Designing a “Target-Tree” for Maximizing Gross Value of Product in Patagonian Sweet Cherry...

Designing a “Target-Tree”for Maximizing Gross Value

of Product in PatagonianSweet Cherry Orchards

Eduardo Daniel CittadiniHerman van Keulen

Pablo Luis PeriNico de Ridder

ABSTRACT. A “target-tree” approach to maximize gross value of prod-uct (GVP; US$/ha) at the farm gate was developed and applied tosweet cherry orchards, integrating eco-physiological information, model

Eduardo Daniel Cittadini is Head of the Group Fruit Production at Estación Experi-mental Agropecuaria Chubut, Instituto Nacional de Tecnología Agropecuaria, Casillade Correo 88, 9100 Trelew, Argentina.

Herman van Keulen is Senior Researcher, Plant Research International, WageningenUniversity and Research Centre; and Professor and Chair of the Group Plant ProductionSystems, Wageningen University, Haarweg 333, 6709 RZ Wageningen, The Netherlands.

Pablo Luis Peri is Head of the Group Forestry, Estación Experimental AgropecuariaSanta Cruz, Instituto Nacional de Tecnología Agropecuaria, Universidad Nacional de laPatagonia Austral; and Researcher, Comisión Nacional de Investigaciones Científicasy Técnicas (CONICET), Chacra 45 “A,” 9400 Río Gallegos, Argentina.

Nico de Ridder is Assistant Professor of the Group Plant Production Systems,Wageningen University, Haarweg 333, 6709 RZ Wageningen, The Netherlands.

Address correspondence to: Eduardo Daniel Cittadini at the above address (E-mail:[email protected]).

The authors would like to thank Ana Belén Pugh for her information on marketingof cherries and to Gerard Laborde for her data on cherry price seasonality. The authorsalso thank Carlos Paz, Alejandro Catalano and Francisco Eylenstein, owners of theorchards in which the experiments were carried out. Finally, the authors wish to thankEnrique Sanchez and Juan Escobar for their critics on the manuscript.

International Journal of Fruit Science, Vol. 6(3) 2006Available online at http://ijfs.haworthpress.com

© 2006 by The Haworth Press, Inc. All rights reserved.doi:10.1300/J492v06n03_02 3

mailto:[email protected]

http://ijfs.haworthpress.com

estimates, and expert knowledge. During the 2003/04 growing season,the effect of the ratio of Fruit Number to Leaf Area (FNLAR; fruit/m2

LA) on Mean Fruit Weight (MFW; g/fruit) was analyzed at both the spurand whole-tree level, for different combinations “training system-cultivar”(Experiment 1). During the 2004/05 season, the effect of FNLAR onMFW, soluble solids content (SSC), titratable acidity (TA), and firmness(F), was evaluated at whole-tree level in a ‘Bing’/‘Mahaleb’ orchard (Ex-periment 2). In Experiment 1, there were no significant interactions be-tween training system and cultivar for the effect of FNLAR on MFW atspur and whole-tree level. No significant differences were observed be-tween vase and Tatura-trellis training systems. The R2-values for the re-lationships per cultivar were higher at whole-tree level than at spur level.At both the spur and whole-tree level, ‘Lapins’ had the highest Y-inter-cept and ‘Van’ the lowest. At spur level, no differences among cultivarswere detected in their sensitivity to increments in FNLAR, but at whole-tree level ‘Van’ showed less sensitivity than ‘Lapins’ and ‘Bing’. InExperiment 2, MFW, TA and SSC decreased linearly with increasingFNLAR (P < 0.05). Firmness showed the same tendency, but the rela-tionship was not significant (P = 0.082). Minimum fruit quality thresh-olds define the suitable market for the fruit (export, domestic, or industry),with their associated price ranges. In addition, in both the domestic andexport markets, price depends mainly on fruit size. FNLAR determinesfruit quality (and indirectly fruit price), and yield, in combination withMFW and leaf area index (LAI). GVP is calculated as the product ofyield and fruit price. The combination ‘Bing’/‘Mahaleb’ on vase wasused to illustrate parameter estimation (LAI and FNLAR) for a “tar-get-tree” in Patagonian orchards, using the results of Experiment 2.Under these conditions, a LAI of 3.07 is required to intercept 75% ofPAR at harvest. With these parameters, and considering “price-fruitquality” relationships based on expert knowledge, maximum GVP wasobtained with 80 fruit/m2 LA and a yield of 18.25 t�ha�1. Although thisexample is limited to a single combination of cultivar and trainingsystem in a specific location, the methodology can be applied to othersituations, provided reliable relevant eco-physiological information is avail-able. doi:10.1300/J492v06n03_02 [Article copies available for a fee fromThe Haworth Document Delivery Service: 1-800-HAWORTH. E-mail address:<[email protected]> Website: <http://www.HaworthPress.com>© 2006 by The Haworth Press, Inc. All rights reserved.]

