Leaf orientation and sunlit leaf area distribution in cotton

AGRICULTURAL AND FOREST METEOROLOGY ELSEVIER

Agricultural and Forest Meteorology 86 (1997) 1-15

Leaf orientation and sunlit leaf area distribution in cotton Somprach Thanisawanyangkura a,b, Herv6 Sinoquet b,,, Pierre Rivet b, Michel Cretenet c, Eric Jallas c a Department of Botany, Faculty of Science, Kasetsart University, 10900 Bangkok, Thailand b INRA Bioclimatologie--PIAF, Domaine de Crouelle, 63039 Clermont-Ferrand Cedex 02, France c Unit~ de Recherche Systkmes de Culture, CIRAD-CA, BP 5035, 34032 MontpeUier Cedex 1, France Received 24 May 1996; revised 21 October 1996; accepted 11 November 1996

Abstract The diurnal leaf orientation behaviour of row-planted cotton plants (Gossypium hirsutum L. cv. 'DES 119') and its relationship to sunlit leaf area distribution at three stages of development were studied in the field. Electromagnetic digitizing was used for plant geometrical structure measurement for three periods of 2h during the day. Cotton leaves showed a diaheliotropic response throughout the day. This heliotropic behaviour varied according to growth stage. In addition to changes in orientation, leaves also moved in space. The distance moved by a leaf between two observation times increased with stage of development, in agreement with petiole and blade lengths. Sunlit leaf area distribution varied according to stage of development. Analysis of interception showed that probability of light interception was greater in the morning and in the afternoon than at noon. Without a diurnal change in canopy structure, cotton plants would intercept less direct radiation in the morning and in the afternoon. Leaf dispersion was regular during the first and last stage of development, but it was clumped during the intermediate stage. Leaf dispersion was more regular in the morning and the afternoon than at noon Leaf dispersion, however, changed primarily with sun direction and not with canopy structure. This would indicate that small changes in leaf location do not significantly affect light interception. Finally, the ecological significance of diaheliotropism and the implications for modelling light interception in cotton plants are discussed. © 1997 Elsevier Science B.V. Keywords: Leaf orientatkm; Sunlit leaf area distribution; Cotton

1. Introduction Change of leaf orientation in response to the change of sun direction occurs in a variety of plants including cotton (Ehleringer and Forseth, 1989). This solar tracking movement, so-called heliotropism, is

* Corresponding author,

generally classified in two types: diaheliotropism and paraheliotropism. Diaheliotropism is a movement making leaves oriented perpendicular to the sun direction: this results in higher leaf irradiance. In contrast, paraheliotropism is a movement where leaves orient themselves parallel to the sun direction: this results in avoidance of high leaf irradiance, decrease in transpiration, and thus in heat loads, and reduction in photoinhibitory effects owing to exces-

0168-1923/97/$17.00 © 1997 Elsevier Science B.V. All fights reserved. PI1 S 0 1 6 8 - 1 9 2 3 ( 9 6 ) 0 2 4 1 7 - 3

2

S. Thanisawanyangkuraet al./Agricultural and Forest Meteorology 86 (1997) 1-15

sive direct radiation (Ehleringer and Forseth, 1989; Fu and Ehleringer, 1989; Yu and Berg, 1994; Isoda et al., 1994; Kao et al., 1994; Wang et al., 1994). Heliotropism has been quantitatively characterized for more than 20years in cotton (Gossypium spp.) (e.g. Lang, 1973; Fukai and Loomis, 1976; Ehleringer and Hammond, 1987; Sassenrath-Cole, 1995). Regarding the two most important species for agriculture, G. hirsutum is reported to be diaheliotropic whereas G. barbadense is not heliotropic (Ehleringer and Hammond, 1987; Sassenrath-Cole, 1995). Researchers wondered about the ecological significance of diaheliotropism, especially at noon when radiation loads on leaves are heavy. In this case, leaves with high irradiance do not improve photosynthesis because they are light saturated, and they are likely to be subject to light, temperature and water stress. Heliotropism is strictly defined as changes in leaf orientation. However, light interception depends not only on leaf angles but also on the spatial distribution of leaves within the canopy. Because consequences of heliotropism on plant physiology or ecology are based on light interception, the effect of heliotropism on light interception could also be studied in terms of changes in leaf locations during the day. At the present time, there is no information about the relationship between diurnal changes in leaf orientation, in spatial leaf arrangement and sunlit leaf area throughout the growing season. A study on changes in leaf orientation and spatial leaf arrangement requires measurement of the threedimensional (3D) distribution of the leaves, which is difficult to perform (Russell et al., 1989; Sinoquet and Andrieu, 1993). Among recently developed 3D digitizing devices, electromagnetic digitizing has shown good accuracy (better than _+ 1 mm in the laboratory) and more advantages compared with other 3D digitizing systems (i.e. articulated arms (Lang, 1973); ultrasonic digitizer (Sinoquet et al., 1991)): it measures both spatial coordinates and orientation angles, and it is insensitive to weather conditions and masking (Moulia and Sinoquet, 1993). The objective of this study is to characterize diurnal changes in leaf orientation and leaf location within the canopy using an electromagnetic digitizing system, and to quantify sunlit leaf area distribution of cotton plants as affected by dynamic changes

in plant geometry during the day and during the growing season.

2. Materials and methods 2.1. Field growing condition

Cotton plants (Gossypium hirsutum L. cv. 'DES 119') were sown on 3375 m 2 of sandy loam soil with pH 8 at the experimental field of CIRAD-CA, Domaine de Lavalette, Montpellier, France (43.62°N, 3.92°E) on 6 May 1995 in NE-SW row direction. Row spacing was 0.80 m and between-plant spacing was 0.18 m after thinning at the four-leaf stage. Pre- and post-emergence herbicides Treflan ® (Trifluraline) and Califor G ®, respectively, were applied. Hand weeding was also done. Compound fertilizer was applied at 400kgha-1 of 0:25:25 (N:P:K) before sowing, and an additional 30 kg ha-l of 46:0:0 (N:P:K) after plant emergence. The crop was irrigated by sprinkler when soil moisture content reached 50% by weight. Insecticide treatments were similar to farm practice in the region. During the growing period, the weather was cold in May (average temperature 16°C) and warmer in June (20°C) and July (25°C). Total precipitation was 48.5 mm. For the measurement days when the leaf area index (LAI, total leaf area per planted area) was 0.12 and 1.09, there were some cloudy periods in the afternoon but the sky was clear when the LAI was 2.84. 2.2. Plant geometrical structure measurement

Electromagnetic digitizer (3Space ®, Fastrak ®, Polhemus, Inc., 1993) and data acquisition software DiplAmi (Sinoquet and Rivet, 1997) were used. The digitizing device consists of a system electronic unit (SEU), one (extendable to four) receiver(s), a single transmitter, and a power supply. The transmitter generates low-frequency magnetic fields which induce currents in coils included in the receiver. Values of induced currents depend on the location and orientation of the receiver in the active volume around the magnetic source. The cartesian coordinates, i.e. x, y and z, and Euler orientation angles, i.e. azimuth, elevation, and twist angles, of the receiver are

