Agricultural and Forest Meteorology 111 (2002) 121–140

Modification of light-acclimation ofPinus sylvestrisshoot architecture by site fertility

Ülo Niinemetsa,∗, Alessandro Cescattib, Aljona Lukjanovac,Mari Tobiasc, Laimi Truusc

a Department of Plant Physiology, Institute of Molecular and Cell Biology, University of Tartu, Riia 23, Tartu 51010, Estoniab Centro di Ecologia Alpina, I-38040 Viote del Monte Bondone (TN), Italy

c Department of Ecophysiology, Institute of Ecology, Tallinn University of Educational Sciences, Kevade 2, Tallinn 10137, Estonia

Received 21 June 2001; received in revised form 17 January 2002; accepted 1 February 2002

Abstract

Adjustment of shoot architectural characteristics—needle angle with respect to shoot axis (α), needle mass per unit shootsilhouette area (MS), needle number per unit shoot axis length (KL), silhouette to total needle area ratio (SS)—to seasonal av-erage daily integrated quantum flux density (Qint) was investigated in coniferPinus sylvestrisL. in an old field (fertile site) andin a raised bog (infertile site). For both sites,KL increased with increasingQint, leading to a greater foliar area and dry mass perunit shoot length, as well as to largerMS and foliar nitrogen content per unit silhouette area at higher irradiance. The five-foldincrease inMS with Qint via architectural modifications allowed biomass concentration in high light environment where thephotosynthetic returns were the highest. However, the negative correlations betweenSS andQint indicated that enhanced needleproduction also led to a lower sunlit needle area fraction and greater self-shading within the shoot at higher irradiance. Simula-tions using canopy gap fractions across the sky hemisphere incident to the shoots in their natural position further demonstrateda decrease in light interception efficiency relative to a flat surface (ϑ) with increasingQint. A shoot architectural model based ona turbid medium analogy suggested that the decreases in the efficiency primarily resulted from increased clumping and largerneedle area density at higher irradiances. The relationships were qualitatively similar at both sites, but the needles were shorterand the packing of needles was higher leading to a lower gap fraction within the shoot volume and to greater self-shadingwithin the shoot at the nutrient-limited site, especially under low irradiance. We conclude that both needle and shoot levelmodifications contribute to plastic alterations in shoot light harvesting efficiency inP. sylvestris, but also that site fertility maysignificantly constrain the light-acclimation in shoot architecture. © 2002 Elsevier Science B.V. All rights reserved.

Keywords:Conifers; Light interception; Needle packing; Nutrient availability; Radiative transfer; Shoot morphology

1. Introduction

In conifers, foliage aggregation at shoot and crownlevel plays a paramount role in light interception ofthe entire canopy (Whitehead et al., 1990; Chen et al.,1997; Nilson and Ross, 1997). Inherently large needle

∗ Corresponding author. Fax:+372-7-366021.E-mail address:ylo@zbi.ee (Ü. Niinemets).

clumping in these species is compatible with a greatergap fraction and thus, with a deeper penetration oflight in the canopy than is possible in a non-aggregatedstand with equal foliar area (Whitehead et al., 1990;Stenberg, 1998). Although the lower shading effi-ciency of unit foliar area and greater incident irradi-ances in various crown layers allow conifer trees tocarry more foliar area per unit ground area than the de-ciduous species can do, enhanced clumping also brings

0168-1923/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.PII: S0168-1923(02)00011-4

122 Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140

Nomenclature

AA(φ, ϕ) projected area of the shootaxis (mm2)

AL total needle area per unit stemaxis length (mm2 mm−1)

AP(α, φn) needle projected area in a specificneedle location on the shoot(Eq. (B.1), mm2)

AP,i projected area of the needles onthe shoot from the age-classi(mm2)

AS(φ, ϕ) measured shoot silhouettearea (mm2)

A′S(φ, ϕ) predicted shoot silhouette

area (Eq. (7), mm2)A′

S spherically averaged shootsilhouette area (Eq. (9), mm2)

AT̄ total needle area per needle (mm2)AT,i total area of the needles on the

shoot from the age-classi (mm2)ATh estimate of shoot silhouette area

without self-shading(Eq. (B.2), mm2)

c parameter of the ellipsoidalneedle angle distribution(Eqs. (5) and (6))

C current-year needlesCn needle circumference (mm)E square error (Eq. (8))f spiral number in the

phyllotactic cycleF(φ, ϕ) probability of photon

interception (Eq. (4))G(c, φ) extinction coefficient (Eq. (5))Idif fraction of diffuse solar

radiation incident to theshoot (diffuse site factor)

Idir fraction of potential directradiation incident to the shoot(direct site factor)

Isum relative amount of global solarradiation incident to theshoot (Eq. (1))

KL needle number per unit shootaxis length (mm−1)

La shoot axis length (mm)

LB(φ, ϕ) beam path length in theshoot volume (mm)

Lk length of non-needled partof the shoot axis (mm)

Ln needle length (mm)Ls total shoot length (mm)M total needle dry mass per shoot (g)MA needle dry mass per unit

total needle area (g m−2)ML needle dry mass per unit shoot

axis length (g mm−1)MP needle dry mass per unit

projected area (g m−2)MS̄ total needle dry mass per

A′S (Eq. (11), kg m−2)

n number of needle age-classeson the shoot

NM needle nitrogen contentper unit dry mass (%)

pdif ratio of diffuse to global solarradiation above the canopy

PM needle phosphorus contentper unit dry mass (%)

Q(γ ) daily integrated above-canopyquantum flux density on ahorizontal plane coming from azenith angleγ (mol m−2 d−1)

Qint seasonal average dailyintegrated photosyntheticallyactive quantum flux densityabove the shoot (mol m−2 d−1)

Q0int Qint above the canopy

(mol m−2 d−1)QSh daily average amount of

intercepted quanta per unit needlearea (Eq. (A.2), mol m−2 d−1)

R(φ, ϕ) mean quantum flux densityper needle area (expressedas half of the total) relativeto quantum flux density ontoa plane orthogonal to the lightbeam (Eq. (A.1))

SP(φ, ϕ) AS(φ, ϕ) to projected shootneedle area ratio

SS(φ, ϕ) AS(φ, ϕ) to total shootneedle area ratio (Eq. (2))

Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140 123

S′S spherically averagedSS(φ, ϕ)

(Eq. (10))T needle thickness (mm)Wd shoot axis diameter (mm)Wn needle width (mm)Ws(φ, ϕ) shoot width (mm)

Greek lettersα angle between needle and

shoot axis (◦)δS fraction of one-sided sunlit

area (Eq. (12))φ shoot rotation angle around

its axis (◦)φn needle rotation angle around

shoot axis (◦)φn,k average needle rotation angle

for the rotation angle classk (◦)φγ shoot rotation angle for the

view directionγ (◦)γ solar zenith angle (◦)η divergence angle, i.e. the rotation

angle between two adjacentneedles (◦)

ϑ shoot efficiency of lightabsorption relative toa flat surface

ϕ shoot inclination angle (◦)ϕγ shoot inclination angle for

the view directionγ (◦)λ0 Markov coefficient

for spatial aggregationρ needle area density (Eq. (3),

m2 m−3)

about decreases in mean irradiance at needle surface.Decreasing efficiencies of light harvesting apparentlydo not significantly curb the daily shoot photosynthe-sis in the upper canopy, where the irradiances may besufficiently large for high rates of foliar photosynthe-sis. However, the incident irradiances progressivelydecline with increasing canopy depth, diminishing theadvantages of shoots with high needle area packing.

To balance the requirements for high shoot photo-synthetic capacity that is more advantageous in highlight and enhanced light harvesting efficiency that

gains in importance with decreasing irradiance, conifershoots acclimate to long-term light availability gradi-ent in the canopy. In particular, needle number per unitshoot length decreases (Niinemets and Kull, 1995a)and the fraction of exposed foliar area increases (Ni-inemets and Kull, 1995a; Sprugel et al., 1996; Sten-berg, 1996; Niinemets, 1997; Stenberg et al., 2001)with decreasing irradiance. These collective changeslead to lower within shoot shading in lower lightavailability. In addition, needles are also thinner withlower dry mass per unit needle area at lower irradi-ance (Kellomäki and Oker-Blom, 1981; Niinemetsand Kull, 1995b; Sprugel et al., 1996; Niinemets,1997) and accordingly, they have a larger exposedsurface area and lower biomass cost for formation.Studies demonstrate that modification of needle pro-jected to total surface area may strongly alter shootlight interception efficiency (Jordan and Smith, 1993).

Apart from the light effects on needle and shootstructure, nutrient availability may also stronglyaffect conifer foliage morphology. Needle length(Smolander et al., 1990; Norgren and Elfving, 1994;Murthy and Dougherty, 1997; Stenberg et al., 1999;Niinemets et al., 2001), width (Niinemets et al.,2001) as well as total foliar area per tree (Vose andAllen, 1988; Albaugh et al., 1998) scale positivelywith nitrogen availability. Greater foliar area produc-tion in response to improved nutrition may implythat within shoot shading increases with increasingN availability. However, improved nutrition generallystimulates shoot length growth as well (Beadle, 1966;Murthy and Dougherty, 1997) and there is evidenceof a lower needle number per unit shoot axis lengthand a higher light interception efficiency at fertilethan at infertile sites in some studies (Beadle, 1966;Smolander and Oker-Blom, 1989; Smolander et al.,1990; Stenberg et al., 1995), but not in others (Murthyand Dougherty, 1997; Stenberg et al., 1999). Giventhat nutrition-related enhanced production of foliararea also leads to lower light availabilities within thecanopy, it is currently unclear whether the observedmodifications in shoot morphology truly result fromchanges in nutrient availability or are attributable to anindirect effect of altered light conditions on structuralacclimation. Because shoot morphology versus lightrelationships may constitute a valuable basis for scal-ing of foliar light interception efficiency in stand-levelproductivity and light interception models (Baldocchi

124 Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140

and Meyers, 1998; Stenberg, 1998), it is important tocomprehend how general and widely applicable arethese dependencies. Furthermore, general correlationsbetween foliage structural and functional character-istics may provide a simplified parameterisation ofcomplex landscape- and biome-level productivitymodels (Reich et al., 1997). From these perspectives,determination of the possible role of nutrition onshoot morphology versus light relationships is verygermane.

