Spatial and Temporal Variation of Light inside Peach Trees

J. AMER. SOC. HORT. SCI. 119(4):669–677. 1994.

Spatial and Temporal Variation of Light inside Peach Trees Michel Génard1 and Frédéric Baret2 Institut National de la Recherche Agronomique, Centre de Recherches Agronomique, Domaine SaintPaul, Montfavet cedex 84143, France Additional index words. Prunus persica, hemispherical photographs, multivariate analysis, shoot position, daily variation Abstract. Gap fractions measured with hemispherical photographs were used to describe spatial and temporal variations of diffuse and direct light fractions transmitted to shoots within peach trees. For both cultivars studied, spatial variability of daily diffuse and direct light transmitted to shoots was very high within the tree. Diffuse and daily direct light fractions transmitted to shoots increased with shoot height within the tree and for more erect shoots. Temporal variations of hourly direct light were also large among shoots. Hourly direct light fractions transmitted to shoots were analyzed using recent developments in multivariate exploratory analysis. A gradient was observed between shoots sunlit almost all day and other shoots almost never sunlit. Well sunlit shoots were mostly located at the top of the tree and were more erect. Shoots located in the outer parts of the tree crown were slightly but significantly more sunlit than others for one cultivar. Principal component analysis additionally discriminated shoots according to the time of the day they were sunlit. This classification was related to shoot compass position for one cultivar. Spatial location of the shoot in the tree explained only a small part of light climate variability. Consequences of modeling light climate within the tree are discussed.

Development and growth of organs such as flower buds and fruit depends mostly on the assimilate supply from the nearest leaves (Hansen, 1967). Light intercepted by leaves is a key factor governing photosynthesis. Thus, light distribution within the canopy explains part of the variability of flower-bud development and fruit growth within the tree (Lakso, 1980; Seeley et al., 1980). Light also influences fruit temperature and, thus, fruit metabolic processes (Jackson, 1980). This action of temperature could explain the variation of gustatory quality with light, as observed by Barritt et al. (1987) and Marini et al. (1991). Moreover, light is known to be important in anthocyanin accumulation in fruit (Erez and Flore, 1986) and, thus, on fruit color (Doud and Ferree, 1980; Heinicke, 1966). Generally, when branches were shaded and experienced low light levels, they performed poorly (Robinson et al., 1991; Rom, 1991). Microclimatic conditions at canopy, tree, or leaf level vary during the day. Daily patterns of light and temperature explain why leaves or fruit receiving similar daily amounts of light behave differently (Moran and Rom, 1991). Several models were recently proposed to predict temporal variation of light interception for the whole canopy (Johnson and Lakso, 1991). The models may simulate the effect of row orientation, tree density spacing, and shape on light interception (Palmer, 1989) and, therefore, are very useful for investigating variations of yield components and fruit quality at the canopy level. Some models also consider spatial variation of light interception within the tree (Norman and Welles, 1983). However, such models are not accurate enough to be used to describe and understand fruit growth and quality within a tree. Thus, additional studies focusing on spatial and temporal variation of light interception within the tree are required. Our study represents the first step of a larger program dedicated Received for publication 12 July 1993. Accepted for publication 3 Feb. 1994. We thank J. Besset, H. Defrance, and G. Ducailar for technical assistance and F. Lescourret, C. Bruchou, R. Delécolle, and R. Marini for helpful comments on this paper This study was funded by Region PACA under project grant 91/05416. The cost of publishing this paper was defrayed in part by the payment of page charges. Under postal regulations, this paper therefore must be hereby marked advertisement solely to indicate this fact. 1 Station d’Agronomie. 2 Station de Bioclimatologie.

