water conduction pathways andmechanically more robust leaves to support heavier and larger(Figure 10C) laminas. Thus, meeting the requirements for sup-port may be an important condition for high leaf-level plastic-ity.

Plasticity in MA and NA was negatively associated with peti-ole to total leaf length ratio, implying a trade-off between leafaggregation and plasticity in MA and NA. This apparent com-promise possibly reflects conflicting requirements for bio-mass investments in petiole versus leaf lamina. Although wecannot discriminate among various hypotheses of the controlson leaf plasticity using correlative arguments, our study hintsat important trade-offs between stand light interception, totalleaf area and foliar architecture as well as between photo-synthetic and structural plasticity that may be mediated by drymass investments in support. Given that morphological ratherthan physiological characteristics are generally employed inselection for more productive poplar genotypes (Dickmann etal. 2001), such compromises may need consideration.

Acknowledgments

We thank Professor Dr. Robert W. Pearcy and Dr. Ilse Laureysens forthoughtful comments on the study, and Ahmad Farid and Nadine

152 NIINEMETS ET AL.

TREE PHYSIOLOGY VOLUME 24, 2004

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 13

Calluy for nitrogen analyses. We gratefully acknowledge the logistic

support from Eta-com B. and the City Council of Boom (Belgium) at

the field site. Funding for this study was provided by the Province of

Antwerpen, the Belgian Federal Office for Scientific, Technical and

Cultural Affairs (OSTC; research fellowship to Ü. Niinemets), the Es-

tonian Science Foundation (Grant No. 4584), the German Academic

Exchange Service (equipment grant to Ü.N.), and the Province of

Trento (Grant REM).

References

Armstrong, A. 1997. The United Kingdom network of experiments onsite/yield relationships for short rotation coppice. Forestry Commis-sion Research Information, Note 294, Forestry Commission, Edin-burgh, U.K., 6 p.

Asner, G.P., C.A. Wessman and S. Archer. 1998. Scale dependence ofabsorption of photosynthetically active radiation in terrestrial eco-systems. Ecol. Appl. 8:1003–1021.

Baldocchi, D.D. and B.A. Hutchison. 1986. On estimating canopyphotosynthesis and stomatal conductance in a deciduous forestwith clumped foliage. Tree Physiol. 2:155–168.

Barigah, T.S., B. Saugier, M. Mousseau, J. Guittet and R. Ceulemans.1994. Photosynthesis, leaf area and productivity of five poplarclones during their establishment year. Ann. Sci. For. 51:613–625.

Barnéoud, C., P. Bonduelle and J.M. Dubois. 1982. Manuel de populi-culture. Afocel, Paris, 320 p.

Bunce, J.A. 1990. The effect of leaf size on mutual shading andcultivar differences in soybean leaf photosynthetic capacity. Photo-synth. Res. 23:67–72.

Campbell, G.S. 1986. Extinction coefficients for radiation in plantcanopies calculated using an ellipsoidal inclination angle distribu-tion. Agric. For. Meteorol. 36:317–321.

Campbell, G.S. and J.M. Norman. 1998. An introduction to environ-mental biophysics. 2nd Edn. Springer-Verlag, New York, 286 p.

Campbell, G.S. and J.M. Norman. 1989. The description and mea-surement of plant canopy structure. In Plant Canopies: their Growth,Form and Function. Eds. G. Russell, B. Marshall and P.G. Jarvis. So-ciety for Experimental Biology seminar series No. 31. CambridgeUniversity Press, Cambridge, pp 1–19.

Casella, E. and R. Ceulemans. 2002. Spatial distribution of leaf mor-phological and physiological characteristics in relation to local ra-diation regime within 3-year-old Populus clones under coppiceculture. Tree Physiol. 22:1277–1288.

Cescatti, A. 1998. Effects of needle clumping in shoots and crowns onthe radiative regime of a Norway spruce canopy. Ann. Sci. For.55:89–102.

Ceulemans, R. and J.G. Isebrands. 1996. Carbon acquisition and allo-cation. In Biology of Populus and its Implications for Managementand Conservation. Eds. R.F. Stettler, H.D. Bradshaw, P.E. Heilmanand T.M. Hinckley. Natural Research Council of Canada, Ottawa,pp 355–399.