We investigated light-acclimation of shoot morpho-logical features in canopies of the wide-spread conifer,Pinus sylvestrisL., at two sites of contrasting nutrientavailability to gain informative insight into the poten-tial effects of the modifications in the dimensions of in-dividual needles, needle number per unit shoot length,needle clumping and angular distribution on theefficiency of light interception. The fertile site was awell-drained old field with rapid nutrient turnover ratesand the infertile site was a raised bog with extremelylow peat mineralization rates and limited drainage. Asthe previous studies demonstrated (Niinemets et al.,2001), the needles were shorter (15–50 mm at theinfertile versus 36–67 mm at the fertile site) and thin-ner (0.40–0.67 mm versus 0.49–0.71 mm), with lowerN contents (0.59–1.32% versus 1.29–1.76%) and theshoots were shorter at a common irradiance in thenutrient-limited site. We also observed limited plas-ticity in the adjustment of needle morphological char-acteristics to light availability in the nutrient-limitedsite (Niinemets et al., 2001). Therefore, we hypothe-sised that constrained needle morphological plasticity,as well as shoot length growth translates into lowershoot morphological plasticity and reduced light in-terception efficiency at the infertile site. Given thatlow nutrient availability also led to decreased foliarphotosynthetic rates (Niinemets et al., 2001) and thatlow net assimilation rates may predispose leaves tophotoinhibition (Osmond et al., 1999), avoidanceof photoinhibitory damage should favour a largerself-shading within the shoot in nutrient-limitedsites.

To determine the shoot light interception efficiencyin the shoot natural location in the canopy and tounderstand the significance of specific morphologi-cal changes in needle morphology in terms of lightinterception, we developed a shoot radiative trans-fer model that is based on a turbid medium analogy

and therefore, is readily usable to model stand lightclimate.

2. Material and methods

2.1. Study sites

The fertile stand was a monospecificP. sylvestrisplantation (1400 trees/ha) on an old field at Ahuna-palu, Estonia (58◦19′N, 27◦17′E, elevation ca. 60 ma.s.l.). The trees were 29–31-year-old and 17–21 mtall. The soil was a sandy pseudo-gley with moder-ately acidic humus horizon. The pH determined in1 M KCl solution (pHKCl) was 4.3 for the uppermost0–5 cm soil horizon. We choose three 19–20 m talltrees in the center of the forest for sampling of foliagemorphological characteristics.

The nutrient-limited site was the Männikjärve raisedbog, Endla State Nature Reserve, Estonia (58◦52′N,26◦13′E, elevation 5–30 m a.s.l.) on thick—up to 8 min the middle of the bog—Sphagnumpeat (Veber,1974). The dominating trees ofP. sylvestrisandBetulapubescensEhrh. were scattered (200 trees/ha) and theaverage height of ca. 50–100-year-old trees was only1–2 m. The organic soil was acidic throughout theentire profile (average± S.D. of pHKCl for the upper5–15 peat layer was 2.59 ± 0.29, n = 17). For themorphological measurements, we selected 22 treeswith the heights ranging from 0.8 to 2 m in the centralareas of the bog. To attain a larger gradient in nutrientavailability, seven larger trees (height 2.9–8.7 m) withapparently better nutrition were chosen at the edge ofthe bog and on the adjacent dried peatlands, whichgenerally tend to have greater nutrient availabilities be-cause of faster soil mineralisation rates (Hånell, 1988).The trees sampled in Endla were 20–150-year-oldaccording to the increment cores taken at the groundlevel (average± S.E. = 43 ± 8). Additional detailsof both sites are communicated in Niinemets et al.(2001).

2.2. Foliage sampling

Branches at different heights in the canopy wereselected for the measurements. Three to four shootsat terminal branch positions were marked for detailedmorphological analyses and the hemispherical photo-

Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140 125

graphs were taken above these shoots for estimation oflong-term incident irradiance. Whole branches werefurther harvested, put in plastic bags and immediatelytransported to the laboratory, where the morphologicalcharacteristics of the marked shoots were measured.Always in Ahunapalu and generally in Endla, onlycurrent-year shoots (C) were used. However, in sev-eral cases annual shoot length growth increments werevery short (<2–10 mm) in the nutrient-limited site andapparently, analysis of current-year shoots alone didnot provide representative information of the shootlight interception capacity. Therefore, shoots with upto four needle age-classes (C,C+1,C+2,C+3) weremeasured in Endla. Needle longevity was consider-ably larger in Endla than in Ahunapalu (unpublisheddata of Niinemets and Lukjanova 1999), such that theneedle number per unit shoot axis length was essen-tially constant for the needle age-classesC, C + 1andC+ 2 in Endla. Thus, analysis of multiple needleage-classes at the infertile site was unlikely to bias theconclusions regarding the light and nutrition effectson shoot morphology. Moreover, in Endla, the needledry mass-based weighted age of the sampled shootsranged from 0.0 (current-year shoots) to 2.1 years andaveraged (±S.D.) 0.29±0.49 years, indicating that theshoot characteristics were primarily determined by thefoliage formed in the current-year in this site as well.

Foliar sampling for morphological analyses wasconducted for both sites in September 1998. Furthersamples were taken in July–August 1999 in Endla andin October–November 1999 in Ahunapalu. The dataof various years were pooled for statistical analyses,because the initial analysis (ANOVA) demonstratedthat the time of sampling did not significantly alterthe relationships of shoot structure versus canopylight availability and site nutrient conditions.

2.3. Estimations of quantum flux density withinthe canopy

The seasonal average daily integrated photosynthe-tically active quantum flux densities (Qint, mol m−2

d−1) in the canopy were determined combining esti-mation of relative solar irradiance by hemisphericalphotography and the measurements of components ofglobal solar radiation in a nearby meteorological sta-tion. From the hemispherical photographs taken abovethe sample shoots in each branch, the diffuse site factor

(Idif , fraction of diffuse solar radiation penetrating thecanopy) and direct site factor (Idir, the fraction of po-tential direct radiation reaching the foliage) were com-puted as described in detail in Niinemets et al. (1999,2001). From these values, the “global site factor”—the relative amount of global solar radiation incidentto the sample shoots, (Isum)—was found as

Isum = pdif Idif + (1 − pdif )Idir (1)

wherepdif is the ratio of diffuse to global solar ra-diation above the canopy (Niinemets et al., 1999).Measurements of Tõravere Actinometric Station(58◦16′N, 26◦28′E) were used to calculate an estimateof pdif .

A correlation betweenQint and global solar radia-tion measured at Tõravere Actinometric Station wasemployed to determine a conversion factor (globalsolar radiation to quantum flux density) of 1.92 molMJ−1 (Niinemets et al., 2001). Using this conversionfactor, an average value ofQint above the canopy,Q0

int = 40.4 mol m−2 d−1, was determined for thetime-span 1 May–31 July 1999. During this period,leaf growth and development occurred at both sites.Qint for each sample location in the canopy was com-puted as the product ofQ0

int andIsum.

2.4. Estimation of shoot silhouette area

Shoot photographs were taken by Zenit-EM cameraequipped with Yupiter 21M lens (focal length 200 mm)using a black and white high resolution 35 mm film.The shoots were placed in front of a white projec-tor screen and they were illuminated in a dark roomonly by highly intense diffuse irradiance reflectedfrom the screen (1000 W halogen projector lamp).The distance between the camera and the shoot wasgenerally fixed at 3.5 m, but the distance was 5–6 mfor larger shoots. Objects of known area (0.7–15 cm2)and various shapes were photographed together withshoots for latter determination of shoot area and lin-ear dimensions. We held the shoot axis perpendicularto the view direction (inclination angle,ϕ = 0◦) andstudied only the variation in the shoot projection inrelation to the shoot rotation angle around its axis (φ).Each shoot was always photographed from the rota-tion angles of 0 and 90◦ and in some cases also from180◦ (Fig. 1). For the projection 0◦, 0◦ (φ, ϕ), the up-per surface of the shoot is perpendicular to the view

126 Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140

Fig. 1. A representativePinus sylvestrisshoot with three needle age-classes (C, C + 1, C + 2, whereC denotes the current-year needles)from the nutrient-limited site. The shoot axis was held perpendicular to the view direction (inclination angle,ϕ = 0◦) and from eachshoot at least two projections were generated by rotating the shoot around its axis by 90◦. The shoot characteristics as defined innomenclature.

direction and the shoot is viewed from the side for theprojection 90◦, 0◦. The projection 180◦, 0◦ is oppositeto the projection 0◦, 0◦ and served as a control of the0◦, 0◦ projection measurements. In the shoot specificlocation in the canopy,ϕ is the angle of the shoot axisrelative to the plane of projection. For definition ofshoot rotation angle,φ, we first define vectorr thatpoints to the upper face of the shoot and that is normalto an hypothetical plane dividing the shoot into upperand lower faces. Thus,φ is the angle between the vec-tor r and the plane going through the shoot axis andthe view direction (Stenberg, 1996; Stenberg et al.,1998).