J. AMER. SOC. HORT. SCI. 119(4):669–677. 1994.

to the analysis of factors governing fruit growth and maturation. Estimates of light transmitted to the shoot are necessary to compute light intercepted by the shoots or fruit. The present work describes spatial and temporal variation of light transmitted to shoots in a peach tree. Spatial variation of diffuse and daily direct light transmitted to shoots is first presented. Then, we focus on the daily pattern and spatial variation of hourly direct light transmitted to shoots. For this purpose, we applied recent developments of multivariate exploratory analysis. Materials and Methods ‘Alexandra’ peach and ‘Fantasia’ nectarine trees (Prunus persica L. Batsch) on ‘GF 305’ rootstock were planted in 1985 in the orchard of the Institut National de la Recherche Agronomique’s Gotheron experimental station, southeastern France (44°30´N, 4°30´E). Trees were spaced 4 × 6 m with north–south oriented rows and trained in goblet shape. They received routine horticultural care. Light penetration within peach trees was evaluated with a fisheye photography technique. Two trees of each cultivar were sampled in the same experimental plot. Respectively, 98, 94, 88, and 94 fruit-bearing shoots were randomly selected on the four trees. This represented about one-fourth of the population of shoots bearing fruit. A photograph was taken above the middle of each shoot sampled, looking upward to the sky. Thus we evaluated light transmitted to the shoot. Additional tests indicated that the images and the resulting gap fractions were not very sensitive to the location of the camera above the shoot, because shoots were not very close together. Shoot spatial location within the tree canopy (Fig. 1) was characterized by its height above the ground (HEI, cm), its distance to the vertical tree axis (DIST, cm), its inclination (INC, degrees; INC < 0 for drooping shoots, INC = 0 for horizontal shoots, and INC > 0 for erect shoots), and its compass position classed in eight categorical variables (ENE, NNE, NNW, WNW, WSW, SSW, SSE, ESE) represented by their categories (0/1 variables). The goblet shape of the trees generated a high correlation between HEI and DIST (R = 0.64, P < 0.001). This trivial relationship between HEI and DIST was suppressed using the residuals (DIS1) of the 669

threshold level could vary from one operator to another, a consistent level was generally obvious, and a small variation in the threshold level induced very little change in the computed vegetation and sky fractions. Further, the same operator digitized and assigned thresholds for all the films. Once the binary image was produced (0 vegetation and 1 sky), the gap fraction Po(θ,φ) was computed by 10° zenith (θ) and 10° azimuth (φ) sectors. The measured gap fraction is an estimate of the fraction of the shoot that is in a sun fleck. The proportion of mixed pixels that were difficult to classify was minimal, except for very oblique viewing angles. Therefore we limited our analysis to the 0° to 80° zenith angle range. Linear interpolation was used to evaluate gap fraction for 80° to 90° zenith angle when the gap fraction tended toward 0° for 90° zenith angle (Baret et al., 1993). Light climate was simulated for each shoot using gap fractions derived from hemispherical photographs. Multiple scattering generated by light fluxes scattered by leaves in the tree and the soil background can be neglected in the photosynthetically active radiation domain (400 to 700 nm) because leaves and soil absorb strongly (Andrieu and Baret, 1993). Incoming radiation was supposed to be the sum of a directional component corresponding to sun beams and a diffuse component generated by scattering in the atmosphere. For simplification, the diffuse component was assumed to be isotropic. The light fraction transmitted to a given shoot was therefore computed for both the direct (Is) and the diffuse (Id) components. Hourly values of the fraction of Is transmitted to a shoot corresponded to Po(θ,φ) observed in the sun direction. The daily fraction of direct sun light transmitted to a shoot was computed, assuming that direct irradiance varied as a cosine function of sun zenith angle θ (Baret et al., 1993):

[1]

Because Id was assumed to be isotropic, it did not depend on sun position or on the time in the day and was computed as

Fig. 1. Schematic description of the shape of the tree and the variables used to characterize shoot position. The tree crown is divided into sectors according to their compass position. HEI is shoot height, DIST is the distance to the vertical tree axis, DIS1 is the distance to the internal border of the goblet, and DIS2 is the distance to the local axis of a sector.

regression of DIST on HEI (calculated independently for each tree). DIS1 corresponds to the distance between the shoot and the internal border of the goblet. The absolute value DIS2 of DIS1 characterized the position of the shoot relative to the local axis of a sector of the goblet (Fig. 1). We used an 8-mm fish eye lens (Nikon, Rochester, N.Y.) for which the projection law was very close to the theoretical polar projection. Black and white Kodak technical PAN film (Nikon Corp., Tokyo) provided highly contrasted slides. When glitter occurred due to specular reflection of leaves, confusion between the sky and vegetation was possible. To minimize this problem, we took the photographs only under diffuse irradiance conditions. A diaphragm priority program was set to improve the clearness. The slides were then digitized with a camera connected to a digitization board (512 × 512, 256 gray levels) attached to a personal computer. The images were then assigned a threshold by visual comparison to the original slide projected simultaneously aside. Although the 670

[2]