S. Thanisawanyangkura et al. /Agricultural and Forest Meteorology 86 (1997) 1-15

,,±z [ l

--~

, !

m

'~ -~b

;....;4 /01/ ~,'

I

West

North- (--) - /

I \

I / ~ . ~ ' - ~ " " ~ ...... ~- East

-~ ~¢i~....... ~,r/'~ ~

I

, "

.~.~..~

~

y

, t~ast

South Fig. 1. Schematic diagram showing angular determination of leaf orientation and sun positi~m: 4~n, leaf normal azimuth; O~n,leaf normal inclination; 4h, azimuth of midrib; al, inclination of midrib; 01, leaf twist; ~, normal of midrib; ~', normal of leaf blade; Az, solar azimuth; h, solar elevation; /2, solar direction. The leaf changes from position a to position b by twisting without change in midrib azimuth and midrib inclination. Thus, ~' changes from ~'. to n'b and 01 is the difference between ~. and n'b angles.

determined with a resolution of 0.0005 cm cm-1 of range, and 0.025 ° (when the receiver is located within 76 cm of the transmitter). Operation with a separation of 305 cm between the transmitter and the receiver is possible with reduced accuracy (Polhemus, Inc., 1993). Digitized positions on cotton plant were the nodes on main stem and branches, the insertion of the leaves with petiole orientation, and the proximal and distal tips of the midrib with blade orientation and midrib direction. Azimuth of mid rib (~b 1) is defined as the projection of the mid rib onto the horizontal plane relative to the north direction (Fig. 1). Inclination of mid rib ( a l) is defined as the angle between the mid rib and the horizontal plane. Twist of leaf blade (0~) is a rotation angle around mid rib. Twist angle is zero when the plane of the leaf blade includes a horizontal axis perpendicular to the midrib direction. Digitizing was conducted three times a day by assuming that the leaves did not change their orientation significantly during the 2 h measurement period: in the morning (07:00-09:00h True Solar Time), at noon ( l l : 0 0 - 1 3 : 0 0 h ) , and in the afternoon (15:0017:00h) every 5 days. After some practice with digitizing, two people were able to measure 190 leaves

3

within about 2 h with minimal disturbance of the canopy. In this study, three growth stages were compared when the LAI was 0.12, 1.09, and 2.84 and total number of leaves per plant was 6, 35, and 63 leaves, respectively. At the LAI 0.12 stage, four plants, the growth and development of which corresponded to those of mean plants, were visually selected at the centre of the plot and considered as representative of the crop. Three of the same four plants were studied at the LAI 1.09 and LAI 2.84 stages because of time limitation for digitizing owing to a larger number of leaves. Such small numbers of measured plants were chosen to reduce the digitizing duration and thus to avoid significant leaf movements during the measurement periods. At each growth stage, leaf area was estimated from an allometric relationship between mid rib length (L, cm) and leaf area (S, cm 2) measured by a scanner (Hewlett Packard Scanjet IICX ®, Hewlett Packard Inc., Palo Alto, CA, USA): SLAI0.12 = -- 0.309 + 0.926L 2 ( n = 75, r 2 = 0.930** ) SEA11.09 =

--4.382

+ 0.895L 2

( n = 100, r 2 = 0.939** ) SLAI 2.84 =

--

1.237 + 0.857L 2

( n = 46, r E = 0.958* * )

2.3. Methods of data analysis 2.3.1. Leaf orientation Angular determination is defined as shown in Fig. 1. The azimuth of leaf normal (~b,) is the angle between the projection of the leaf normal onto the horizontal plane and the north direction. ~b, was calculated as follows: (~n = ( ~ l -

arctan(tan

0,/sin

o/1)

where (kl is the azimuth of mid rib, o~1 is the inclination of mid rib, and 01 is the twist of leaf blade around mid rib. The inclination of leaf normal (c~n) is the angle between the leaf normal and the vertical axis. It was calculated as follows:

~n = arccos(cos 0,cos '~1)

4

S. Thanisawanyangkura et al./Agricultural and Forest Meteorology 86 (1997) 1 - 1 5

2.3.2. Cosine of the angle of incidence The cosine of the angle between leaf normal and sun direction (cos /3) indicates how the leaf faces the Sun, and therefore it may be used as an indicator for heliotropic behaviour (Ehleringer and Forseth, 1989): a value of cos/3 close to 1.0 indicates a leaf with a strong diaheliotropic behaviour. In contrast, a leaf with cos/3 close to zero shows strong paraheliotropic behaviour. The cosine (cos/3) is calculated as (Ross, 1981) cos/3 = cos ansinh + sin ancos hcos( Az - thn) where h is the solar elevation and Az is the solar azimuth. In most radiative transfer models, the distribution of leaf azimuth is assumed to be uniform. To test the effect of this assumption on cos/3, a value of cosine of incidence (cos/3)u with uniform leaf azimuth distribution was also calculated for each individual leaf as follows (e.g. Sinoquet et al., 1993). When solar elevation is higher than leaf inclination: (cos/3)u = cos ausin h When solar elevation is less than leaf inclination: (cos/3)u = cos tznsin h(2{[arccos( - t a n h / t a n an) ] -

tan [ arccos( - tan h / t a n a n) ] }/ 7r - 1)

For a population of N leaves, the average values of cos/3 and (cos/3)", called G-functions (see Ross, 1981), were computed by weighting individual cos/3 and (cos/3)u by leaf area S: G = ~ S,- cos/3i i=l

S~

i=l

2.3.4. Sunlit leaf area and light interception From digitizing data, the coordinates of leaf elements in space were recalculated to create an image of the plant with actual leaf orientation and leaf position. Sunlit leaf area was estimated from pictures of the plants viewed in the sun direction, made by the 'smooth curve' function of Microsoft Excel ® Version 5.0 (Microsoft Corp., Redmond, WA, USA). Leaves seen on the picture were coloured by hand to measure the sunlit leaf area Sp with a Li-Cor 3100 Leaf Area Meter ® (Li-Cor Inc., Lincoln, NE, USA). On the plant pictures, a reference area SR was defined by the soil surface area occupied by a plant. It was estimated as the smallest ellipse including the leaf area projected onto the soil surface. Plant pictures allowed us to derive light interception probability from sunlit leaf area estimated from digitizing (PDigit): PDigit = S p / ( S R s i n

h)

PDigit was compared with two interception models, i.e. the Beer-Lambert law model (PBeor) and binomial model (PBinom), to quantify the pattern of leaf dispersion within the reference area:

Boor: 1 exp[ i cos l iS,, Rsinh,] N eBinom ~--- 1 - 1--I [1 - cos/3iSJ(SRsin h)] i=1

i

G ° = E Si(cos/3)

where D a_ b is the change in blade location between observation times a and b, and (Xa,Y~,Z a) and (Xb,Yb,Z b) are the spatial coordinates of the central point of the leaf blade at observation times a and b, respectively.