The negatives were scanned by FilmScan 200(Epson Europe, Amsterdam, The Netherlands) witha resolution of 1200 dpi. This resulted in a pixelsize of 0.015–0.035 mm2 for the camera distance of3.5 m from the shoot and of 0.050–0.075 mm2 for thecamera distance of 6 m from the shoot. The initialimage processing and conversion from 8 (greyscale)to 1 bit (black and white) bitmap was conducted byCorel Photo-Paint 8.369 (Corel Corporation, Ottawa,Ont., Canada). The areas of the shoot silhouettes,AS(φ,ϕ), were computed by self-made computersoftware (Pindala 1.0 by I. Kalamees). The esti-mates for the projections 0◦, 0◦ and 180◦, 0◦ wereaveraged.

2.5. Needle and shoot morphological characteristics

Shoot images were also used to measure total shootlength (Ls), shoot axis length (La), length of needledpart of shoot axis, shoot width (Ws(φ, ϕ)) and shootaxis diameter (Wd, Fig. 1) by UTHSCSA Image Tool2.00 alpha by C.D. Wilcox, S.B. Dove, W.D. McDavidand D.B. Greer (Department of Dental Diagnostic Sci-ence at the University of Texas Health Science Center,San Antonio, Texas, ftp://maxrad6.uthscsa.edu). Tenrandomly chosen needles in the image were measuredfor needle length (Ln) and for the angle between nee-dle and shoot axis (α) and averages were calculated(Fig. 1). The lengths of the stem axis correspondingto each needle age cohort (C, C + 1, C + 2, C + 3,whereC denotes the current-year needles) were alsoestimated from the images. Separate measurements ofthe shoot axis lengths by precision callipers demon-strated that the needle age-classes were reliablydetected from the images. For regressions betweenthe direct measurements by callipers and the mea-surements from the images, the slope was 0.96 andr2 = 0.92 (the regression was forced through zero)for the current-year needle cohort and the slope was1.02 andr2 = 0.95 for the 1-year-old needles.

Five to ten needles were randomly taken fromeach shoot needle age-class for determination of total

Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140 127

shoot needle area. Needle length (Ln), thickness (T)and width (Wn) of the sample needles were measuredby precision callipers and the projected needle area,AP, was calculated asWnLn and total needle area,AT̄ as the product of needle circumference (Cn) andLn assuming that the needle cross-section geome-try can be approximated by half-ellipse (Niinemetset al., 2001). The mass of the sample needles wasdetermined after drying at 70◦C for at least 48 hand needle dry mass per unit total area (MA) andprojected area (MP) was calculated. All remainingneedles on the shoot were also dried and each needleage-class was weighed separately. The total needlearea for every needle age-class (AT,i) was found asthe dry mass of needles in this age-class divided bythe age-class specific estimate ofMA. Projected nee-dle areas (AP,i) were computed analogously using thevalues ofMP. For each shoot projection, the ratio ofsilhouette needle area,AS(φ, ϕ), to total needle area(SS(φ, ϕ); Carter and Smith, 1985; Oker-Blom andSmolander, 1988) was further calculated as

SS(φ, ϕ) = AS(φ, ϕ)−WdLa∑i=ni=0AT,i

(2)

wherei is the needle age andn the number of needleage-classes on the shoot and the productWdLa correctsthe shoot silhouette area for the area of shoot axis. Be-cause the shoot axis is also partly covered by needles,WdLa overestimates the contribution by shoot axis.However, the correction was generally small (0.04–10% of total shoot silhouette area) and was only rel-evant (>5%) for shade shoots, needles of which weresparsely spaced with large needle-to-needle distances.Needle silhouette to projected area ratio (SP(φ, ϕ))was computed in a similar manner using the estimatesof AP,i . One-sided needle area density (ρ, m2 m−3)was estimated for the shoot cylinder with an ellipticalcross-section as

ρ = 2∑i=ni=0AT,i

πWs(90◦,0◦)Ws(0◦,0◦)(Ls − Lk)(3)

where Lk is the length of non-needled part of theshoot axis (Fig. 1). According to the Cauchy’s theo-rems, the mean projection coefficient of a randomlyoriented convex solid is 0.5 and thus, in radiativetransfer equations, the area that is equivalent to that ofa flat leaf is half of the total surface area of a non-flatleaf (Chen and Black, 1992).

We also calculated the ratios of needle silhouette,projected and total needle area and total needle drymass (M) to total shoot axis length andM to the lengthof needled part of shoot axis to characterize the pack-ing of foliar area and biomass in the shoot (Niinemetsand Kull, 1995a). The needle number per unit shootaxis length—needle frequency (KL)—was found asa further estimate of the packing of needles in theshoot. For the latter characteristic, all needles werecounted in 1998. In 1999, the needle number for eachneedle age-class was calculated as the needle mass ofthe specific age-class divided by the average needlemass for this age-class. The measured variables werefurther employed in derivation of the parameters forthe shoot architectural model.

2.6. Shoot radiative transfer model

In our shoot model, the shoots were consideredas an aggregation of shading elements (needles) in aconfined space (shoot volume). Some properties ofthese objects (surface area of needles and dimensionsof the shoot volume) were measured, while the angu-lar distribution and level of clumping in needle spatialdistribution were indirectly estimated by inverting theradiative transfer model.

Approximating the shoot as a cylinder with ellipticcross-section filled by a turbid medium, the probabil-ity of photon interception can be computed accordingto the theory of light penetration in an aggregatedmedium (Nilson, 1971). The model predicts the prob-ability of photon interception,F(φ, ϕ), in the shootvolume as a function of the extinction coefficient(G-function, Ross, 1981), of the Markov coefficientfor spatial aggregation (λ0, Nilson, 1971), of theneedle area density in the shoot volume (ρ, Eq. (3))and of the beam path length in the shoot volumefor specific rotation (φ) and inclination (ϕ) angle,LB(φ, ϕ):

F(φ, ϕ) = 1 − exp[−G(c, φ)λ0ρLB(φ, ϕ)] (4)

The Markov coefficient is a measure of within shootspatial aggregation and can theoretically range from1 (random distribution) to 0 (completely aggregateddistribution). TheG(c, φ) function is given by theellipsoidal leaf angle distribution (Campbell, 1986;Campbell and Norman, 1989), which assumes that theneedle surfaces are distributed parallel to the surface

128 Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140

of an oblate or prolate spheroid:

G(c, φ) =√c2 cos2 φ + sin2 φ

Bc(5)

The angular distribution of the needle surface isreferred to the angles between the normal to the nee-dle surface and the vertical axis when the shoot islying on a plane. The parameterc is the ratio of theellipsoid horizontal to vertical semiaxis (c > 1 for anoblate spheroid andc < 1 for a prolate spheroid) andB is dependent onc as

B = 1 + ln[(1 + ε1)/(1 − ε1)]2ε1c2

; ε1 = √1 − c−2,

if c > 1

B = 1 + sin−1 ε2

cε2; ε2 = √

1 − c2,

if c < 1

(6)

Implicit in the use of theG(c, φ) function as outlinedhere is that the foliage normals have random azimuthaldistribution (Campbell and Norman, 1989).

2.7. Parameterisation of the shoot model and theefficiency of light interception

The shoot silhouette area at each specific angleA′

S(φ, ϕ) was predicted as the probability of photoninterceptionF(φ, ϕ) times the projected area of theshoot cylinderASC(φ, ϕ) and further accounting forthe shading effect of the cylindrical shoot axis with aprojected areaAA(φ, ϕ):

A′S(φ, ϕ)=AA(φ, ϕ)+ F(φ, ϕ)

× [ASC(φ, ϕ)− AA(φ, ϕ)] (7)

To apply the radiative transfer model to our measuredshoot data, the coefficient of the angular distributionof needle normals (c) and the Markov coefficient ofspatial aggregation (λ0) are required. These parame-ters are estimated for each shoot by minimising thesquare error (E) between the shoot silhouette areasmeasured by the photographic technique,AS(φ, ϕ)and predicted by the shoot radiative transfer model,A′

S(φ, ϕ), in various angular directions:

Min E =∑φ

∑ϕ

[A′S(φ, ϕ)− AS(φ, ϕ)]

2 (8)

In our measurements, the inclination angle was con-stant and accordingly,E was minimised for the rota-tion angles only. After determination ofλ0 andc, theshoot model was employed to predict the silhouettearea at different inclination and rotation angles with aresolution of 5◦ for bothϕ andφ and the sphericallyaveraged shoot silhouette area,A′

S, was computed as

A′S = 1

2π

∫ π/2

−π/2

∫ π/2

−π/2A′

S(φ, ϕ) cosϕ dϕ dφ (9)

The integration over the angles was performed with aquadratic spline function going through the predictedvalues ofA′

S(φ, ϕ). The estimate ofA′S was used to

calculate the spherically averaged silhouette to totalarea ratio,S′

S:

S′S = A′

S∑i=ni=0AT,i

(10)

and total needle dry mass per unit shoot projected area:

MS̄ = M

A′S

(11)

Considering a collimated light beam incident on theshoot from a specific view direction, the fraction ofone-sided sunlit area (δS) for a population of shootswith uniform azimuthal distribution was computed asthe ratio of the silhouette area and half of the totalneedle area projected onto an orthogonal plane:

δS(φ, ϕ) = 2AS(φ, ϕ)

ATG(c, φ)= 2SS(φ, ϕ)

G(c, φ)(12)

Implicit in Eq. (12) is that the angular distribution ofneedle normals is the same for the shaded and sunlitneedle area fraction.