Hourly and daily direct and diffuse fractions of light transmitted to a shoot were averaged from 25 June to 5 July, when hemispherical photographs were taken. That period corresponded to the end of shoot growth, when trees reached their maximal leaf area index. Spatial variation of daily diffuse and direct light fractions was studied using simple correlations and by multiple linear regressions with shoot spatial location characterized by quantitative variables. In these regressions, we used linear components of shoot spatial location variables and their interaction to second order. The stepwise procedure of the statistical package S plus (Chambers and Hastie, 1992) was used. We tested the effect of shoot compass position using a Kruskal-Wallis one-way analysis of variance (Scherrer, 1984). Variability among shoots of hourly fractions of direct light transmitted was described with principal component analysis (PCA) using hours as variables. The data set was represented by a matrix Is, where each record corresponded to a given shoot and each variable to the observed light fraction transmitted during one J. AMER. SOC. HORT. SCI. 119(4):669–677. 1994.

of the 15 h of the day [(5:00), (6:00) ... (19:00)]. PCA synthesizes a multivariate dataset by a reduced number of uncorrelated components as explained by Broschat (1979) and Iezzoni and Pritts (1991) to horticultural research. Each successive component represents a decreasing fraction of the variance of the data set (Lebart et al., 1984). To test the significance of the PCA, we compared sphericity in the original data to the sphericity of randomized data (null hypothesis). Under the null hypothesis, the variables are not supposed to be highly correlated, leading to a low variance associated with the first principal components. An approximate randomization test was applied (Noreen, 1989). For each variable independently, the records were shuffled and the variance of the first principal components was computed. This process was repeated 1000 times. Comparison between the observed variances and the simulated distribution of variances provided the probability that the data set had a random structure. In this case, shoots would be similarly illuminated. Following Génard and Bruchou (1993), we chose to present three main outputs of the PCA: i) the percentages of variation accounted for by the principal components, ii) the correlations between the variables and PCA components, and iii) the scores of records on the components. Variables are displayed in the planes generated by the PCA components by means of points whose coordinates are correlations with components. In the plane generated by two PCA components, a vector generated by the origin and the point corresponding to a variable is defined. The cosines of the angle between two vectors associated with two variables represents the correlation between the projection of these variables in the PCA plane. Vectors with a length close to 1 are well correlated with the plane of the two components and are the most meaningful. For each cultivar, we plotted the scores of records of both trees studied on PCA components to evaluate tree-to-tree variation of the direct light transmitted. The trees were compared on factorial planes using a 95% confidence region for the mean value of PCA scores (Scheffé, 1959). To investigate spatial variation of the hourly direct light transmitted, we used the earlier data table Is and an extra data table Xp composed of the same records as for Is, the variables being the four quantitative spatial location variables (HEI, INC, DIS1, DIS2) and the categorical compass position variable. PCA with instrumental variables (PCAIV) was used as described by Sabatier et al. (1989) and Lebreton et al. (1991). This technique was also applied for horticultural research by Génard and Bruchou (1992). The aim of PCAIV is to relate the first data table Is to the second table Xp. As discussed by Lebreton et al. (1988a, 1988b, 1991), Xp can be made up of quantitative or of categorical variables represented by their categories (0/1 variables). PCAIV proceeds in two steps to explain Is by Xp. First, Is was projected on Xp using simultaneous multiple regressions of the 15 Is variables on the 12 Xp variables (four spatial location variables and the eight compass position catego-

ries). A new data set was then built with the fitted Is values (Îs). It represented the part of hourly transmitted direct light linearly explained by variables characterizing shoot position. In the second step, Îs variations were described using a PCA. Thus, PCAIV is a PCA for which the axes are constrained to be linear combination of Xp variables. We calculated the ratio R2 of the variances of Îs and Is. This ratio is the fraction of the variance of the original data Is explained by the explanatory variables Xp and is analogous to a squared correlation coefficient. We calculated also the R2 adjusted for sample size. We used a randomization approach to evaluate the significance of the relationships between Is and Xp according to Manly (1991): we compared the R2 value obtained from the standard PCAIV with the distribution of R2 values computed when shuffling the records of Is. During the randomization, Is values for a record were maintained together to preserve the correlations between Is variables. This randomization differed from the previous one because here each record (shoots) was preserved, but the link with the corresponding shoot position variables was destroyed. This procedure was repeated 1000 times. If Is is not significantly related to Xp, it will be concluded that shoot illumination does not depend on shoot spatial location variables. Results Diffuse and daily direct light fractions transmitted. Diffuse light fractions transmitted to a shoot varied from 0% to 100%, with a large standard deviation of 30% and 25%, respectively, for ‘Alexandra’ and ‘Fantasia’. The average value was similar for the two cultivars (50% and 47%). About 30% of shoots received 0.05) to the model. Further, the R2 adjusted for

Table 1. Variation of diffuse light with compass position (percentage of incoming diffuse light).