S, i

2.3.3. Diurnal change of leaf blade location To quantify the diurnal change in leaf blade location, motion of the central point of leaf blade between two observation times was calculated as follows: O a _ b --'- ~ ( X a -- X b ) 2 --~ ( V a -- y b ) 2 dr- ( Z a -- Z b ) 2

where N is the leaf number and S i is the area of leaf i. PBoer estimates the probability of light interception for random leaf dispersion whereas PBinom corresponds to a case of regular dispersion. In the latter, leaf dispersion is explicitly related to leaf size, meaning that a leaf does not shade itself. The ratio PBinom/PBeer therefore indicates departure from randomness owing to leaf size distribution. Otherwise, the ratio PDigit/PBinom may be regarded as a m e a -

s. Thanisawanyangkura et al. /Agricultural and ForestMeteorology 86 (1997) 1-15 sure of leaf dispersion owing to non-randomness of leaf locations. 2.3.5. Comparative effects of plant structure variation and sun position variation Cos/3, sunlit leaf area, and leaf dispersion were computed for each combination of plant geometry and sun direction (i.e... in the morning, at noon and in the aftemoon). The cross comparison allowed the distinction between effects of change in plant structure and sun direction on the attributes of light interception.

50 45 -

J ~4o

SunPoeltion Morn~g

a,

5

SunPo~i~on Noon

Sun F~ition Afternoon

.:

~,

".;

0

b,

Sun Fos~n Morning

45

SunFosit~n Noon

Sun Position Afternoon

_~~4o t,*,m~,g "~

025 E

~20

S~'~

Noon

~/

3. Results 5

3.1. Leaf area development of cotton plant

0

; ...........

.._.

,

,

* o,

Cotton plants grown after early planting in May developed well under the conditions in the North Mediterranean area, although their growth and development was relatively slow at the beginning because of the low temperature (Fig. 2). The LAI reached 0.12 at 38days after emergence (DAE). The vegetative branches started development at 43 DAE when LAI was 0.4. Then, plants developed indeterminately numbers of vegetative and reproductive branches with numerous leaves. The LAI attained 4.0 at 70 DAE. At this time, plants had more than 100 leaves.

50 45

== ~ Eo 25

5

....... C,

S u n - ~ i l ~ ~ - SunPc6i~ Morning Noon

Sun~ i l ~ o n Afternoon

.............

]

Morning

.

o Leaf Azimuth C)

Fig. 3. Leaf azimuth distribution o f cotton at three times o f day

and three stages of development:(a) LAI 0.12; (b) LAI 1.09; (c) LAI 2.84.

Capsule Formation /

5

~,

F%:= u0

Flowering

Appearance

t

4

3.2. Leaf orientation distribution

o/ ~

•

/

Vegetative

2

BroncO1 I

0

--

33

"l

38

43

_/

r

r

I

r

r

48

53

58

63

68

73

Days after germination Fig. 2. Leaf area development o f cotton (Gossypium hirsutum L.

cv. 'DES 119').

3.2.1. Leaf azimuth distribution The distribution of leaf azimuth angles changed significantly during the day. Leaf azimuths tended to lead the sun position in the morning, but lagged at noon and in the afternoon during the early stage of development (Fig. 3). At the LAI 0.12 stage, cotton plant maintained 46% of its total leaf area with a similar leaf azimuth (i.e. in the same 30 ° class of azimuth) from morning to noon and 74% from noon to afternoon. The proportion of leaf area with similar leaf azimuth showed little difference at LAI 1.09 and LAI 2.84

6

S. Thanisawanyangkura et al./Agricultural and Forest Meteorology 86 (1997) 1-15

stages (55% and 46% from morning to noon, and 53% and 52% from noon to afternoon, respectively). Change in leaf azimuth from morning to noon was therefore greater than the change from noon to afternoon, particularly at the LAI 0.12 and LAI 2.84 stages. It was also noted that the degree of change in leaf azimuth distribution from noon to afternoon tended to increase as the plant developed more leaves.

NORTH

-~---

'

' ~e

'*

'

Morning Oderltation

SOUTH NORTH

3.2.2. Leaf inclination distribution Cotton leaves changed their inclinations significantly ( P < 0.01) during the day. At the more advanced stages, cotton leaves tended to stand more erectly in the morning (average leaf inclination of 25 °, 40 ° and 42 ° at LAI 0.12, LAI 1.09, and LAI 2.84 stages, respectively) and at noon (25 °, 33 ° and

.,~~ .

•

Sun Pos~ic~

50

~4° / ~ Noon

7'

0"-10=

+

/

a.

i

J 20%30°

30o-40o

40°-50o

50o-60*

60--70*

70o-80°

Noon

~!

10*-20*

20*-soo

30--40.

40o-soo

Sun I:~ #ion Noon

50*-60o

60*-70.

70"-80-

Sun Posit~ Morn~g

I

i

C.

Noon

Afternoon

~

Morning

. . . . . . . . . 10*-20" 2o*-3o*

30--40*

40.-50 ~

Leaf Inclln&tlon

~

ii i i

"" 50.-60 o

80.-90.

i

Sun Position

40

0~-10o

C, I • Afternoon Orientation

curve) as a function of leaf orientation (a) in the morning, (b) at noon, and (c) in the afternoon, at LAI 0.12 stage. Each point represents the orientation of one leaf described by its inclination angle and azimuth angle.

Afternoon

20

~= to

EAST

Fig. 5. Distribution of cotton leaf orientation and lines of cosine of incidence angle, cos/3, of 0.6 (dashed curve) and 0.8 (continuous

Sun Position Morning

Afternoon

a o

~

80".90-

3O

0o-10*

~

. . . . .

10.-20 o

40

o

"

Mornbg - Sun Position ~f~lr Af. . . .

Morning

Sun FosiioI1 Noon

•

i

EAST

• Noon Orientation

.

=

-s0

, ,

Sun Position

Noon

= ~

i

,

NORI~

60

o

, ,

60--70 o

70o.80*

41 ° , respectively), whereas there was virtually no change in the afternoon (38 °, 34 ° and 37 °, respectively). However, there was generally no direct relation between solar elevation and leaf inclination distribution, except at noon at the LAI 0.12 stage (Fig. 4). It was also noted that nearly one-third of the total leaf area remained at a similar inclination (i.e. in the same 10 ° class) from morning to noon, and also from noon to afternoon for all three stages of development.

80~.90*

(*)

Fig. 4. Leaf inclination distribution of cotton at three times of day and three stages of development: (a) LAI 0.12; (b) LAI 1.09; (c) LAI 2.84.

3.2.3. Distribution o f cosine o f incidence angle Figs. 5 - 7 show the effect of the combination of leaf inclination and leaf azimuth on the cosine of incidence angle. Most cotton leaves had a c o s / 3

S. Thanisawanyangkura et al. / Agricultural and Forest Meteorology 86 (1997) 1-15 NOFm~

the morning and in the afternoon. In comparison with a uniform distribution, the actual azimuth distribution gave about a 9 - 1 6 % advantage to the cotton plant in cos/3 in the morning and in the afternoon, but only 3 - 7 % at noon. As cos/3 values result from the relative geometry of both the sun position and the leaf orientation, diumal changes in cos/3 may be due to changes in canopy geometry and sun course. Assuming a situation with a change in the sun direction without any change in canopy geometry, or change in canopy structure without any change in sun direction, allows us to distinguish the change in cos/3 owing to sun position or leaf orientation (Table 2).