We further employed the parameterised shoot radia-tive transfer model to calculate the seasonal averagedaily amount of intercepted quanta per unit needlearea (QSh, mol m−2 d−1, Eq. (A.1)) and to quantifythe shoot efficiency of light interception relative to ahorizontal flat surface,ϑ , as described in AppendixA. In these simulations, the azimuthal angles of theshoot axes were assumed to be uniformly distributedand the shoot inclination angle from the horizontalwas fixed at 0◦ to investigate only the changes in shootmorphology on light interception efficiency. Possiblelimitations resulting from these simplifications areoutlined in Appendix A.

Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140 129

The main emphasis of the current study was toanalyse a large number of shoots from specific lightand nutrient availability conditions (the number ofshoots measured was 167) and we had to reduce theamount of shoot silhouette areas measured from dif-ferent view directions for a single shoot. Thus, theinversion of the shoot model (Eq. (4)), that allowsreconstruction of the light climate on the needle sur-face and that was required for determination of thelight interception efficiency of the shoots in theirnatural location in the canopy (Appendix A), wasaccomplished with only the shoot projections mea-sured from view directions 0◦, 0◦ (also calledAmax

S ,Stenberg et al., 1999) and 90◦, 0◦ (Fig. 1). Never-theless, elegant studies (Oker-Blom and Smolander,1988; Stenberg et al., 1998, 1999) investigating thecorrelations between various shoot projections haveshown that the spherically averaged shoot silhouetteareaA′

S is generally strongly related toAS(0◦, 0◦).Thus, we conclude that the projections measured inour study provide valuable information about shootarchitecture. Moreover, we tested the validity of theinversion routine for derivation of the values ofc andλ0 with sets of shoots ofAbies alba(unpublished dataof Cescatti et al., 1999), projections of which weremeasured from 30 view directions. The model param-eters determined using the projections from 30 viewdirections and from the two view directions measuredalso in the current study were essentially equivalent,indicating the validity of our parameterisation.

2.8. Shoot silhouette area without self-shading

In the shoot model that is based on a turbidmedium analogy, the shading elements do not havefinite sizes. Although needle linear dimensions alterneedle area density, as well as light path length inthe shoot model (Eq. (4)), the model does not allowexplicit consideration of separate effects of needlelength, width and inclination angle (Fig. 1) on thegap fraction in the shoot volume. Such effects maybe analysed with a 3D architectural model, whereneedles in their specific location are simulated usinga Monte Carlo ray-tracing approach (Disney et al.,2000). However, these models are generally of limiteduse for prediction of the stand light climate, becausethey are impossible to invert and therefore, difficult toparameterise.

Given these limitations, we calculate a theoreticalestimate of shoot silhouette area,ATh, without needleoverlap within the shoot based on a simple geometricalapproximation (Eq. (B.2), Appendix B). The estimate,ATh, considers species-specific constraints on needlerotation angle around the shoot axis, but neglects thepossible needle overlap between the adjacent needles,as well as the shading of needles by the shoot axis. Thedifference between the ratioATh/

∑AT,i and SS is

the measure of within shoot shading and the differencebetweenATh/

∑AT,i and needle projected to total

area ratio estimates the inherent limitations on lightuse efficiency resulting from specific average needleangle with respect to shoot axis.

2.9. Foliage chemical analyses

For most samples, needle nitrogen content wasdetermined by an elemental analyser CHN-O-Rapid,Foss Heraeus GmbH, Hanau, Germany) and phospho-rus content by inductively coupled plasma emissionspectroscopy (Integra XMP, GBC Scientific Instru-ments, Melbourne, Australia). For some specimens,standard Kjeldahl digestion was applied for N andP determination. Comparisons of N and P contentsof the control samples measured by various meth-ods indicated that different methods gave essentiallyidentical results for both N and P with the differences≤1% (Niinemets et al., 2001).

3. Results

3.1. Needle morphological characteristics in relationto light and site nutrient availability

Both needle dry mass per unit total (Niinemets et al.,2001) and projected needle area (MP, Fig. 2A) werepositively related to seasonal integrated average dailyquantum flux density (Qint) at the fertile and infer-tile site. As the analyses of covariance (separate slopemodel) demonstrated, the slope of these relationshipswas significantly lower (P < 0.001) at the infertilethan at the fertile site, indicating limited adjustmentof foliar morphology to high irradiance at the infertilesite (Fig. 2A). According to a multiple linear regres-sion analysis, both foliar nitrogen content per unit drymass (NM) and Qint positively affectedMA and MP

130 Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140

Fig. 2. Dependencies on seasonal average integrated quantum flux density (Qint) of: (A) needle dry mass per unit projected needle area;(B) needle frequency; (C) needle area density (Eq. (3)) and (D) needle inclination angle (α, Fig. 1). Separate linear regressions were usedto fit the data from the fertile (filled circles) and infertile sites (open circles). Parameterr2 is the fraction of explained variance andP theprobability that the regression slope is zero. Dashed line denotes the non-significant regression in D.

(Table 1) further confirming the strong effect of nutri-ent availability on needle dry mass per unit area.

Because of increases in needle length (Ln),width (Wn) and thickness with increasing irradiance(Niinemets et al., 2001), the average total area perneedle (AT̄) was positively related toQint at the fertilesite (r2 = 0.62, P < 0.001). In the bog, light didnot significantly alter needle length and moderatelyaffected needle width and thickness (Niinemets et al.,2001), leading to a non-correlation betweenAT̄ andQint (r2 = 0.00,P > 0.8). For all data pooled,AT̄ wassignificantly related to bothQint and NM (Table 1).Apart from the positive effects of nutrient availabil-ity and light on needle linear dimensions (Niinemetset al., 2001), irradiance stimulated more needle growthin width andNM in length such that the ratioLn/Wn,which scales negatively with needle shading by otherneedles and shoot axis, was negatively correlated withQint and positively withNM (Table 1).

3.2. Dependence of shoot morphologyon irradiance

Increases in irradiance led to a larger needle fre-quency (Fig. 2B), greater needle mass (ML, for linearregressions:r2 = 0.63,P < 0.001 for the fertile andr2 = 0.18, P < 0.01 for the infertile site) and totalarea (AL, r2 = 0.48, P < 0.001 for the fertile andr2 = 0.14, P < 0.02 for the infertile site) per unitshoot axis length. Despite the faster increases in drymass than in area (MA = ML/AL, Fig. 2A), largerneedle frequency along with greater area of individualneedles also resulted in a positive correlation betweenneedle area density andQint (ρ, Eq. (3), Fig. 2C).Modification of needle angle with respect to shootaxis (α, Fig. 1) with irradiance (Fig. 2D) was an-other source of changes inρ. The needles were moreclosely positioned with respect to the shoot axis athigher than at lower irradiance (Fig. 2D) and this also

Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140 131

Table 1Foliage architectural characteristics in relation to seasonal integrated quantum flux density (Qint, mol m−2 d−1) and needle nitrogen content(NM, %): results of linear multiple regression analyses on all the data pooled

Dependent variable Intercept P Slopes r2

Qint P NM P

Needle characteristicsAverage total needle area (AT̄, mm2) −127 0.001 0.137 0.005 169 0.001 0.82Needle mass per unit projected area

(MP, g m−2)97.5 0.001 2.77 0.001 46.0 0.001 0.65

Needle length to width ratio (Ln/Wn) 17.4 0.001 −0.458 0.001 24.0 0.001 0.85

Shoot characteristicsNeedle number per unit shoot axis

length (KL , mm−1)3.21 0.001 0.0356 0.001 −1.80 0.001 0.81

Needle area density (ρ, m2 m−3) 0.428 0.2 0.0281 0.001 −0.504 0.02 0.48Needle dry mass per unitS′

S (MS̄, kg m−2) 0.167 0.1 0.0202 0.001 −12.3 0.8 0.73Needle angle with respect to shoot axis (α, ◦) 61.4 0.001 −0.388 0.005 −11.3 0.002 0.21Shoot silhouette to total area ratio for

projection 0◦, 0◦ (SS(0◦, 0◦))0.173 0.001 −0.00247 0.001 0.0345 0.001 0.76

Spherically averagedSS (S′S) 0.188 0.001 −0.00254 0.001 0.0170 0.05 0.77

Markov coefficient of spatial aggregation (λ0) 0.686 0.001 −0.00664 0.001 0.140 0.001 0.62Parameter of the ellipsoidal needle angle

distribution (c)1.01 0.001 0.00157 0.5 0.0980 0.1 0.04

Sunlit fraction of one-sided needle area (δS) 0.690 0.001 −0.00890 0.001 0.0896 0.005 0.75Ratio of measured to calculated (without

within shoot shading) silhouette areas(AS(0◦, 0◦)/Ath)

0.610 0.001 −0.00874 0.001 0.0337 0.2 0.71

Shoot efficiency of light absorptionrelative to a flat surface (ϑ)

0.280 0.001 −0.00324 0.001 0.000262 0.9 0.60

r2 is the fraction of variance explained andP indicates the probability that the specific regression coefficient is zero.

likely contributed to a lower gap fraction and greaterneedle area density at higher irradiance.