Mean SD

Sample size Mean SD

Sample size

NNE

ENE

54 34 30

59 31 30

42 29 19

51 24 36

J. AMER. SOC. HORT. SCI. 119(4):669–677. 1994.

ESE SSE Alexandra 48 50 29 30 27 29 Fantasia 40 46 24 21 21 26

SSW

WSW

WNW

NNW

49 28 33

25 15 11

53 25 16

49 26 16

41 23 19

45 26 25

62 24 14

50 27 22

671

Table 2. Diffuse light (DIF) as affected by shoot height (HEI), inclination (INC), distance to the internal border of the goblet (DIS1), and distance to the local axis of a sector (DIS2). Total and simplified regression models were presented for each cultivar. Regression model

R2

R2 adjusted

0.47

0.45

0.45

0.44

0.35

0.33

0.29

0.28

Alexandra Total model DIF = 0.214 HEI + 0.104 INC + 0.079 DIS1 + 0.036 DIS2 + 0.002 HEI × INC + 0.001 HEI × DIS1 – 0.004 DIS1 × DIS2 Simplified model DIF = 0.240 HEI + 0.310 INC Fantasia Total model DIF = 0.187 HEI + 0.346 INC + 0.265 DIS1 – 0.146 DIS2 + 0.002 HEI × DIS1 + 0.002 HEI × DIS2 – 0.011 DIS × DIS2 Simplified model DIF = 0.211 HEI + 0.357 INC

Fig. 2. Daily variation of the average hourly direct light fraction transmitted to a shoot for ‘Alexandra’ and ‘Fantasia’.

sample size for the complete models was almost equal to those of models based only on the best predictors of diffuse light that are HEI and INC (Table 2). Daily fractions of direct light transmitted to shoots were highly correlated to the diffuse light fraction (r = 0.934, P < 0.001). Therefore, similar interpretation of the spatial variation of daily direct and diffuse fractions applies. Direct light: variation during the day. The average direct light transmitted to a shoot shows a symmetric daily pattern (Fig. 2): Is increased beginning at sunrise (5:00 AM), then plateaued from about 07:00 to 17:00 HR when shoots intercepted about half the direct incoming sun light, and finally decreased until sunset. As presented in the methods, variability among shoots was described for each cultivar using a PCA on the hourly fraction of direct light transmitted to a shoot as variables. The first three principal components accounted (P < 0.001) for 52%, 14%, and 11% for ‘Alexandra’ and 40%, 21%, and 13% for ‘Fantasia’ of the variance of hourly direct light fraction. Variance of successive factors decreased sharply after the first principal component for ‘Alexandra’, indicating an almost one-dimensional pattern of light

Fig. 3. PCA scores of shoots. Shoots are displayed in the PC1–PC2 plane. Numbers 1 and 2 correspond to the two ‘Alexandra’ trees and 3 and 4 correspond to the two ‘Fantasia’ trees. The 95% confidence region of each tree is drawn.

672

J. AMER. SOC. HORT. SCI. 119(4):669–677. 1994.

climate variation among shoots. Conversely, it decreased more regularly for ‘Fantasia’, suggesting a more complex pattern. Distribution of shoots in the plane defined by the first two principal components (Fig. 3) is uniform for each of the four trees sampled. Hence, variation among shoots does not allow clear grouping of shoots. There were no significant differences between trees of the same cultivar, since confidence region of mean scores overlap on each factorial plane (Fig. 3). A correlation plot shows for both cultivars that the first component (PC1) corresponds to a gradient between shoots rarely sunlit and shoots sunlit almost all the day (Fig. 4). During most of the day, ≈25% of shoots received a fraction of the direct component >90%, while >30% of shoots is shadowed (Table 3). This first principal component was strongly linked with the daily fraction of direct light transmitted (r = 0.99, P < 0.001). The second principal component (PC2) corresponded for both cultivars to a gradient between shoots better illuminated in the morning and shoots better sunlit in the afternoon. The third principal component (PC3) discriminated shoots better sunlit in midday. Therefore, shoot illumination pattern in the PC2 × PC3 plane was described by the time course (Fig. 4). To illustrate this fact we grouped the shoots into three sets according to their PC2 and PC3 values and calculated the percentage of shoots receiving