Ceulemans, R., R.F. Stettler, T.M. Hinckley, J.G. Isebrands and P.E.Heilman. 1990. Crown architecture of Populus clones as determinedby branch orientation and branch characteristics. Tree Physiol. 7:157–167.

Ceulemans, R., J.-Y. Pontailler, F. Mau, J. Guittet and B. Legay. 1993.Leaf allometry in young poplar stands: reliability of leaf area indexestimation, site and clone effects. Biomass Bioenergy 4: 315–321.

Chen, J.M. and J. Cihlar. 1995a. Plant canopy gap-size analysis theoryfor improving optical measurements of leaf-area index. Appl. Opt.34:6211–6222.

Chen, J.M. and J. Cihlar. 1995b. Quantifying the effect of canopy archi-tecture on optical measurements of leaf area index using two gap sizeanalysis methods. IEEE Trans. Geosci. Remote Sens. 33:777–787.

Chen, J.M., P.M. Rich, S.T. Gower, J.M. Norman and S. Plummer.1997. Leaf area index of boreal forests: theory, techniques andmeasurements. J. Geophys. Res. 102:29,429–29,443.

Chen, S.G., R. Ceulemans and I. Impens. 1994. A fractal-basedPopulus canopy structure model for the calculation of light intercep-tion. For. Ecol. Manage. 69:97–110.

Deraedt, W. and R. Ceulemans. 1998. Clonal variability in biomassproduction and conversion efficiency of poplar during the estab-lishment year of a short rotation coppice plantation. Biomass Bio-energy 15:391–398.

Dickmann, D.I., J.G. Isebrands, T.J. Blake, K. Kosola and J. Kort.2001. Physiological ecology of poplars. In Poplar Culture in NorthAmerica. Eds. D.I. Dickmann, J.G. Isebrands, J.E. Eckenwalder andJ. Richardson. NRC Research Press, Ottawa, pp 77–118.

Ellsworth, D.S. and P.B. Reich. 1993. Canopy structure and verticalpatterns of photosynthesis and related leaf traits in a deciduous for-est. Oecologia 96:169–178.

Farque, L., H. Sinoquet and F. Colin. 2001. Canopy structure and lightinterception in Quercus petraea seedlings in relation to light regimeand plant density. Tree Physiol. 21:1257–1267.

Farquhar, G.D. 1989. Models of integrated photosynthesis of cellsand leaves. Philos. Trans. R. Soc. Lond. B Biol. Sci. 323: 357–367.

Ferris, R., M. Sabatti, F. Miglietta, R.F. Mills and G. Taylor. 2001.Leaf area is stimulated in Populus by free air CO2 enrichment(POPFACE) through increased cell expansion and production. PlantCell Environ. 24:305–315.

Fleck, S., Ü. Niinemets, A. Cescatti and J.D. Tenhunen. 2003. Three-dimensional lamina architecture alters light harvesting efficiency inFagus: a leaf-scale analysis. Tree Physiol. 23:577–589.

Gielen, B., C. Calfapietra, M. Sabatti and R. Ceulemans. 2001. Leafarea dynamics in a closed poplar plantation under free-air carbondioxide enrichment. Tree Physiol. 21:1245–1255.

Gielen, B., C. Calfapietra, A. Clau, M. Sabatti and R. Ceulemans.2002. Crown architecture of Populus spp. is differentially modifiedby free-air CO2 enrichment (POPFACE). New Phytol. 143:91–99.

Givnish, T.J. 1978. Ecological aspects of plant morphology: leaf formin relation to environment. Acta Biotheor. 27:83–142.

Givnish, T.J. 1986. Biomechanical constraints on crown geometry inforest herbs. In On the Economy of Plant Form and Function. Pro-ceedings of the Sixth Maria Moors Cabot Symposium, “EvolutionaryConstraints On Primary Productivity: Adaptive Patterns of EnergyCapture in Plants,” Harvard Forest, August 1983. Ed. T.J. Givnish.Cambridge University Press, Cambridge, pp 525–583.

Gutschick, V.P. and F.W. Wiegel. 1988. Optimizing the canopy photo-synthetic rate by patterns of investment in specific leaf mass. Am.Nat. 132:67–86.

Heilman, P.E. and X. Fu-Guang. 1994. Effects of nitrogen fertiliza-tion on leaf area, light interception, and productivity of short-rota-tion Populus trichocarpa × Populus deltoides hybrids. Can. J. For.Res. 24:166–173.