Shoot silhouette to total needle area ratio was neg-atively related to irradiance (Fig. 3A), indicating alower fraction of exposed foliar area at higherQint. Inaddition to changes inρ (Fig. 2C), increases in needleclumping (the Markov coefficient of spatial aggrega-tion, Fig. 3B) withQint were also responsible for thenegative relationship betweenQint andSS (Eq. (4)). Incontrast, the ratio of shoot silhouette areasAS(90◦, 0◦)to AS(0◦, 0◦) was essentially independent of integratedquantum flux density (Fig. 3C), suggesting that needlerotation angles relative to shoot axis were not relatedto irradiance. Moreover, the parameter of needle el-lipsoidal distribution,c (Eqs. (5) and (6)), which char-acterises the inclination angle distribution of needlesurfaces, was also independent of irradiance (Fig. 3D).The absolute values of thec-parameter varied from0.8 to 1.5 and thus, the needle inclination angle

distributions derived by inverting the shoot radiativetransfer model (Eq. (4)) were close to spherical for allshoots independently of the shoot light environment.

Similarly toMA andMP (Fig. 2A), needle dry massper unit silhouette area (MS̄, Eq. (11)) was stronglyrelated toQint (Fig. 4A). This relationship was theoutcome of combined alterations in bothMA andSS(Fig. 3A; MS = MA/SS) with irradiance. Given thatthe nutrient content per unitSS is equal to elementcontent per unit dry mass timesMS, the five-foldmodification inMS̄ was also paralleled by similar al-teration in needle nitrogen (Fig. 4B) and phosphorus(Fig. 4C) contents per unit shoot silhouette area.

3.3. Site nutrient availability effects on shootarchitecture

Analyses of covariance (site as the main effect andirradiance as covariate) demonstrated that either the

132 Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140

Fig. 3. Effects ofQint on: (A) measured shoot silhouette to total needle area ratio; (B) the Markov coefficient of spatial aggregation(Eq. (4)); (C) ratio of silhouette areas for the projections 90◦, 0◦ and 0◦, 0◦ and (D) the parameter of the ellipsoidal needle angle distribution(Eqs. (5) and (6)). Linear regressions were fitted to the data from the fertile (filled circles) and infertile sites (open circles).

slopes, which were detected as a significant interac-tion term or intercepts, which were manifested in asignificant main effect in a model lacking the interac-tion term (tested after the interaction term was foundinsignificant), of the light relationships depicted inFigs. 2–4 were generally different between the sites.The slope of the needle area density versusQint rela-

Fig. 4. Needle dry mass per unit spherically averaged shoot silhouette area (A) and needle nitrogen (B) and phosphorus (C) contents perunit spherically averaged silhouette area in relation toQint. Data from the fertile (filled circles) and infertile sites (open circles) were fittedby linear regressions.

tion was larger at the infertile site (Fig. 2C,P < 0.05)and the slopes ofSS (Fig. 3A, P < 0.02) and needleN (Fig. 4B,P < 0.001) and P (Fig. 4C,P < 0.001)content versusQint dependencies were steeper at thefertile site. The slopes of the other relationships werenot statistically different (P > 0.2), but at a commonirradiance, needle frequency (Fig. 2B) and needle

Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140 133

Fig. 5. Needle frequency (A) and shoot silhouette to total area ratio (B) dependence on needle nitrogen content. Data from the fertile(filled circles) and infertile sites (open circles) were fitted by linear regressions.

angle with respect to shoot axis (Fig. 2C) were signif-icantly larger at the infertile (P < 0.001 for both) andthe Markov coefficient (Fig. 3B) was greater at thefertile site (P < 0.001). The variables characterisingneedle angular distributions (Fig. 3C and D) were notsignificantly different between the sites (P > 0.06).

The variableMS̄ (Fig. 4A, Eq. (11)) was not sta-tistically different between the sites (P < 0.1). Thus,the site differences inSS (Fig. 3A) andMA (Fig. 2A)were of similar magnitude (MS = MA/SS).

In addition to site-to-site variability, there wasevidence of a direct control of shoot morphology byfoliar nitrogen contents. For all data pooled, therewas a strong negative correlation between needle fre-quency (Fig. 5A) and needle nitrogen content, whileSS increased with increasing nitrogen content at the

Fig. 6. Correlations of spherically averaged silhouette to total needle area ratio (Eq. (10)) with (A) needle number per unit shoot axislength; (B) needle inclination angle and (C) needle length to needle width ratio. Each symbol represents an individual shoot (filled circlescorrespond to the data from the fertile and open circles from the infertile site). A single regression was fitted to all data in A and C andseparate linear regressions were fitted through the data from the fertile and infertile sites in B. To linearise the relationships, the explainingvariable was log-transformed in panel A.

infertile site (Fig. 5B). Because the needle nitrogencontent andQint were positively correlated (Niinemetset al., 2001),SS versusNM relationship was negativeat the fertile site (Fig. 5B). Nevertheless, when the co-variation betweenNM andQint was accounted for bya multiple regression analysis, there was a consistentpositive effect ofNM on SS(0◦, 0◦) andS′

S (Table 1).

3.4. Effects of needle morphological acclimation onshoot light climate

A simple shoot architecture model (Appendix B)suggested that shoot silhouette area should dependon needle frequency, needle inclination angle, needlelength and width (Eq. (B.2)). We observed a log-linearnegative dependence between needle frequency and

134 Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140

S′S (Fig. 6A). However, the fraction of exposed area

increased with increasing needle angle (Fig. 6B), nee-dle length (r2 = 0.28,P < 0.001) and length to widthratio (Fig. 6C). Thus, as expected, increases in bothneedle length and inclination angle were likely to in-crease the gap fraction in the shoot volume and therebyreduce the shoot needle area density and self-shadingwithin the shoot.

ForS′S versus needle frequency (Fig. 6A) and length

to width ratio (Fig. 6C) relationships, the data fromboth sites were seemingly a part of the same rela-tionship (slopes and intercepts were not significantlydifferent atP < 0.05 according to the ANOVA anal-yses). The separate regressions for fertile (r2 = 0.24for Fig. 6A andr2 = 0.18 for Fig. 6C) and infertile(r2 = 0.07 for Fig. 6A andr2 = 0.15 for Fig. 6C)sites were also statistically significant (P < 0.01),indicating that the trends were valid both within andacross sites. In contrast, the slope of needle inclina-tion angle versusS′

S relation was lower for the bogthan for the forest (Fig. 6B). This possibly illustratesthe circumstance that the effect of needle inclinationangle on the gap fraction depends on both needlelength and needle density.

3.5. Shoot light interception efficiency inrelation to Qint and nutrition

The ratio of measuredAS(0◦, 0◦) to calculated shootsilhouette area without needle overlap (ATh, AppendixB) declined with increasing irradiance (Fig. 7A),

Fig. 7. Dependencies onQint of: (A) the ratio of measured to theoretical silhouette area without self-shading (Eq. (B.2)); (B) sunlitone-sided needle area fraction (Eq. (12)) and (C) the shoot efficiency of light interception relative to a flat surface (Eq. (A.3)). Linearregressions were fitted to the data from the fertile (filled circles) and infertile sites (open circles).

suggesting that modifications in shoot architectureresulted in significant increases in self-shading withinthe shoot. This assumption was further supported bya lower sunlit needle area fraction (δS, Eq. (12)) athigherQint (Fig. 7B).

Sunlit needle area fraction (r2 = 0.75,P < 0.001),as well asS′

S (r2 = 0.70,P < 0.001) were positivelycorrelated with the Markov coefficient for spatial ag-gregation (λ0), but they were negatively related to nee-dle area density (r2 = 0.55 for δS andr2 = 0.58 forS′

S, P < 0.001 for both), indicating that alterations inboth needle packing and clumping modified the frac-tion of exposed foliage area. However, the slopes of theδS andS′

S versus needle area density relationships wereless steep at the infertile site (P < 0.001 according to aseparate slope covariation analysis). The slopes did notdiffer for δS versusλ0 dependence (P > 0.2), but theslope was slightly lower forS′

S versusλ0 (P < 0.05).Similarly to δS and AS/ATh ratio, which charac-

terise the light interception efficiency of all shoots inabove-canopy conditions, the shoot efficiency for lightinterception relative to a flat surface (ϑ , Appendix A)in the shoot natural position was negatively related toQint (Fig. 7C). Nevertheless, the mean irradiance onneedle surface was still considerably higher for theupper canopy needles, becauseϑ varied two-fold, butthe incident integrated quantum flux density four-fold(Fig. 7C).

AlthoughAS/ATh ratio, δS andϑ at both sites con-verged to the same value at high irradiance, the slopesof all relationships depicted in Fig. 7 were larger for

Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140 135

the fertile than for the infertile site (P < 0.005 forFig. 7A andP < 0.01 for Fig. 7B and C), indictinga lower plasticity at the infertile site. Moreover,δSwas also positively related toNM, but other efficiencycharacteristics did not correlate withNM (Table 1).Thus, the differences in the efficiency were less in theshoot natural location in the canopy than expected onthe basis of site differences in shoot morphologicalcharacteristics.