OO

• 3~; .

¢' • I ) •

~

FAST

a.

7~" "2,

• MorningOrientation

FAST

-

• NoonOrientation

b.

7

) NORPH

!' , ",~

FAST

EAST

~/ ~/~C.

•Afternoon Orientation

1

a.

• MorningOrientation]

b.

• Noon Orientation

SOU134 NORTH

Fig. 6. Distribution of cotton leaf orientation and lines of cosine of incidence angle, cos/3, of 0.6 (dashed curve) and 0.8 (continuous curve) as a function of leaf orientation (a) in the morning, (b) at noon, and (c) in the afternoon, at LAI 1.09 stage. Each point represents the orientation of one leaf described by its inclination angle and azimuth angle.

greater than 0.6 at each period o f day for all three stages of development. At the L A I 0.12 stage when the leaves did not suffer from mutual shading, cotton leaves oriented diaheliotropically at noon, but less in the morning and in the afternoon. ']'he leaves showed non-significantly different orientation behaviour in the morning and in the afternoon ( P > 0.05) (Table 1). This pattern of diurnal heliotropic behaviour still remained at the L A I 1.09 and L A I 2.84 stages, where mutual shading became more important. Table 1 shows that if the leaves were distributed uniformly in azimuth, their cos/3 values (i.e. Gfunction G u) would have been smaller than c o s / 3 obtained for the actual distribution, particularly in

"~

SOLm4 NORTH

••

~ "

•

FAST

C. • AfternoonOrientation

SOUTH

Fig. 7. Distribution of cotton leaf orientation and lines of cosine of incidence angle, cos/3, of 0.6 (dashed curve) and 0.8 (continuous curve) as a function of leaf orientation (a) in the morning, (b) at noon, and (c) in the afternoon, at LAI 2.84 stage. Each point represents the orientation of one leaf described by its inclination angle and azimuth angle.

8

S. Thanisawanyangkura et a l . / A g r i c u l t u r a l and Forest Meteorology 86 (1997) 1 - 1 5

Table 1 The cosine of the angle of incidence of the Sun on the cotton leaves, cos/3, and on leaves with a uniform azimuth distribution, (cos/3) u, and the enhanced light interception owing to cotton leaf orientation at three times of day and at three stages of development when LAI was 0.12, 1.09 and 2.84

LAI 0.12 cos/3 (cos/3) ~ Advantage by orientation LAI 1.09 cos fl (cos/3) ~ Advantage by orientation LAI 2.84 cos/3 (COS /3 ) ~ Advantage by orientation

Morning

Noon

Afternoon

0.697 ~ 0.539 c + 15.8%

0.862 b 0.819 b + 4.3%

0.682 a 0.516 ~ + 16.6%

0.597 ~ 0.505 b +9.2%

0.813 ~ 0.745 d +6.8%

0.613 ~ 0.522 b +9.1%

0.645 a.b 0.496 ~ + 14.9%

0.685 a 0.657 ~,b + 2.8%

0.616 b 0.511 ~ + 10.5%

would have led to a decrease in cos/3 of 15-20% and 35-50%, respectively. This means that, if there was no leaf movement between the afternoon and the next morning, the orientation of cotton leaves would have been significantly less favourable to intercept solar direct beam. The afternoon sun direction showed a symmetrical behaviour with greatest values of cos/3 for the afternoon leaf orientation: reduction in cos/3 owing to the morning and noon leaf orientation would have been 30-45% and 15-25%, respectively. Otherwise, the greatest values of cos/3 were for the noon sun direction, but they were much less influenced by the leaf orientation.

3.2.4. Diurnal change of leaf blade location The location of leaf central point varied significantly ( P < 0.01) during the day depending on the

The cos/3 and (cos/3) ~ values of each stage of development followed by the same letter are not significantly different ( P > 0.05), tested by the Student-Newman-Keuls method.

100 9O

Diurnal Total Movement .... From Momlrtg to Noon . . . . . . . From Noon to Afternoon

8o

For the morning sun direction, the morning leaf orientation allowed the greatest values of cos/3, whereas the noon and afternoon leaf orientation

~ ~° 80

~

so

"6

40

~

30

.......

a. L A I 0 . 1 2

0

Table 2 The cosine of the angle of incidence of the Sun on the cotton leaves, cos/3, calculated for the leaf angles measured at three times of day and for three positions of the Sun Orientation

LAI 0.12 Morning Noon Afternoon LAI 1.09 Morning Noon Afternoon LA12.84 Morning Noon Afternoon

0-2

I

4-6

Noon

Afternoon

0.697 b 0.578 ~ 0.339 d

0.866 ~ 0.862 a 0.731 b

0.380 a 0.507 ~ 0.682 b

0.597 ~,d 0.507 d,e 0.394 f

0.712 b 0.813 a 0.772 a,b

0.441 ~'f 0.515 d,e 0.613 c

50 ¢ 45

Diurnal Total Movement .... . . . . . . .

30

.:

0.401 d 0.487 c 0.616 a

Results are given for the three development stages when LAI was 0.12, 1.09 and 2.84. The results for the leaf angle appropriate to the Sun's position are highlighted by printing in italics. The cos/3 values of each stage of development followed by the same letter are not significantly different ( P > 0.05), tested by the StudentNewman-Keuls method.

;t"~

2-4

0-2

0.648 a 0.685 a,b 0.724 b

8-10

~,

.a= 35

Morning

6-8

Distance of Leaf Movement (cm)

40

Sun position

0.645 ~ 0.522 ~ 0.406 d

p

24

From Morning to Noon From Noon to Altemoon

\\

4~

64 8-10 16-12 12-14 Dlmtance of Leaf Movement (cm)

14-16

16-18

18-20

5o 45 40 "¢ 35 o 30 25

_~ ~I :~

Diurnal Total Movement .'~ .'

....

:' / . 4 \ // "\

2010150 5 F:: t

"\

0*2

4-6

From Morning to Noon

. . . . . . . From

~

Noonto Afternoon

,

12-14 16-18 20-22 Dlmnee of Leaf Movement (cm)

8-10

.84 : ~ - - " ~ .

24-26

26-30

Fig. 8. Diurnal change in leaf blade location at three stages of development: (a) LAI 0.12; (b) LAI 1.09; (c) LAI 2.84.