a high fraction of the direct light component (>80%) in the morning, midday, or afternoon (Table 4). Direct light: spatial variation. As presented in the methods, spatial variation of hourly direct light fraction transmitted to a shoot was investigated using PCAIVs. The first step of the PCAIV applied on each cultivar indicated that factors characterizing shoot spatial location in the tree accounted for one-fourth of the variation of the hourly fraction of direct light transmitted (R2 = 26% for ‘Alexandra’ and R2 = 21% for ‘Fantasia’, P < 0.001), but R2 adjusted for sample size was weak (21% for ‘Alexandra’ and 15% for ‘Fantasia’). The first PCAIV principal component accounted for 88% and 58% (Table 5) of the variation explained by shoot location variables for ‘Alexandra’ and ‘Fantasia’ respectively (P < 0.001). As hourly direct light variables were all positively correlated to the first PCAIV component (Table 5), it is interpreted similarly to the first PCA component described above: a gradient between shoots rarely sunlit and shoots sunlit almost all the day. As for the daily light fraction, this first component was positively related to shoot height and shoot inclination (Table 5). Although few descriptors of compass position were well correlated with the first PCAIV component, nonsignificant correlation ratio values (Jambu, 1989) indicated no clear effect of compass position on the

Fig. 4. Correlation plots of the PCAs. Variables are displayed by means of vectors. The closer the vectors are together, the stronger their correlation is. Vectors with length close to 1.0 are the most important for interpreting the components. Numbers 5 to 19 represent the hours of the day (5:00 AM to 7:00 PM).

J. AMER. SOC. HORT. SCI. 119(4):669–677. 1994.

673

Table 3. Percentage of ‘Alexandra’ and ‘Fantasia’ shoots receiving a fraction of direct light (DIR) component 90% for at least 7 h per day. DIR ≤ 10% 36 32

Alexandra Fantasia

10% < DIR ≤ 90% 0 0

DIR > 90% 31 21

Table 4. Percentage of ‘Alexandra’ and ‘Fantasia’ shoots receiving a fraction of direct light component >80% in the morning, midday, or afternoon, according to their values on PC2 and PC3 components. Shoots (%)

PC2 < 0 and PC3 > 0 PC3 < 0 PC2 > 0 and PC3 > 0 PC2 < 0 and PC3 >0 PC3 < 0 PC2 > 0 and PC3 > 0

7–8 h Alexandra 68 0 13 Fantasia 64 0 34

11–13 h

16–17 h

54 50 22

43 0 39

21 46 32

13 0 66

whole (Table 5). DIS2 was correlated to the first component of PCAIV for ‘Fantasia’ only. This means that outer parts (high DIS2) of the tree crown were more sunlit than internal parts for ‘Fantasia’. Comparison between PCAs and PCAIVs first principal component patterns according to HEI and DIST (Fig. 5) showed that the model issued from the first component of PCAIVs approximated the reality as expressed by the first component of PCAs. Therefore, the main pattern of light distribution within the tree is correctly depicted by HEI for both cultivars and DIS2 only for ‘Fantasia’. Nevertheless, local variations are large and only partly accounted for by INC. Figure 5 shows that light distribution patterns are different for the two cultivars. For ‘Fantasia’, the second principal component accounted for 30% of the variability and was similar to that of the former PCA for ‘Fantasia’. It was interpreted as a gradient between shoots located in the eastern part of the tree that were sunlit in the morning and those located in the western part that were sunlit in the afternoon (Table 5). Shoots located at the borders of the crown seemed more sunlit in the morning. The second principal component of the PCAIV for ‘Alexandra’ accounted only for 4% of the variance. The third principal component of the PCAIV accounted for 3% and 7% of the variance respectively for ‘Alexandra’ and ‘Fantasia’ and these components were not interpreted. PCAIV did not exhibit a clear gradient between shoots better sunlit in midday and other shoots,

Table 5. Correlation between the two first PCAIV principal components (PC1 and PC2) for ‘Alexandra’ and ‘Fantasia’ and variables characterizing, respectively, time (hours per day) and shoot spatial location. PC1 Explained variance Time variable (h) 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Spatial location variable HEI (height) INC inclination) DIS1 (distance to internal border of goblet) DIS2 (distance to local axis of sector) NNE ENE ESE SSE SSW WSW WNW NNW Compass positiony zVariance