Heilman, P.E., G. Ekuan and D. Fogle. 1994. Above- and below-ground biomass and fine roots of 4-year-old hybrids of Populustrichocarpa × Populus deltoides. Can. J. For. Res. 24:1186–1192.

Heilman, P.E., T.M. Hinckley, D.A. Roberts and R. Ceulemans. 1996.Production physiology. In Biology of Populus and its Implicationsfor Management and Conservation. Eds. R.F. Stettler, H.D. Bradshaw,P.E. Heilman and T.M. Hinckley. National Research Council Canada,Ottawa, pp 459–489.

Horn, H.S. 1971. The adaptive geometry of trees. Princeton Univer-sity Press, Princeton, NJ, 144 p.

TREE PHYSIOLOGY ONLINE at http://heronpublishing.com

LEAF AGGREGATION AND LIGHT INTERCEPTION IN POPLAR 153

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from

Page 14

Isebrands, J.G. and D.A. Michael. 1986. Effects of leaf morphologyand orientation on solar radiation interception and photosynthesisin Populus. In Crown and Canopy Structure in Relation to Productiv-ity. Eds. T. Fujimori and D. Whitehead. Forestry and Forest ProductsResearch Institute, Ibaraki, pp 359–381.

Isebrands, J.G. and N.D. Nelson. 1982. Crown architecture of short-rotation, intensively cultured Populus. II. Branch morphology anddistribution of leaves within the crown of Populus ‘Tristis’as related tobiomass production. Can. J. For. Res. 12:853–864.

Jarvis, P.G. and J.W. Leverenz. 1983. Productivity of temperate, de-ciduous and evergreen forests. In Encyclopedia of Plant Physiology.Vol. 12D. Eds. O.L. Lange, P.S. Nobel, C.B. Osmond and H. Ziegler.Springer-Verlag, Berlin, pp 233–280.

Kucharik, C.J., J.M. Norman and S.T. Gower. 1999. Characterizationof radiation regimes in nonrandom forest canopies: theory, mea-surements, and a simplified modeling approach. Tree Physiol. 19:695–706.

Lacaze, R., J.M. Chen, J.-L. Roujean and S.G. Leblanc. 2002. Re-trieval of vegetation clumping index using hot spot signatures mea-sured by POLDER instrument. Remote Sens. Environ. 79:84–95.

Lang, A.R.G. 1986. Leaf area and mean leaf angle from transmissionof direct sunlight. Aust. J. Bot. 34:349–355.

Laureysens, I., J. Bogaert, R. Blust and R. Ceulemans. 2003a. Bio-mass production of 17 poplar clones in a short-rotation coppice cultureand its relation to soil characteristics. For. Ecol. Manage. In Press.

Laureysens, I., W. Deraedt, T. Indeherberge and R. Ceulemans.2003b. Population dynamics in a 6-year old coppice culture of poplar.I. Clonal diffrences in stool mortality, shoot dynamics and shoot diam-eter distribution in relation to biomass production. Biomass Bioenergy24:81–95.

Lüttge, U. 1997. Physiological ecology of tropical plants. Springer-Verlag, Berlin, 384 p.

Meir, P., B. Kruijt, M. Broadmeadow, E. Barbosa, O. Kull, F. Cars-well, A. Nobre and P.G. Jarvis. 2002. Acclimation of photosynthet-ic capacity to irradiance in tree canopies in relation to leaf nitrogenconcentration and leaf mass per unit area. Plant Cell Environ.25:343–357.

Miller, J.B. 1967. A formula for average foliage density. Aust. J. Bot.15:141–144.

Niinemets, Ü. and S. Fleck. 2002a. Leaf biomechanics and biomassinvestment in support in relation to long-term irradiance in Fagus.Plant Biol. 4:523–534.

Niinemets, Ü. and S. Fleck. 2002b. Petiole mechanics, leaf inclina-tion, morphology, and investment in support in relation to lightavailability in the canopy of Liriodendron tulipifera. Oecologia 132:21–33.

Niinemets, Ü., O. Kull and J.D. Tenhunen. 1998. An analysis of lighteffects on foliar morphology, physiology, and light interception intemperate deciduous woody species of contrasting shade tolerance.Tree Physiol. 18:681–696.

Niklas, K.J. 1988. The role of phyllotactic pattern as a “developmen-tal constraint” on the interception of light by leaf surfaces. Evolu-tion 42:1–16.