S. Thanisawanyangkura et al. / Agricultural and Forest Meteorology 86 (1997) 1-15

9

1.09 and 3697-4029 cm 2 at LAI 2.84. Although S R increased with LAI, the ratio of S R to the leaf area of a plant decreased, from 1.74 at LAI 0.12 to 1.25 at LAI 1.09 and 0.98 at LAI 2.84. This means that the efficiency of space occupation by the leaves of a plant decreased with the stage of development. The fraction of sunlit leaf area varied significantly ( P < 0.01) during the day according to the stage of development (Fig. 9). Sunlit leaf area increased from morning to noon, and decreased from noon to afternoon (Table 3). However, sunlit leaf area in the morning did not differ significantly ( P > 0.05) from that in the afternoon at all three stages. It was noted that sunlit leaf area decreased from LAI 0.12 stage to LAI 1.09 stage but increased from LAI 1.09 stage to LAI 2.84 stage, with a more uniform distribution during the day. Table 3 shows that sunlit leaf area responded to the combination of sun direction and plant geometry in a similar way to cos ft. For the morning and

period of the day and stage of plant development (Fig. 8). Morning motion of the leaf blade (i.e. change in location between morning and noon) was 2.4cm at LAI 0.12 stage, which differed significantly ( P < 0.01) from that at the LAI 1.09 stage (4.7cm) and LAI 2.84 stage (5.0cm). Afternoon motion was not so different from that in the morning: 2.7cm, 3.2cm and 4.1cm when the LAI was 0.12, 1.09 and 2.84, respectively. Total dally changes in leaf blade location were therefore 5.1 cm, 7.9 cm and 9.1 cm at the LAI 0.12, 1.09 and 2.84 stages, respectively.

3.3. Sunlit leaf area and probability of light interception

3.3.1. Sunlit leaf area The soil surface a r e a S R occupied by a plant varied according to its stage of development, from 255-345 cm 2 at LAI 0.12, to 1962-1991 cm 2 at LAI

Sun PoslUon

Morning

Afternoon.

Noon

-20

-20

-20

-10

-100 0 cm

:

~

I0

10

lO

b.

cm

C.

20

2O -20

-10

0

10

-20

20

-10

0

10

20

-50

-50

cm

2o

-20

-10

0

-25

0

10

20

25

so

30

60

-50 -25

0

~ r "

0 cm

"

~

(~,~ 25

25

25

d. crn

-50

f.

e.

50 -25

0

50

25

-90

~0 0

',

-50

50

50

cm

-2S

0

25

50

-90

-90

-60

.so

-30

cm

o

g"

30 -60

cm -30

0

C~,

-30 cm

4 3m

30

30

cm

-50

i.

30

-60

-30

0

30

60

-60

cm

-30

0

Fig. 9. Reconstruction of le:ff area distribution of cotton projected perpendicular to solar beam direction at three times of day by digitizing data at three stages of development: (a-c) LAI 0.12; (d-f) LAI 1.09; (g-i) LAI 2.84.

10

S. Thanisawanyangkura et al. / Agricultural and Forest Meteorology 86 (1997) 1-15

Table 3 The sunlit leaf area of cotton as a percentage of total leaf area calculated for the plant structures measured at three times of day and for three positions of the Sun Plant

Sun position

structure

Morning

LAI 0.12 Morning Noon Afternoon LAI 1.09 Morning Noon Afternoon LAI 2.84 Morning Noon Afternoon

Noon

Afternoon

60.8 s,t,. 81.1 w 54.9 p,q,r,s 73.3 v 34.6 c,d,e,f,g,h,ij 63.6 t,u 36.8 d,e,f,g,h,i,j,k 30.7 b,c,d,e,f 24.1 a,b

33.9 c,d,e,f,g,h,i 42.5 h,i,j,k,l,m,n 55. 7 q,r,s,t

37.1 d,e,f,g,h,i,j,k,l 25.5 a,b,c 42.2 g,h,i,j,k,l,m,n 32.9 c,d,~,e,g,h 42.3 h,ij,k,l,m,n 37.9 d,e,f,g,h,i,j,k,I

46.4 k,l,m,n,o,p,q 43.9 i,j,k,l,m,n,o 39.2 e,f,g,h,i,j,k,l,m 48.0 m,n,o,p,q 32.9 c,d,e,f,~,n, 50.6 n,o,p,q....

33.5 c,d,~,f.g,h 39.5 f,g,h,i,j,k,l,m 49.6 n,o,p,q,r

Results are given for the three development stages when LAI was 0.12, 1.09 and 2.84. The results for the plant structure appropriate to the Sun's position are highlighted by printing in italics. The values of sunlit leaf area percentage followed by the same letter are not significantly different (P > 0.05) for all pairwise comparison. afternoon sun directions, the m o r n i n g and afternoon plant geometry allowed the largest fraction of sunlit area, respectively. A d o p t i n g an afternoon leaf distrio

Og(a)

o

08

b u t i o n for the m o r n i n g sun direction or a m o r n i n g plant structure for the afternoon sun w o u l d have reduced the fraction of sunlit area by 3 0 - 4 3 % . Similar to cos t , the fraction of sunlit area for the n o o n sun direction was m u c h less influenced b y changes in plant geometry.

3.3.2. Probability o f light interception F r o m the digitizing data, the probability of light interception within the reference area S R was greater in the m o r n i n g and in the afternoon than at noon, at all three stages of d e v e l o p m e n t (Fig. 10). The same b e h a v i o u r was found with light interception probabilities calculated b y the B e e r - L a m b e r t model (PBeer)

and the b i n o m i a l model (eBinom) (Fig. 10). The probability of light interception was also affected b y diurnal changes in plant geometry. Fig. 10 also shows that plant structure in the m o r n i n g allowed more light interception in the m o r n i n g than plant structure at n o o n and in the afternoon, at all three stages of development. In a symmetrical way, probability of light interception in the afternoon sun direction was the highest for the afternoon plant geometry. Plant structure at n o o n showed only little advantage for light interception at noon.

Br(a)

oTBn(a)

08

06

06

06

04

0.4

0.4

~

0.0

o

Morning

Noon

1oO8tDg(b) -I

Noon

Noon

Afternoon

Noon

Aftsmoon

Morning

lO I Bn (c)

o.,

0.4

0.2 -

09

0.0 .

0.0

i

Alternoon

plant StrlAllamoon

I

Noon

I

Afternoon

PlantmM~g St ....

[] A~n

o.,t

0.4

02

Altemoon

1

'

Morning

,.o ~ Br (c)

Noon

I ,f Noon

lO (] b Bn ).