PC2

Alexandra 88%

Fantasia 58%

Alexandra 4%z

Fantasia 30%

0.95 0.93 0.92 0.90 0.94 0.93 0.96 0.93 0.95 0.95 0.98 0.95 0.94 0.89 0.89

0.64 0.69 0.82 0.82 0.87 0.84 0.80 0.85 0.87 0.85 0.78 0.50 0.64 0.60 0.64

-------------------------------

–0.04 –0.12 –0.39 –0.51 –0.44 –0.47 –0.54 –0.33 0.06 0.37 0.49 0.84 0.71 0.76 0.69

0.89 0.33 0.07 –0.05 0.14 0.14 –0.01 0.02 –0.05 –0.28 –0.08 0.00 0.12

0.61 0.42 0.07 0.23 –0.05 0.14 –0.19 –0.05 –0.17 –0.08 0.35 0.07 0.21

---------------------------

0.02 –0.02 –0.20 –0.43 –0.08 –0.37 –0.43 –0.22 0.04 0.73 0.16 0.25 0.87

explained by PC2 for Alexandra was too low to be properly interpreted. ratio.

yCorrelation

674

J. AMER. SOC. HORT. SCI. 119(4):669–677. 1994.

as it was found through the third component of the PCA. Thus, variation expressed by that component was not linked to shoot spatial location variables. To illustrate these results, Table 6 presents the direct light transmitted to the shoot as a function of the period of the day and the shoot spatial location variables. The relatively small percentage of variability explained by variables characterizing shoot spatial location cannot be explained by variability among trees. Indeed, when the PCAIVs were performed for each tree, R2 adjusted for sample size was as small as for trees pooled by cultivar (16% and 25% for ‘Alexandra’ and 15% and 25% for ‘Fantasia’). The large remaining percentage of unexplained variance was related to the high local variability of light climate within a tree. Discussion Variation among shoots of diffuse and daily direct light fractions transmitted to a shoot was large for both cultivars. Many shoots received 2.0 m

Mean SD

INC < 0°

Mean SD

45° 45°

Mean SD

DIS1 < 0.5 m

Mean SD

0.5 < DIS1 < l.0 m

Mean SD

DIS1 > 1.0 m

Mean SD

DIS2 < 0.3 m

Mean SD

0.3 < DIS2 < 0.6 m

Mean SD

DIS2 > 0.6 m

Mean SD

NNE

Mean SD

ENE

Mean SD

ESE

Mean SD

SSE

Mean SD

SSW

Mean SD

WSW

Mean SD

WNW

Mean SD

NNW

Mean SD

zHEI

676

32 35 45 38 69 33 21 23 46 38 55 38 44 40 52 38 47 35 51 38 43 39 51 35 54 44 48 36 51 37 47 33 48 39 23 33 56 40 43 40

11–13 Alexandra 32 35 n = 56 58 40 n = 81 78 33 n = 55 39 34 n=8 55 41 n = 118 60 42 n = 66 53 42 n = 66 59 41 n = 89 54 40 n = 37 60 40 n = 104 49 41 n = 73 62 40 n = 15 64 40 n = 30 66 43 n = 30 55 43 n = 27 68 40 n = 29 48 35 n = 33 27 29 n = 11 42 42 n = 16 54 45 n = 16

16–17

7–8

22 29

42 39

49 40

53 40

79 32

74 33

65 36

34 39

47 41

50 40

54 40

59 40

52 43

55 40

48 41

46 41

51 38

66 36

50 41

47 40

51 41

62 40

43 43

78 25

55 41

63 40

55 42

66 36

48 44

56 42

47 41

51 40

53 40

36 35

26 26

35 41

39 45

60 40

62 33

50 42

11–13 Fantasia 45 39 n = 69 53 38 n = 81 61 39 n = 32 29 33 n = 14 46 39 n = 92 62 37 n = 76 52 39 n = 62 49 38 n = 89 57 39 n = 31 49 38 n = 115 56 41 n = 62 36 19 n=5 45 40 n = 19 59 39 n = 36 49 36 n = 21 51 32 n = 26 49 41 n = 19 35 27 n = 25 72 34 n = 14 53 44 n = 22

16–17 38 37 43 40 64 43 37 39 42 40 50 40 44 39 50 41 32 37 47 41 44 39 6 6 34 41 37 37 19 29 36 39 35 35 73 40 72 32 60 36

= height, INC = inclination, DIS1 = distance to internal border of goblet, DIS2 = distance to local axis of sector.