4. Discussion

4.1. Influence of irradiance on needle frequency andsilhouette to total needle area ratio

Our results (Figs. 2–4 and Niinemets et al., 2001)demonstrate that inP. sylvestrisneedle and shootmorphology plastically responded to integrated quan-tum flux density (Qint) at both sites. An increase inneedle frequency with increasingQint as found inour (Fig. 2B) and in other studies (Kellomäki andOker-Blom, 1983; Carter and Smith, 1985; Niinemetsand Kull, 1995a) is a central modification in shootarchitecture that is primarily responsible for a greaterneedle area density within the shoot (Fig. 2C). Giventhat shoot length also increased with increasingQint(unpublished data of Niinemets et al., 1999), these re-sults jointly indicate that enhanced light availabilitiesstrongly stimulate foliage production.

Lower needle frequency and needle area densitywere also manifested in higher shoot silhouette tototal needle area ratio (SS; Fig. 3A) and the sun-lit needle area fraction (Fig. 7B) in low irradiance.Both the spherically averagedSS and sunlit fractionof needle area increased two-fold along the canopylight gradient at the fertile site, implying a similarincrease in the efficiency of light interception in thisshade-intolerant conifer. Although there is evidencethat species may widely differ in maximum valuesof SS (Leverenz and Hinckley, 1990), the range ofvariation in SS along the canopy light environmentsobserved in our study is similar to that determinedfor shade-tolerant conifersAbies amabilis(Sprugelet al., 1996; Stenberg et al., 1998) andPicea abies(Niinemets and Kull, 1995a; Niinemets, 1997; Sten-berg et al., 1999), indicating inherent constraints onstructural adjustment of exposed foliage area.

Apart from the strong influences on whole canopylight interception capacity, modifications in shootmorphological structure to local light environmentalso affect foliage photosynthetic competence. Simul-taneous changes in both needle dry mass per unit area(Fig. 2A and Niinemets et al., 2001) and inSS resultedin a five-fold increase in needle dry mass per unitASfrom the bottom to the top of the canopy (Fig. 4A).This change was paralleled by a strong increase in nee-dle nitrogen (Fig. 4B) and phosphorus (Fig. 4C) con-tents per unit silhouette area. Thus, enhanced needlepacking led to investment of photosynthetic biomassin high light environments where the photosyntheticreturns are the greatest. Our results along with the pre-vious findings (Niinemets and Kull, 1995a; Sprugelet al., 1996; Niinemets, 1997; Stenberg et al., 1998,1999, 2001) conclusively demonstrate that structuralscaling of foliage photosynthetic potentials is an im-portant way to increase the whole canopy carbon gaincapacity. Modelling studies provide evidence thatlight availability-dependent architectural changes inthe foliage light interception efficiency and photosyn-thetic potentials optimise the canopy photosynthesisboth in broad-leaved species (e.g., Gutschick andWiegel, 1988), as well as in conifers (Stenberg, 1996,1998).

4.2. Needle angular distribution and clumpingalong the light availability gradients

We observed that the ratio of rotated shoot pro-jectionsAS(90◦, 0◦) to AS(0◦, 0◦) did not essentiallyvary with light availability (Fig. 3C). Thus, the needlerotation (azimuthal) angles were close to uniformlydistributed and accordingly, the primary assumptionof the use of theG(c, φ) function (Eqs. (5) and (6))was fulfilled. As we demonstrated, using certain phyl-lotactic constraints for leaf rotation angle also resultsin a basically uniform rotation (azimuthal) needleangle distribution (Appendix B). The maintenanceof rotational symmetry along the light gradients inP. sylvestris(Fig. 3C) is similar to previous findingsfor another shade-intolerant coniferPinus contorta(Carter and Smith, 1985), but contrasts with obser-vations in shade-tolerant conifersAbies (Carter andSmith, 1985) andPicea (Carter and Smith, 1985;Niinemets and Kull, 1995a), shoots of which be-come flatter with decreasing irradiance, allowing to

136 Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140

decrease the fraction of shaded foliar area and therebyimprove the efficiency of light interception.

The coefficient of ellipsoidal distribution of needlenormals,c, which is a measure of the distribution ofthe angles between needle surface and the zenith axis(with the shoot lying on a plane, Eqs. (5) and (6)) wasindependent of irradiance (Fig. 3D). This suggests thatthe needle angular distribution is close to spherical inall light conditions. Use of only two projections forderivation of bothc andλ0 (Eq. (4)) may imply thatthe non-correlation with irradiance may arise from er-rors in c estimation. However, the values ofc andλ0estimated for two or 30 view directions were very sim-ilar for the shoots compared and we conclude that theneedle angular distribution may safely be consideredspherical inP. sylvestris.

There was evidence that needle angle with respectto shoot axis (α, Fig. 1) was larger at lower irradiance(Fig. 2D), apparently contradicting the constancy ofc. However,c and α measure different shoot char-acteristics.c is a measure of the distribution of theangles between the normal to needle surface and thezenith axis with the shoot lying on a plane, whileαis a measure of the angle relative to the shoot axis.In particular,α and the needle clumping within theshoot volume are not independent. Closer appressionof needles to shoot axis, which is compatible with alower α, also increases needle clumping and needlearea density within the shoot volume (Appendix B).Moreover, needle twisting and curvature was not ex-plicitly considered inα measurements, butc dependson the overall distribution of needle surfaces andalso accounts for the twisting and curvature effectson the inclination of the surfaces. Studies suggestthat such effects may have an important influence onwithin shoot light climate (Hemmerlein and Smith,1994). Especially when the needles are long, theyare generally also twisted and curved. Like in ourstudy, Kellomäki and Oker-Blom (1983) found nostrong effect of canopy height on the values ofα inP. sylvestris. Yet in P. abies, a positive correlationbetweenα and Qint has been reported (Greis andKellomäki, 1981).

We observed a consistent decrease of the Markovcoefficient of spatial aggregation with irradiance(Fig. 3B), implying that independently of modifica-tions in needle area density and beam path lengthwithin the shoot volume (Eq. (4)), the spatial clump-

ing increased with increasing light availability. Al-though, it has been suggested that needle area densityalone provides enough information for predictionof the shootSS values (Oker-Blom and Smolander,1988; Smolander et al., 1994), our results indicate thatchanges in spatial clumping with light also signifi-cantly alter within shoot light climate and the shadowarea cast by the shoot. Because the needles are gen-erally in fascicles inPinus, the foliage is inherentlyclumped in these species. In a previous study, an in-crease inλ0 has been observed with increasing standdensity in shade-intolerant coniferPinus ponderosa(Law et al., 2001).

4.3. Site effects on shoot morphology versuslight relationships

As discussed in the Section 1, previous researchhas demonstrated inconsistent relationships betweennutrient availability and shoot architecture, possiblybecause modifications in total canopy leaf area indexand changes in light availability generally accompanychanges in nutrition regime. We observed that nee-dle number (Fig. 2B) and inclination angle (Fig. 2D)were larger in the nutrient-limited site, leading to agreater needle area density (Fig. 2C) and a lowerSS(Fig. 3A). In addition, the needles were also moreclumped at the infertile site (Fig. 3B), indicating thatat a common needle area density, the gap fractionwithin the shoot volume was larger at the infertilesite. Although the site differences were often mani-fested in different slope values, foliar nutrient contentswere also directly related to shoot architectural fea-tures (Fig. 5, Table 1), confirming that nutrient avail-ability may strongly control shoot structural acclima-tion. Similarly to our study, Smolander et al. (1989,1990) have reported a negative effect onNM on KLand a positive onα andSS in a study withP. sylvestris.However, the incident irradiances at shoot locationswere not measured in their study.

In another investigation withP. sylvestris, it wasobserved that needle area density should not monoton-ically increase with light availability contrary to ourobservations (Fig. 2C), but may also increase in lowlight (Kellomäki and Oker-Blom, 1983), especially insuppressed trees. Such effect may be an outcome ofa stronger control exerted by low light availabilitieson shoot length growth than on needle formation,

Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140 137

like was the case with low nutrient availability. Thus,arrested shoot length growth in response to environ-mental stresses and soil nutrient availability may be ageneral mechanism affecting the shoot structure.

A simple model further demonstrated (Eq. (B.2),Appendix B) that alterations in needle architecturewith nutrient availability (Table 1 and Niinemets et al.,2001) may provide illuminating insight into site-differences in shoot structure. The ratioATh/

∑AT,i

differs from the projected to total area ratio of singleneedles (α = 90◦, φ = 0◦) by including the shadingresulting from phyllotactic constraints, as well as byconsidering the changes in needle projection resultingfrom different inclination angles of needles relative tothe shoot axis. Thus, the ratioAS/ATh (Fig. 7A) givesan estimate of within shoot shading that is indepen-dent ofα effects on the average projection of singleneedles.

In addition toα, needle length may also substan-tially alter shoot light climate. With increasing needlelength the light intercepting surfaces of common thick-ness are located farther away from the shoot axis andthe gap fraction within the shoot volume increases, andthe shading by the axis decreases. Moreover, needlelength may have indirect effects on the needle incli-nation angle. Thus, site-differences inα (Fig. 2D) andneedle length (Table 1 and Niinemets et al., 2001) mayexplain the site effects on needle clumping (Fig. 3B)and different slopes in the relationships betweenSSand needle area density.