0.4

Morning

I

Morning

0.0

Nternoon

"! o T Dg (c)

~. . . . mNoon r l Arernoon

0.0

Morning

1° T Br (b)

0.0

Morning

0.0

A~emoon

P

08

0.0

Morning

Noon

Alfernoon

Morning

Nooe

Anemooe

Fig. 10. Probability of light interception of cotton plant as a function of sun position and diurnal leaf orientation (Dg, measured; Br, by the Beer-Lambert law; Bn, by binomial law), at the three stages of development: (a) LAI 0.12; (b) LAI 1.09; (c) LAI 2.84.

s. Thanisawanyangkura et al. /Agricultural and Forest Meteorology 86 (1997) 1-15 3.3.3. D i u r n a l c h a n g e s in l e a f d i s p e r s i o n o f p l a n t

There was no significant difference ( P > 0.05) between PB~er and PBinom, meaning that finite leaf size did not have a significant effect on leaf dispersion. In contrast, the difference between the actual light interception (i.e. PDigit estimated from digitizing) and the models (either the Beer-Lambert model or the binomial model) was significant ( P < 0.05). Dispersion of cotton leaves expressed by the ratio PDigit//PBinom varied significantly ( P < 0.01) during the day according to the stage of development (Table 4). Cotton leaves presented a regular dispersion (PDigit/PBinom > 1.0) during the day (particularly at the LAI 0.12 and 2.84 stages). Conversely, the leaf dispersion tended to be random (PDigit//PBinom = 1.0) in the morning and in the afternoon but clumped (PDigit//PBinom < 1.0) at noon at the L A I 1.09 stage. At all three stages, leaf dispersion tended to be more regular in the morning and in the afternoon than at noon (Table 4). However, comparison of the figures shows that most changes in leaf dispersion were related more to changes in sun direction than to changes in plant geometry. This pattern is clearly

Table 4 The ratio of the measured probability of light interception, PDigit, to the probability of light interception for a binomial distribution of leaves (PBinom) at three times of day and for three positions of the Sun Plant structure Sun position Morning

Noon

Afternoon

1.26 ':

1.33 "

1.30 ~' 1.20 ,t

1.22 b.d

1.24 b.c 1.20 d

1.24 b.c

1.26 c

0.97 '' 0.96 ~ 0.93 ~

0.79 b 0.81 b

0.93 a 0.96 a

0.82 b

0.96 a

1.29 ~'

1.06 c

1.26 t. 1.30 ~.b

1.11 c

1.29 b 1.30 a.b

1.10 c

1.38 a

LA10.12

Morning Noon Afternoon LAI 1.09

Morning Noon Afternoon LAI 2.84

Morning Noon Afternoon

The results are presented for three stages of development of the cotton plants when LAI was 0.12, 1.09 and 2.84. The results for the plant structure appropriate to the Sun's position are highlighted by printing in italics. The PDigit//PBinom values of each stage of development followed by the same letter are not significantly different (P > 0.051}.

11

observed at L A I 1.09 and L A I 2.84 stages, whereas differences in leaf dispersion at L A I 0.12 stage are smaller and not clearly related to changes in sun direction or plant structure.

4. Discussion 4.1. H e l i o t r o p i s m in c o t t o n

Cotton leaves exhibit varying heliotropic response during the day with difference in degree of response according to the growth stage. As heliotropism is defined as change in leaf orientation to track or avoid direct sunlight, it may be quantified by leaf angle distributions (Lang, 1973; Fukai and Loomis, 1976) or by the incidence angle fl (Ehleringer and Hammond, 1987). On the one hand, leaf azimuth and inclination distributions observed in our study are not obviously related to the sun position (Fig. 3 and Fig. 4). Only leaf azimuths at the L A I 2.84 stage clearly follow the Sun's course. Lang (1973) and Fukai and Loomis (1976) found a closer relationship between leaf angle distribution and sun direction, especially for leaf azimuths. In our study, the lack of a clear relationship may be due to the relationship between leaf inclination and leaf azimuth, as can be seen in Figs. 5 - 7 , and as reported by Lang (1973). One could, however, notice that sensitivity of cos/3 to leaf azimuth is low when foliage inclination is planophile (i.e. with most leaf inclinations between 0 ° and 30°), especially at noon when solar elevation is high (e.g. Sinoquet and Andrieu, 1993). On the other hand, Table 1 shows that values of cos/3 are greater at noon than in the morning or the afternoon. This does not mean that leaf heliotropism is greater at noon: in fact, any kind of planophile canopy would show a similar diurnal patterns of cos/3, even if leaves do not move during the day. Therefore, Table 2 only shows the combined effects of sun direction and plant geometry. It gives some evidence that there are changes in leaf angles to allow the leaves to better face the Sun during the day. Decrease in solar tracking movement (i.e. in values of cos/3; see Table 1) as the plant develops might relate to leaf age and plant maturity (Lang and Begg, 1979; Ehleringer and Forseth, 1989). First,

12

S. Thanisawanyangkura et aL / Agricultural and Forest Meteorology 86 (1997) 1-15

leaf movement is reported to be related to petiole mechanics, so that motion of older leaves is likely to decrease with petiole lignification. Second, heliotropism is reported to be driven by blue light signals (Ehleringer and Forseth, 1989; Firn, 1994) or intra-leaf irradiance gradients (Fisher and Fisher, 1983), suggesting that heliotropism is likely to decrease as mutual shading increases with plant development.

4.2. Leaf dispersion Light interception in heliotropic canopies has only been considered from a leaf orientation point of view. However, changes in leaf angles cannot occur without change in leaf blade location, which may have consequences on leaf dispersion. In our study, the digitizing technique allowed us to observe diurnal changes in leaf location. Even though coordinate measurements may be affected by operator error during digitizing in the field (about 1 cm), significant diurnal changes in leaf blade location were identified. It was relatively small at earlier stage in the plant's growth, because of relatively small leaf size and short petiole length, but it was significantly greater than the location error at the later stages (4-5 cm). Although leaf blades moved during the day, this did not greatly influence leaf dispersion (Table 4): for a given sun direction, any kind of plant geometry, as at morning, noon or afternoon, leads to similar leaf dispersion. This is especially the case at LAI 1.09 and 2.84 stages. This would mean that leaf motion during the day does not allow leaves to better exploit or avoid direct sunlight by tracking light or shade microsites. Table 4, however, shows diurnal changes in leaf dispersion: it is relatively more regular in the morning and in the afternoon, and more clumped at noon. As changes in leaf dispersion are not related to small leaf displacements, this results from the spatial leaf distribution at the plant level, i.e. plant architecture. In his pioneer work on leaf dispersion, Nilson (1971) has already pointed out that leaf dispersion may change with solar elevation. Experimental work on maize (Prtvot, 1985) and artificial plants (Andrieu and Sinoquet, 1993) also showed directional changes in leaf dispersion on immobile canopy structures. In