Niklas, K.J. 1991a. The elastic moduli and mechanics of Populustremuloides (Salicaceae) petioles in bending and torsion. Am. J. Bot.78:989–996.

Niklas, K.J. 1991b. Flexural stiffness allometries of angiosperm andfern petioles and rachises: evidence for biomechanical conver-gence. Evolution 45:734–750.

Niklas, K.J. 1992. Petiole mechanics, light interception by lamina,and “economy in design.” Oecologia 90:518–526.

Nilson, T. 1971. A theoretical analysis of the frequency of gaps inplant stands. Agric. Meteorol. 8:25–38.

Pearcy, R.W. and W. Yang. 1998. The fuctional morphology of lightcapture and carbon gain in the Redwood forest understorey plant,Adenocaulon bicolor Hook. Funct. Ecol. 12:543–552.

Pellis, A. 2002. Bladoppervlakte, bladkarakteristieken en biomassa-productie van een populieren-hakhoutcultuur één jaar naterugsnijden. Licientiaat in Biologie Dissertation, Dept. of Biol-ogy, Fac. of Sci., Univ. Antwerpen, 48 p. In Dutch.

Ridge, C.R., T.M. Hinckley, R.F. Stettler and E. Van Volkenburgh.1986. Leaf growth characteristics of fast-growing poplar hybridsPopulus trichocarpa × P. deltoides. Tree Physiol. 1:209–216.

Ross, J. 1981. The radiation regime and architecture of plant stands.Dr. W. Junk, The Hague, 342 p.

Strauss-Debenedetti, S. and F.A. Bazzaz. 1991. Plasticity and accli-mation to light in tropical Moraceae of different successional posi-tions. Oecologia 87:377–387.

Takenaka, A. 1994. Effects of leaf blade narrowness and petiolelength on the light capture efficiency of a shoot. Ecol. Res. 9:109–114.

ter Steege, H. 1996. Winphot 5: a programme to analyze vegetationindices, light and light quality from hemispherical photographs.Tropenbos Guyana Reports, 95-2. Tropenbos Guyana Programme,Tropenbos, Georgetown, Guyana, 40 p.

Valladares, F. and R.W. Pearcy. 1999. The geometry of light intercep-tion by shoots of Heteromeles arbutifolia: morphological and physio-logical consequences for individual leaves. Oecologia 121:171–182.

Valladares, F., J.M. Chico, I. Aranda, L. Balaguer, P. Dizengremel,E. Manrique and E. Dreyer. 2002a. The greater seedling high-lighttolerance of Quercus robur over Fagus sylvatica is linked to a greaterphysiological plasticity. Trees 16:395–403.

Valladares, F., J.B. Skillman and R.W. Pearcy. 2002b. Convergence inlight capture efficiencies among tropical forest understory plantswith contrasting crown architectures: a case of morphological com-pensation. Am. J. Bot. 89:1275–1284.

Vogelmann, T.C., P.R. Larson and R.E. Dickson. 1982. Translocationpathways in the petioles and stems between source and sink leavesof Populus deltoides Bartr. ex Marsh. Planta 156:345–358.

Weber, J.C., R.F. Stettler and P.E. Heilman. 1985. Genetic variationand productivity of Populus trichocarpa T. & G. and its hybrids. I.Morphology and phenology of 50 native clones. Can. J. For. Res.15:376–383.

Wilkinson, L. 1990. SYSTAT: The system for statistics. SYSTAT,Inc., Evanston, IL, 677 p.

Zavitkovski, J. 1981. Small plots with unplanted plot border can dis-tort data in biomass production studies. Can. J. For. Res. 11:9–12.

Zavitkovski, J. 1982. Characterization of light climate under canopiesof intensively-cultured hybrid poplar plantations. Agric. Meteorol.25:245–255.

Zwieniecki, M.A., L. Hutyra, M.V. Thompson and N.M. Holbrook.2000. Dynamic changes in petiole specific conductivity in red ma-ple (Acer rubrum L.), tulip tree (Liriodendron tulipifera L.) and north-ern fox grape (Vitis labrusca L.). Plant Cell Environ. 23:407–414.

154 NIINEMETS ET AL.

TREE PHYSIOLOGY VOLUME 24, 2004

by guest on May 23, 2011

treephys.oxfordjournals.orgD

ownloaded from