4.4. Changes in light interception efficiency inresponse to light and nutrient availabilities

There were strong negative relationships of sun-lit needle area fraction (Fig. 7B) and seasonal lightinterception efficiency in shoot natural location inthe canopy (Fig. 7C, Appendix A) withQint. Apartfrom the fundamental tradeoff between the potentialpositive effects of enhanced photosynthetic biomassconcentration in high light and low light interceptionefficiency, greater self-shading in the upper canopymay also constitute a significant means to decreasephotoinhibitory damage (Osmond et al., 1999), es-pecially when the average foliar photosynthesis ratesare low. As the studies demonstrate, modification incrown architecture in high light environments mayoften serve as an irradiance avoidance mechanism

(Valladares and Pugnaire, 1999). Because the needlephotosynthetic capacities were strongly curtailed atthe infertile site (Niinemets et al., 2001), we initiallysuggested that the site differences in shoot architec-ture should also be manifested in larger differences inlight interception efficiency in high irradiance, wherethe probability for photoinhibition is the largest.However, there was evidence of lower light inter-ception efficiency in the nutrient-limited site only atlower irradiance (Fig. 7C) contradicting the hypothe-sis of the role of photoinhibition avoidance on shootarchitectural differences between the sites.

Shoot structure effects on needle temperature mayprovide an alternative explanation of similar valuesof SS and light interception efficiency at both sites.Increases in shoot needle area packing increase thethickness of boundary layer around the needles andtherefore, result in higher average needle temperatures(Smith and Carter, 1988; Germino and Smith, 1999).Given that the temperatures in the canopy increasewith increasing height in the canopy (Niinemets et al.,1999), avoidance of excessively high temperaturesmay be a more important goal of acclimation thanphotoprotection.

5. Conclusions

Our measurements demonstrate that shoot architec-ture plastically adjusts to the light gradient within thecanopy in the shade-intolerant coniferP. sylvestris. Asthe result of plastic structural changes, needle lightharvesting efficiency is modified two-fold and foliagephotosynthetic potentials per unit shoot silhouette areafive-fold within the canopy of this species.

Our data along with the literature evidence outlinedalso indicate that changes in shoot length growthpatterns, as well as needle morphology versus lightrelationships may strongly alter the light versus shootstructure relations. In particular, our results show thatthe plasticity for modification of needle and shootcharacteristics declines with decreasing site fertility.Because of limited plasticity, long-term irradiance ver-sus shoot architecture relationships are not necessarilygeneral. Such a limited generality suggests that site-specific parameterisations of shoot structure versusirradiance dependencies are required to scale coniferlight interception and photosynthesis to a stand-level.

138 Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140

Acknowledgements

We acknowledge the financial support from theEstonian Science Foundation (Grants 3235, 4584), theEstonian Minister of Education (Grants 0180517s98and 0281770Bs01) and the Bayreuther Institut für Ter-restrische Ökosystemforschung (BITÖK), Universityof Bayreuth, Germany (BBWFT Grant 0339476C).The enthusiastic contributions of Olevi Kull, MaarikaMäesalu, Anu Sõber (Institute of Ecology, TallinnUniversity of Educational Sciences) are appreciated.We also thank Anne Jõeveer (Tõravere Meteorolog-ical Station, Estonia) for the unpublished data ofglobal solar radiation during 1998 and 1999; KaiKimmel (Endla State Nature Reserve) for allowingus to conduct the research in Endla; Andres Koppel(Estonian Agricultural University) for providing theaccess to the slide-scanner and Indrek Kalamees forpreparing the software for area calculation from thebitmap images.

Appendix A. The shoot efficiency of lightinterception

In the simulations, we assume that the inclinationangle with respect to the horizontal is 0◦ for all shootsand that the azimuthal angle distribution of the shootaxes is random. Calculating the shoot silhouette area(A′

S) for various shoot inclination and rotation anglesby Eq. (7), the mean quantum flux density per unitneedle area (expressed here as half of the total) relativeto the quantum flux density onto an horizontal planeis given for a non-shaded shoot as

R(φ, ϕ) = 2A′S(φ, ϕ)

AT cosφ= 2SS(φ, ϕ)

cosφ(A.1)

The absolute quantum flux densities on the needlesurface depend both on the shading provided by thecanopy, as well as the above-canopy quantum fluxdensity. We used the estimates of fractional transmis-sion of irradiance,Isum (Eq. (1)), determined from thehemispherical photographs for various zenith angles(γ ), to quantify the within-canopy shading. The zenithdistribution of above-canopy diffuse and direct quan-tum flux densities on a horizontal plane,Q(γ ), wascomputed for the period 1 May–31 July 1999 accord-ing to Cescatti (1997), considering the vertical angles

of sun and assuming an uniform distribution of dif-fuse radiation. Thus, the integral of the above-shootquantum flux density from the specific zenith angle isequal toIsum(γ )Q(γ ). Integrating this over the zenithangles, the daily average amount of intercepted quantaper unit one-sided needle area (QSh, mol m−2 d−1) foreach shoot was estimated as

QSh =∫ π/2

0R(φγ , ϕγ )Isum(γ )Q(γ )dγ (A.2)

whereφγ is the shoot rotation andϕγ the inclina-tion angle for the view directionγ (Stenberg et al.,1998) andR(φγ , ϕγ ) for a specific view direction iscalculated by Eq. (A.1). Given that the shoot incli-nation angles may actually increase with increasingirradiance inP. sylvestris(fertile site in our studyand Kellomäki and Oker-Blom, 1983), our estimateof QSh may be biased. However, the lower canopyshoots at the fertile site and the shoots at the infertilesite were essentially horizontal (unpublished obser-vations of Niinemets and Lukjanova 1999). Thus,we conclude that the simplification that all shootsare horizontal may have led to an overestimation ofQSh for the upper canopy shoots at the fertile siteonly.

The shoot efficiency of light absorption relative toa flat surface,ϑ , was further computed as the ratioof QSh and the radiation absorbed by a hypotheticalplanar leaf with an equivalent leaf area in the samelocation:

ϑ = QSh∫ π/20 Isum(γ )Q(γ )dγ

(A.3)

The efficiencies for interception of direct and diffuseirradiance may be separately calculated in an analo-gous manner. However, because both the efficienciesfor interception of diffuse (r2 = 0.92,P < 0.001 fora linear correlation) and direct light (r2 = 0.72,P <0.001) were strongly correlated with that for the totalquantum flux density, only the calculations with totalquantum flux density are presented.

Appendix B. Theoretical estimate of shootsilhouette area without self-shading (ATh)

We follow Wang and Jarvis (1993) to calculate theprojection area of a needle with the needle inclination

Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140 139

angleα and rotation angleφn(AP(α, φn)) and for thedirection of illumination perpendicular to the shootaxis (shoot projectionφ = 0◦, ϕ = 0◦, Fig. 1) as

AP(α, φn) = LnWn

√cos2 α + sin2 α sin2 φn (B.1)

whereLn is needle length andWn needle width and theinclination angle,α, is defined as the angle between theneedle and shoot axis (Fig. 1) and the needle rotationangle,φn, as the clockwise angle between the needleand the direction of illumination. We determined therotation angle distribution for the needles employingthe constraints resulting from the phyllotactic patternin P. sylvestris. In this species, the needles are ar-ranged in spirals with a phyllotactic fraction of 8/21(Vakarelov, 1998). The numerator indicates that eightneedle circuits around the stem axis separate the super-imposed needles and the denominator that there are 21needles in these needle spirals. Thus, the divergenceangle (η), i.e. the rotation angle between two adjacentneedles, is equal to 8/21 times 360◦ (137.14◦, Niklas,1988). Given these phyllotactic constraints, theφndistribution is discrete and uniform with the values ofφn = 0◦, 137.14◦, 274.29◦, ... Because the completionof one phyllotactic cycle requires 21 needles, but insome shoots there were less needles present, we dividethe 21 rotation angles into seven rotation angle classes.For this, the values ofφn > 360◦ were converted toangles ofφn between 0◦ and 360◦ asφn = η− 360f ,wheref is the spiral number. The average needle angleclasses were, thus,φn,k = 17.1◦ + kk=0, ... ,6360◦/7.Using the average measured values ofLn, Wn and�, AP(�, φn) was computed for eachφn,k and theobtained values were averaged (AP(α, φn)). The es-timate of the shoot silhouette area,ATh, was furtherfound as

ATh = AP(α, φn)KL(Ls − Lk) (B.2)

whereLs is the length of stem axis,Lk the length ofnon-needled part of the stem axis andKL is the nee-dle number per unit needled length of the stem axis.The estimate ofATh was independent of whether all21 needle rotation angle classes or only seven aver-age needle angle classes were used in calculations ofAP(α, φn).

References

Albaugh, T.J., Allen, H.L., Dougherty, P.M., Kress, L.W., King,J.S., 1998. Leaf area and above- and below-ground growthresponses of loblolly pine to nutrient and water additions. For.Sci. 44, 317–328.

Baldocchi, D., Meyers, T., 1998. On using eco–physiological,micrometeorological and biogeochemical theory to evaluatecarbon dioxide, water vapor and trace gas fluxes over vegetation:a perspective. Agric. For. Meteorol. 90, 1–25.

Beadle, N.C.W., 1966. Soil phosphate and its role in moldingsegments of the Australian flora and vegetation. Ecology 47,992–1007.

Campbell, G.S., 1986. Extinction coefficients for radiation inplant canopies calculated using an ellipsoidal inclination angledistribution. Agric. For. Meteorol. 36, 317–321.

Campbell, G.S., Norman, J.M., 1989. The description andmeasurement of plant canopy structure. In: Russell, G.,Marshall, B., Jarvis, P.G. (Eds.), Plant Canopies: Their Growth,Form and Function. Society for Experimental Biology SeminarSeries, 31, Cambridge University Press, Cambridge, pp. 1–19.