J. AMER. SOC. HORT. SCI. 119(4):669–677. 1994.

On the average, direct light transmitted to a shoot varies with time due to changes in sun position, as noted by Charles-Edwards and Thorpe (1976) and Moran and Rom (1991). PCA showed that various time course patterns were observed among shoots. However regardless of cultivar or tree, there was a clear gradient between shoots always sunlit (high daily fraction of direct light) and shoots almost never sunlit. For peach trees, DeJong and Doyle (1985) showed that hourly variation of the light fraction received by shoots was linked to shoot compass position. In our study, shoot compass position explained a significant part of the variability only for ‘Fantasia’. For this cultivar, the borders were also more sunlit. Differences between the two cultivars could be due to peculiarities of tree architecture and foliage distribution. Variables characterizing shoot spatial location in the tree crown explained a small part of light climate variability. High local variations of sunlight within the tree observed in this study show that general assumptions used for modeling light interception (Johnson and Lakso, 1991) are not sufficient to describe the light climate at shoot level accurately. Improvements in the study of light climate within trees will require explanatory variables varying locally within the tree crown and linked with light interception, such as foliage density distribution. Models of tree architecture connected with Monte Carlo techniques, as used by Oikawa and Saeki (1977), are very attractive for simulating the light climate within the tree. However this approach is very expensive in computer and research time. The technique used in this study to characterize light climate at the shoot level may be of great interest when relating local irradiance conditions to growth and development of leaves, buds, flowers, and fruit through physiological processes such as photosynthesis. Literature Cited Andrieu, B. and F. Baret. 1993. Indirect methods of estimating crop structure from optical measurements, p. 285–322. In: C. Varlet-Grancher, R. Bonhomme, and H. Sinoquet (eds.). Crop structure and microclimate: Characterizations and applications. INRA, Paris. Baret, F., B. Andrieu, and M.D. Steven. 1993. Gap frequency and canopy architecture of sugar-beet and wheat crops. Agr. For. Meteorol. 65:207–227. Barritt, B.H., C.R. Rom, K.R. Guelich, S.R. Drake, and M.A. Dilley. 1987. Canopy position and light effect on spur, leaf, and fruit characteristics of ‘Delicious’ apple. HortScience 22:402–405. Broschat, T.K. 1979. Principal component analysis in horticultural research. HortScience 14:114–117. Chambers, J.M. and T.J. Hastie. 1992. Statistical models in S. 1st ed. Wadsworth and Brooks, Pacific Grove, Calif. Charles-Edwards, D.A. and M.R. Thorpe. 1976. Interception of diffuse and direct-beam radiation by a hedgerow apple orchard. Ann. Bot. 40:603–613. DeJong, T.M. and J.F. Doyle. 1985. The effect of row orientation on light distribution in hedgerow peach tree canopies. Acta Hort. 173:159–166. De Salvador. F.R. and T.M. DeJong. 1989. Observations of sunlight interception and penetration into the canopies of peach trees in different planting densities and pruning configurations. Acta Hort. 254:341–346. Doud, D.S. and D.C. Ferree. 1980. Influence of altered light level on growth and fruiting of mature ‘Delicious’ apple trees. J. Amer. Soc. Hort. Sci. 105:325–328. Elfving, D.C., I. Schechter, R.A. Cline, and W.F. Pierce. 1990. Palmetteleader and central-leader tree forms compared for light distribution, productivity, and fruit quality of McIntosh apple trees. HortScience 25:1386–1388. Erez, A. and J.A. Flore. 1986. The quantitative effect of solar radiation on ‘Redhaven’ peach fruit skin color. HortScience 21:1424–1426. Génard, M. and C. Bruchou. 1992. Multivariate analysis of within-tree factors accounting for the variation of peach fruit quality. Scientia Hort. 52:37–51. Génard, M. and C. Bruchou. 1993. A functional and exploratory approach to studying growth: The example of the peach fruit. J. Amer. Soc. Hort. Sci. 118: 317–323.