the case of cotton, data on diurnal changes in leaf dispersion are rare. Sassenrath-Cole (1995) used light measurements to infer patterns of leaf dispersion at two stages of plant development. In the case of mature G. hirsutum, leaf dispersion was regular at both early morning and noon. Unfortunately, leaf dispersion was quantified by a Markov model (Nilson, 1971) in the morning and a binomial model (Fukai and Loomis, 1976) at noon; this prevents a comparison of the degree of regularity at the two times of day. Most workers (e.g. Fukai and Loomis, 1976; Ehleringer and Forseth, 1989; Gutschick, 1991) wondered about the ecological significance of cotton diaheliotropism. They agreed that diaheliotropism is beneficial for photosynthesis when solar altitude is low but wondered about the cost-efficiency of diaheliotropism at noon: it does not improve photosynthesis as leaves with a high cos fl are light saturated, and it increases risks of water, temperature and light stress. It was concluded that maximizing light capture even if it is poorly used allows greater shading of competitors. Our results suggest that cotton plants do not necessarily attempt to maximize light interception throughout the day. First, even if leaf inclination distribution changes during the day, overall the foliage remains planophile. In other words, cotton plants would be unable to change leaf inclination by a large amount. Consequently, values of cos/3 at noon (i.e. at high solar elevation) must be high whatever the leaf azimuth distribution. Second, most work on cotton heliotropism (Lang, 1973; Fukai and Loomis, 1976; and some of ours) shows that leaf azimuth distribution follows the Sun's course. The correlation between leaf and sun azimuth at noon could have been misinterpreted: if plants tend to face the Sun in the morning and afternoon by adjusting leaf azimuths, they cannot make an abrupt shift from east-facing to west-facing and have to gradually move. South-facing of leaves at noon could just be an intermediate azimuth angle between those of morning and afternoon. This assumption is corroborated by the weak influence of leaf azimuth on cos/3 for planophile canopies at high sun elevation. Third, spatial distribution of leaves within the plant leads to more clumpiness (or less regularity) at noon (Table 4). This could be a way to decrease light interception at noon when leaves keep planophile

s. Thanisawanyangkura et al. /Agricultural and Forest Meteorology 86 (1997) 1-15

throughout the day. Plant architecture could then be designed in a way which compensates for the drawback of a planophile foliage. Another interpretation could be that new leaves locate themselves in light gaps found for the low sun directions. Caldwell (1987) suggested that plants grew in a way that allowed them to expose new leaves in light microsites. Table 4 also shows large differences in leaf dispersion between the stages of plant development. Leaf dispersion was regular at the LAI 0.12 and 2.84 stages but clumped at the intermediate growth stage. The regular distribution at the early stage might be related to cotton leaf arrangement on the main stem, which shows a spiral phyllotaxis with 135 ° divergence ( 3 / 8 of 360°), and there are only 5 - 6 main stem leaves at the LAI 0.12 stage. Spatial arrangement of the leaves is therefore regular throughout the day because of physical limitations of mutual shading. At the LAI 2.84 stage, the regular leaf dispersion is in agreement with previous reports on leaf dispersion in a closed canopy of G. hirsutum (Sassenrath-Cole, 1995). The clumped dispersion observed at LAI 1.09 stage could have been related to the soil surface area covered by a plant. At intermediate stages, it is to be expected that the leaf area would be concentrated above a small soil surface area (e.g. all leaves attached to the main stem). However, this assumption disagrees with observed rates of soil surface occupation by the plant, i.e. the ratios of soil surface area occupied by a plant to the plant leaf area. According to plant mapping information, the clumped leaf dispersion is mostly due to leaf clusters, as plants developed 6 - 8 branches with only 2-3 fully expanded leaves on each branch at the LAI 1.09 stage. This indicates a clear but non-explicit relationship between plant architecture and leaf dispersion, and the difficulties of inferring leaf dispersion parameters from plant geometry measurements. 4.3. Implications f o r .light interception modelling

In models based on the Beer-Lambert law, the probability of light interception may be written P = 1 - exp[( - G / s i n h)/~L] where /x describes leaf dispersion and L is the leaf

13

area index. In most models, G is computed assuming a random leaf azimuth distribution and random leaf dispersion (/z = 1). Our results suggest that such assumptions would greatly underestimate probability of light interception. First, the actual azimuth distribution increases the value of G by 5-15% (Table 1). Second, leaf dispersion is mostly regular (Fig. 10 and Table 4). In both morning and afternoon, leaf regularity expressed in terms of /.~ gives values about 1.4 and two at the LAI 0.12 and 2.84 stages, respectively. The magnitude of the resulting error in estimating the interception probability could be counterbalanced by an overestimation of LAI of about 160-220%. This illustrates the weak ability of simple models to estimate light interception in cotton plants. Some models take into account diaheliotropism. Gutschick (1991) proposed a model where all leaves are completely diaheliotropic, i.e. all leaves have a cos/3 equal to unity. This assumption overestimates the projection coefficient G, but the overestimation could compensate for the random leaf dispersion hypothesis. In a similar way, Mann et al. (1980) assumed that a fraction of leaf area (17%) showed a cos/3 equal to unity whereas the remainder was planophile. Another approach would be to assume that the only sunlit leaves are diaheliotropic. This would relate light interception and sunlit area as follows: P = 1 - e x p [ ( - 1/sin h ) L s + ( - Gp/sin h)Lp]

(1) P = ( I / s i n h)Ls

(2)

where Ls is the sunlit leaf area index, Gp is the G-function with a planophile inclination and Lp is the shaded leaf area index. Eq. (1) corresponds to the Beer-Lambert law with sunlit leaf area index L~ facing the Sun and shaded foliage Lo with a planophile inclination. Eq. (2) states that only the sunlit area intercepts direct sunlight. In such a model, the unknown L s has to be iteratively inferred as there is no explicit solution. Although this makes the computation of P more complicated, this formalism attempts to relate leaf orientation to the assumed light signals responsible for heliotropism. However, it does not take into account departure from foliage randomness.

14

S. Thanisawanyangkura et al. / Agricultural and Forest Meteorology 86 (1997) 1-15

Finally, the model of Fukai and Loomis (1976) allows heliotropism to be taken into account, as the model parameters are spatial variations in leaf area density, in leaf inclination and azimuth distributions, and in leaf dispersion. Spatial distribution of leaf area is taken into account by dividing the vegetation space into two-dimensional (2D) cells. There is no assumption about changes in canopy structure throughout the day, as the model inputs are field measurements of plant geometry. Such a model provides a general framework to model light microclimate. The model does not generate diurnal changes in leaf orientation resulting from heliotropism, but it could be fed with angular distributions proposed by Verstraete (1987) to describe heliotropic foliages. By using a binomial law for light interception, Fukal and Loomis (1976) found that leaf dispersion in cotton canopies is regular within a cell: about half of the regularity was related to leaf size (as a fraction of cell size) and the remainder was due to leaf arrangement within the 2D cells. However, it should be kept in mind that taking into account the spatial variations of leaf area density with an array of 2D cells results in considering leaf clumpiness at a canopy level. In our study, finite leaf size appeared to have no significant effect on leaf dispersion (Fig. 10) because it was expressed as a fraction of the soil surface area occupied by the plant. Moreover, results on leaf dispersion might have been different if we used the soil surface area available per plant (i.e. the inverse of plant density) instead of the projection of the plant onto the soil: leaf dispersion would appear to be clumped in the early stage (LAI 0.12) because of the very small horizontal projection of the plant. That may be the reason why leaf dispersion in the early stage of G. hirsutum was reported to be clumped by Sassenrath-Cole (1995).