Carter, G.A., Smith, W.K., 1985. Influence of shoot structure onlight interception and photosynthesis in conifers. Plant Physiol.79, 1038–1043.

Cescatti, A., 1997. Modelling the radiative transfer in discontinuouscanopies of asymmetric crowns. Part I. Model structure andalgorithms. Ecol. Model. 101, 263–274.

Chen, J.M., Black, T.A., 1992. Defining leaf area index for non-flatleaves. Plant Cell Environ. 15, 421–429.

Chen, J.M., Rich, P.M., Gower, S.T., Norman, J.M., Plummer, S.,1997. Leaf area index of boreal forests: theory, techniques andmeasurements. J. Geophys. Res. 102, 29429–29443.

Disney, M., Lewis, P., North, P., 2000. Monte Carlo ray-tracingin optical canopy reflectance modelling. Remote Sensing Rev.18, 163–196.

Germino, M.J., Smith, W.K., 1999. Sky exposure, crownarchitecture and low-temperature photoinhibition in coniferseedlings at alpine tree line. Plant Cell Environ. 22, 407–415.

Greis, I., Kellomäki, S., 1981. Crown structure and stem growthof Norway spruce undergrowth under varying shading. SilvaFenn. 40, 86–93.

Gutschick, V.P., Wiegel, F.W., 1988. Optimising the canopyphotosynthetic rate by patterns of investment in specific leafmass. Am. Nat. 132, 67–86.

Hånell, B., 1988. Postdrainage forest productivity of peatlands inSweden. Can. J. For. Res. 18, 1443–1456.

Hemmerlein, M.T., Smith, W.K., 1994. Structural scaling oflight interception efficiency inPicea engelmanniiand Abieslasiocarpa. Tree Physiol. 14, 1139–1148.

Jordan, D.N., Smith, W.K., 1993. Simulated influence of leafgeometry on sunlight interception and photosynthesis in coniferneedles. Tree Physiol. 13, 29–39.

Kellomäki, S., Oker-Blom, P., 1981. Specific needle area of Scotspine and its dependence on light conditions inside the canopy.Silva Fenn. 15, 190–198.

Kellomäki, S., Oker-Blom, P., 1983. Canopy structure and lightclimate in a young Scots pine stand. Silva Fenn. 17, 1–21.

140 Ü. Niinemets et al. / Agricultural and Forest Meteorology 111 (2002) 121–140

Law, B.E., Van Tuyl, S., Cescatti, A., Baldocchi, D.D., 2001.Estimation of leaf area index in open-canopy ponderosapine forests at different successional stages and managementtreatments in Oregon. Agric. For. Meteorol. 108, 1–14.

Leverenz, J.W., Hinckley, T.M., 1990. Shoot structure, leaf areaindex and productivity of evergreen conifer stands. Tree Physiol.6, 135–149.

Murthy, R., Dougherty, P.M., 1997. Effect of carbon dioxide,fertilisation and irrigation on loblolly pine branch morphology.Trees 11, 485–493.

Niinemets, Ü., 1997. Distribution patterns of foliar carbon andnitrogen as affected by tree dimensions and relative lightconditions in the canopy ofPicea abies. Trees 11, 144–154.

Niinemets, Ü., Kull, O., 1995a. Effects of light availability and treesize on the architecture of assimilative surface in the canopyof Picea abies: variation in shoot structure. Tree Physiol. 15,791–798.

Niinemets, Ü., Kull, O., 1995b. Effects of light availability and treesize on the architecture of assimilative surface in the canopyof Picea abies: variation in needle morphology. Tree Physiol.15, 307–315.

Niinemets, Ü., Oja, V., Kull, O., 1999. Shape of leaf photosyntheticelectron transport versus temperature response curve is notconstant along canopy light gradients in temperate deciduoustrees. Plant Cell Environ. 22, 1497–1514.

Niinemets, Ü., Ellsworth, D.S., Lukjanova, A., Tobias, M.,2001. Site fertility and the morphological and photosyntheticacclimation ofPinus sylvestrisneedles to light. Tree Physiol.21, 1231–1244.

Niklas, K.J., 1988. The role of phyllotactic pattern as a“developmental constraint” on the interception of light by leafsurfaces. Evolution 42, 1–16.

Nilson, T., 1971. A theoretical analysis of the frequency of gapsin plant stands. Agric. Meteorol. 8, 25–38.

Nilson, T., Ross, J., 1997. Modeling radiative transfer throughforest canopies: implications for canopy photosynthesis andremote sensing. In: Gholz, H.L., Nakane, K., Shimoda, H.(Eds.), The Use of Remote Sensing in the Modelling of ForestProductivity: Forestry Sciences, Vol. 50. Kluwer AcademicPublishers, Dordrecht, pp. 23–60.

Norgren, O., Elfving, B., 1994. Needle size and nitrogenconcentration ofPinus sylvestrisandPinus contorta. Scand. J.For. Res. 9, 165–169.

Oker-Blom, P., Smolander, H., 1988. The ratio of shoot silhouetteto total needle area in Scots pine. For. Sci. 34, 894–906.

Osmond, C.B., Anderson, J.M., Ball, M.C., Egerton, J.G.,1999. Compromising efficiency: the molecular ecology oflight-resource utilisation in plants. In: Press, M.C., Scholes,J.D., Barker, M.G. (Eds.), Proceedings of the 39th Symposiumof the British Ecological Society held at the University of Yorkon Physiological Plant Ecology. Blackwell Science, Oxford,7–9 September 1998, pp. 1–24.

Reich, P.B., Walters, M.B., Ellsworth, D.S., 1997. From tropicsto tundra: global convergence in plant functioning. Proc. Natl.Acad. Sci. U.S.A. 94, 13730–13734.

Ross, J., 1981. The radiation regime and architecture of plantstands. Dr. W. Junk, The Hague.

Smith, W.K., Carter, G.A., 1988. Shoot structural effects on needletemperatures and photosynthesis in conifers. Am. J. Bot. 75,496–500.

Smolander, H., Oker-Blom, P., 1989. The effect of nitrogen contenton the photosynthesis of Scots pine needles and shoots. Ann.Sci. For. 46 (Suppl), 473–475.

Smolander, H., Oker-Blom, P., Kellomäki, S., 1990.Typpipitoisuuden vaikutus männyn neulasten fotosynteesiin javerson itsevarjostukseen. (The effects of nitrogen concentrationon needle photosynthesis and within shoot shading in Scotspine). Silva Fenn. 24, 123–128 (In Finnish).

Smolander, H., Stenberg, P., Linder, S., 1994. Dependence oflight interception efficiency of Scots pine shoots on structuralparameters. Tree Physiol. 14, 971–980.

Sprugel, D.G., Brooks, J.R., Hinckley, T.M., 1996. Effects of lighton shoot geometry and needle morphology inAbies amabilis.Tree Physiol. 16, 91–98.

Stenberg, P., 1996. Simulations of the effects of shoot structureand orientation on vertical gradients in intercepted light byconifer canopies. Tree Physiol. 16, 99–108.

Stenberg, P., 1998. Implications of shoot structure on the rate ofphotosynthesis at different levels in a coniferous canopy usinga model incorporating grouping and penumbra. Funct. Ecol.12, 82–91.

Stenberg, P., Linder, S., Smolander, H., 1995. Variation in theratio of shoot silhouette area to needle area in fertilised andunfertilised Norway spruce trees. Tree Physiol. 15, 705–712.

Stenberg, P., Smolander, H., Sprugel, D.G., Smolander, S., 1998.Shoot structure, light interception and distribution of nitrogenin an Abies amabiliscanopy. Tree Physiol. 18, 759–767.

Stenberg, P., Kangas, T., Smolander, H., Linder, S., 1999. Shootstructure, canopy openness and light interception in Norwayspruce. Plant Cell Environ. 22, 1133–1142.

Stenberg, P., Palmroth, S., Bond, B.J., Sprugel, D.G., Smolander,H., 2001. Shoot structure and photosynthetic efficiency alongthe light gradient in a Scots pine canopy. Tree Physiol. 21,805–814.

Vakarelov, I.I., 1998. Changes in phyllotactic pattern structure inPinusL. due to changes in altitude. In: Jean, R.V., Barabé, D.(Eds.), Symmetry in plants. WS series in mathematical biologyand medicine. World Scientific, Singapore, pp. 213–230.

Valladares, F., Pugnaire, F.I., 1999. Tradeoffs between irradiancecapture and avoidance in semi-arid environments assessed witha crown architecture model. Ann. Bot. 83, 459–469.

Veber, K., 1974. Vegetation history of the Endla mire system. In:Kumari, E. (Ed.), Estonian wetlands and their life. Estoniancontributions to the IBP, 7. Valgus, Tallinn, pp. 160–182.

Vose, J.M., Allen, H.L., 1988. Leaf area, stemwood growth andnutrition relationships in loblolly pine. For. Sci. 34, 547–563.

Wang, Y.P., Jarvis, P.G., 1993. Influence of shoot structure on thephotosynthesis of Sitka spruce (Picea sitchensis). Funct. Ecol.7, 433–451.

Whitehead, D., Grace, J.C., Godfrey, M.J.S., 1990. Architecturaldistribution of foliage in individualPinus radiata D. Doncrowns and the effects of clumping on radiation interception.Tree Physiol. 7, 135–155.