J. AMER. SOC. HORT. SCI. 119(4):669–677. 1994.

Hansen, P. 1967. C14 studies on apple trees. The effect of the fruit on the translocation and distribution of photosynthates. Physiol. Plant. 20:382–391. Heinicke, D.R. 1966. Characteristics of McIntosh and Red Delicious apples as influenced by exposure to sunlight during the growing season. Proc. Amer. Soc. Hort. Sci. 89:10–13. Iezzoni, A.F. and M.P. Pritts. 1991. Applications of principal component analysis to horticultural research. HortScience 26:334–338. Jackson, J.E. 1980. Light interception and utilization by orchard systems. Hort. Rev. 2:208–267. Jambu, M. 1989. Exploration informatique et statistique des données. Dunod, Paris. Johnson, R.S. and A.N. Lakso. 1991. Approaches to modeling light interception in orchards. HortScience 26:1002–1004. Kappel, F., J.A. Flore, and R.E.C. Layne. 1983. Characterization of light microclimate in four peach hedgerow canopies. J. Amer. Soc. Hort. Sci. 108:102–105. Lakso, A.N. 1980. Correlations of fisheye photography to canopy structure, light climate, and biological responses of light in apple trees. J. Amer. Soc. Hort. Sci. 105:43–46. Lebart, L., A. Morineau, and K.M. Warwick. 1984. Multivariate descriptive statistical analysis. Correspondence analysis and related techniques for large matrices. Wiley, New York. Lebreton, J.D., D. Chessel, R. Prodon, and N. Yoccoz. 1988a. L’analyse des relations especes-milieu par l’analyse canonique des correspondances. I. Variables de milieu quantitatives. Acta Oecol. Oecol. Gen. 9:53–67. Lebreton, J.D., D. Chessel, M. Richardot-Coulet, and N. Yoccoz. 1988b. L’analyse des relations especes-milieu par l’analyse canonique des correspondances. II. Variables de milieu qualitatives. Acta Oecol. Oecol. Gen. 9:137–151. Lebreton, J.D., R. Sabatier. G. Banco, and A.M. Bacou. 1991. Principal component and correspondence analyses with respect to instrumental variables: An overview of their role in studies of structure- activity and speciesenvironment relationships, p. 85–114. In: J. Devillers and W. Karcher (eds.). Applied multivariate analysis in SAR and environmental studies. ECSC, EEC, EAEC, Brussels, The Netherlands. Manly, B.F.J. 1991. Randomization and Monte Carlo methods in biology. 1st ed. Chapman and Hall, London. Marini, R.P., D. Sowers, and M.C. Marini. 1991. Peach fruit quality is affected by shade during final swell of fruit growth. J. Amer. Soc. Hort. Sci. 116:383– 389. Moran, R. and C. Rom. 1991. Canopy side and height affect apple spur and fruit quality. Compact Fruit Tree 24:80–82. Noreen. E.W. 1989. Computer-intensive methods for testing hypotheses: An introduction. Wiley, New York. Norman, J.M. and J.M. Welles. 1983. Radiative transfer in an array of canopies. Agron. J. 75:481–488. Oikawa, T. and T. Saeki. 1977. Light regime in relation to plant population geometry. I. A Monte Carlo simulation of light microclimates within a random distribution foliage. Bot. Mag. Tokyo 90:1–10. Palmer, J.W. 1989. The effect of row orientation, tree height, time of year and latitude on light interception and distribution in model apple hedgerow canopies. J. Hort. Sci. 64:137–145. Robinson, T.L., A.N. Lakso, and Z. Ren. 1991. Modifying apple tree canopies for improved production efficiency. HortScience 26:1005–1012. Rom, C.R. 1991. Light thresholds for apple tree canopy growth and development. HortScience 26:989–992. Sabatier, R., J.D. Lebreton, and D. Chessel. 1989. Principal component analysis with instrumental variables as a tool for modelling composition data, p. 341–352. In: R. Coppi and S. Bolasco (eds.). Multiway data analysis. Elsevier Science Publishers, Amsterdam. Scheffé, H. 1959. The analysis of variance. Wiley, New York. Scherrer, B. 1984. Biostatistique. Gaëtan Morin, Chicoutimi, Canada. Seeley, E.J., W.C. Micke, and R. Kammereck. 1980. ‘Delicious’ apple fruit size and quality as influenced by radiant flux density in the immediate growing environment. J. Amer. Soc. Hort. Sci. 105:645–657. Tustin, D.S., P.M. Hirst, and I.J. Warrington. 1988. Influence of orientation and position of fruiting laterals on canopy light penetration, yield, and fruit quality of ‘Granny Smith’ apple. J. Amer. Soc. Hort. Sci. 113:693–699.

677