5. Conclusion Although heliotropism is defined as diurnal changes in leaf orientation related to the Sun's course, light interception also depends on the spatial distribution of leaves within the plant. In this study, heliotropism of G. hirsutum in terms of incidence angle of direct sunlight on the leaves was observed, in agreement with all previous works (Lang, 1973;

Fukai and Loomis, 1976; Ehleringer and Hammond, 1987). However, diurnal patterns of leaf dispersion showed that cotton foliage is more regular in the morning and in the afternoon than at noon. location of new leaves within the plant could then be arranged to exploit light gaps found for the low sun altitudes. One should, however, keep in mind that this work was on a plant basis. To complete this work at a canopy level, interactions between plants will be considered in a further study. This should allow us to identify how leaf dispersion is influenced by intra-plant and inter-plant leaf locations. From a methodological point of view, 3D digitizing provides us with an accurate description of the plant geometry in terms of leaf location and leaf orientation. The data set allowed us to build pictures of the plants measured in the field, from which information on light interception was derived. Using these measurements with rough image synthesis and image analysis gave us an improved method for studying relationships between plant geometry and light interception.

Acknowledgements This research was supported in part by an EEC grant (STD III Project TS3*-CT94-0288, DG 12 HSMU). The authors wish to acknowledge the collaboration of the staff of INRA Bioclimatologie-PIAF, Clermont-Ferrand, and Unit6 de Recherche Syst~mes de Culture, CIRAD-CA, Montpellier.

References Andrieu, B. and Sinoquet, H., 1993. Evaluation of structure description requirements for predicting gap fraction of vegetation canopies. Agric. For. Meteorol., 65: 207-227. Caldwell, M.M., 1987. Plant architecture and resource competition. In: E.D. Schultze and H. ZwSlfer (Editors), Ecological Studies 61. Springer, Berlin, pp. 164-179. Ehleringer, J.R. and Forseth, I.N., 1989. Diurnal leaf movements and productivity in canopies. In: G. Russell, B. Marshall and P.G. Jarvis (Editors), Plant Canopies: their Growth, Form and Function. Cambridge University Press, Cambridge, pp. 129142. Ehleringer, J.R. and Hammond, S.D., 1987. Solar tracking and photosynthesis in cotton leaves. Agric. For. Meteorol., 39: 25-35.

S. Thanisawanyangkura et al. / Agricultural and Forest Meteorology 86 (1997) 1-15 Firn, R.D. 1994. Phototropism. In: R.E. Kendrick and G.H.M. Kronenberg (Editors), Photomorphogenesis in Plants. Kluwer Academic, Dordrecht, pp. 659-681. Fisher, F.J.F. and Fisher, P.M., 1983. Photosynthetic patterning: a mechanism for sun tracking. Can. J. Bot., 61: 2632-2640. Fu, Q.A. and Ehleringer, J.R., 1989. Heliotropic leaf movements in common beans controlled by air temperature. Plant Physiol., 91: 1162-1167. Fukai, S. and Loomis, R.S., 1976. Leaf display and light environments in row-planted cotton communities. Agric. Meteorol., 17: 353-379. Gutschick, V.P., 1991. Joining leaf photosynthesis models and canopy photon-transport models. In: R.B. Myneni and J. Ross (Editors), Photon-Vegetation Interactions: Applications in Optical Remote Sensing and Plant Ecology. Springer, Berlin, pp. 516-535. Isoda, A., Yoshimura, T., Ishikawa, T., Nojima, H. and Takasaki, Y., 1994. Effects of leaf movement on radiation interception in field grown leguminous crops. III. Relation to leaf temperature and transpiration among soybean cultivars. Jpn. J. Crop Sci., 63(4): 657-663. Kao, W.-Y., Comstock, J.P. and Ehleringer, J.R., 1994. Variation in leaf movements among common bean cultivars. Crop Sci., 34: 1273-1278. Lang, A.R.G., 1973. Leaf orientation of a cotton plant. Agric. Meteorol., 11: 37-51. Lang, A.R.G. and Begg, J.E., 1979. Movements of Helianthus annuus leaves and heads. J. Appl. Ecol., 16: 299-305. Mann, J.E., Curry, G.L., DeMichele, D.W. and Baker, D.N., 1980. Light penetration in a row-crop with random plant spacing. Agron. J., 72: 131-142. Moulia, B. and Sinoquet, H., 1993. Three-dimensional digitizing systems for plant canopy geometrical structure: a review. In: C. Varlet-Grancher, R. Bonhomme and H. Sinoquet (Editors), Crop Structure and Light Microclimate: Characterization and Applications. INRA Editions, Paris, pp. 183-193. Nilson, T., 1971. A theoretical analysis of the frequency of gaps in plant stands. Agric. Meteorol., 8: 25-38. Polhemus, Inc., 1993. 3 SPACE ® FASTRAK ® User's Manual, Revision F. Polhemus. Inc., Colchester, VT, 165 pp.

15

Prrvot, L., 1985. Modrlisation des 6changes radiatifs au sein des couverts vrg6taux, application h la trlrd&ection, validation sur un couvert de ma'is. Th~se d'universitr, Paris VI, 178 pp. Ross, J., 1981. The Radiation Regime and Architecture of Plant Stands. Dr. W. Junk, The Hague, 391 pp. Russell, G., Jarvis, P.G. and Monteith, J.L., 1989. Absorption of radiation by canopies and stand growth. In: G. Russell, B. Marshall and P.G. Jarvis (Editors), Plant Canopies: their Growth, Form and Function. Cambridge University Press, Cambridge, pp. 21-39. Sassenrath-Cole, G.F., 1995. Dependence of canopy light distribution on leaf and canopy structure for two cotton (Gossypium) species. Agric. For. Meteorol., 77: 55-72. Sinoquet, H. and Andrieu, B., 1993. The geometrical structure of plant canopies: characterization and direct measurement methods. In: C. Varlet-G-rancher, R. Bonhomme and H. Sinoquet (Editors), Crop Structure and Light Microclimate: Characterization and Applications. INRA Editions, Paris, pp. 131-158. Sinoquet, H. and Rivet, P., 1997. Measurement and visualisation of the architecture of an adult tree based on three-dimensional digitizing device. Trees, in press. Sinoquet, H., Moulia, B. and Bonhomme, R., 1991. Estimating the three-dimensional geometry of a maize crop as an input of radiation models: comparison between three-dimensional digitizing and plant profiles. Agric. For. Meteorol., 55: 233-249. Sinoquet, H., Varlet-Grancher, C. and Bonhomme, R., 1993. Modelling radiative transfer within homogeneous canopies: basic concepts. In: C. Varlet-Grancher, R. Bonhomme and H. Sinoquet (Editors), Crop Structure and Light Microclimate: Characterization and Applications. INRA Editions, Paris, pp. 207-228. Verstraete, M.M., 1987. Radiation transfer in plant canopies: transmission of direct solar radiation and the role of leaf orientation. J. Geophys. Res., 92: 10985-10995. Wang, P., Isoda, A., Wei, G., Yoshimura, T. and Ishikawa, T., 1994. Growth and adaptation of soybean cultivars under water stress conditions. II. Effects of leaf movement on radiation interception. Jpn. J. Crop Sci., 63(4): 699-705. Yu, F. and Berg, V.S., 1994. Control of paraheliotropism in two Phaseolus species. Plant Physiol., 106: t567-1573.