KEYWORDS. Ideotype, LAI, fruit number to leaf area ratio, fruit size,fruit quality, fruit price, Prunus avium

4 INTERNATIONAL JOURNAL OF FRUIT SCIENCE

mailto:[email protected]

http://www.HaworthPress.com

INTRODUCTION

In the valleys of South Patagonia, Argentina (between lat. 42�S and46�S), the total area planted with sweet cherries (Prunus avium L.) hasincreased from 176 ha in 1997 to 507 ha in 2004. In the 2004/05 season,seven packing houses processed 870 t, of which 45% was exported toEurope, 45% was sold as fresh fruit in the domestic market, and 10%was sold to industry. Fruit production will continue to increase, as mostorchards have not yet reached their mature stage and new ones arebeing established, climatic conditions are favorable, land and water areamply available, and most importantly, harvest and marketing are inthe winter season of the Northern Hemisphere. However, economicperformance has not always been satisfactory due to low yields and/orlack of fruit quality. In commercial orchards, fruit yield and price definethe gross value of product (GVP; US$/ha) for growers:

GVP = Yield × Pf (1)

where yield is expressed in kg of cherry fruit per hectare and Pf is theprice at farm gate (US$/kg). Yield can be calculated as:

Yield = MFW × FNLAR × LAI (2)

where, MFW is mean fruit weight (kg/fruit); FNLAR fruit number toleaf area ratio (fruit/ha LA) and LAI (leaf area index) [ha�ha�1].

Empirical evidence indicates that low yields in Patagonian orchardscan be associated with low LAI or with low fruit number, resulting in lowFNLAR. LAI is the main determinant of intercepted PAR (photo-synthetically active radiation), the main driver of photosynthesis (Lappiand Stenberg, 1998) and total dry matter production of crops (Patrick,1988; Wagenmakers, 1994).

As Pf depends on fruit quality, highest GVP usually is not obtained atthe highest yield due to the negative relation between yield and fruit qual-ity. Fruit weight can be manipulated through cultivar selection, trainingsystem and management practices, such as irrigation regime, nutrientsupply and so on. For a specific cultivar, FNLAR of the current year isthe most important factor affecting inter-annual variation in fruit weight(Flore and Layne, 1999; Proebsting, 1990), while management in preced-ing years has a minor influence (Whiting and Lang, 2004). Leaf area (LA)per fruit itself is not important, but represents the production potentialthrough its photosynthetic capacity, and a high value is essential for

Cittadini et al. 5

production of superior sweet cherries (Roper and Loescher, 1987). How-ever, higher availability of sugars may result in higher fruit set (Marcelisand Heuvelink, 1999), with the associated increase in FNLAR, negativelyaffecting MFW. Training system affects light distribution (Cavallo et al.,2001) and therefore canopy photosynthetic rate (Proietti et al., 2000).For the same LA per tree, a better light distribution may contribute toincreased carbohydrate supply to sink organs, allowing higher MFW.

Although individual fruit weight is the main quality characteristicdetermining fruit price, also color, firmness (F), soluble solids content(SSC) and titratable acidity (TA) are important quality characteristicsfor cherries. The relationships between FNLAR on one hand and MFW(Flore and Layne, 1999; Proebsting, 1990; Whiting and Lang, 2004),SSC (Roper and Loescher, 1987) and F (Facteau et al., 1983), on theother, are negative.

Using relationships between the different quality parameters andthe fruit to leaf area ratio in trees, “target-trees,” that is, trees withmatching LAI and FNLAR, maximizing GVP at farm gate, for different“price-fruit quality” relations can be defined. Such trees should inter-cept enough light to support high yields, but without excessive shadingthat could compromise fruit quality and future production. Managementpractices, that is, that promote tree vigor (e.g., winter pruning, irriga-tion, N-fertilization), weaken the tree (e.g., summer pruning) or affectfruit set (e.g., hormonal treatments, boron application, beehive density,and frost protection) can then be tuned to attain orchards with target-trees.

Specification of a “target-tree” can be considered conceptually equiv-alent to specifying the requirements of a crop ideotype in plant breeding(Donald, 1968), that has been described as “the use of physiological ap-proaches in crop breeding” (Sedgley, 1991), and aimed at tailoring ofplants for increased production. In particular, the target-tree resemblesthe “market ideotype” defined by Donald (1968), as “the ideotype thatis characterized by desirable characteristics at the end-point, that is, thequality characteristics of the desired product, such as bread-makingquality of the grain.” To our knowledge, this approach has not been ap-plied to perennial crops; hence the theory needs further elaboration forthis purpose.

The objective of this study was to develop an integrated model(“target-tree”) to maximize gross value of product at farm gate, basedon experimental eco-physiological information (relationship betweenfruit quality and FNLAR), expert knowledge (“price-fruit quality”relationships) and estimated relationships between light interceptionand LAI.


MATERIALS AND METHODS

To study and quantify the relationship between FNLAR and fruitquality, two experiments were performed in a sweet cherry orchard ofSouth Patagonia, Argentina.

Experiment 1

The first experiment was conducted at both the spur and whole-treelevel (Cittadini et al., 2005b). The study was carried out during the2003/04 growing season in two commercial orchards, one trained asTatura-trellis (1872 tree/ha) and the other as a vase (966 tree/ha). Bothorchards were planted in 1997 in the lower valley of the Chubut river(lat. 43�16�S; long. 65�25�W), with the cultivars ‘Bing’, ‘Van’, and‘Lapins’, grafted on the rootstock ‘Mahaleb’.

Routine horticultural care for commercial fruit production was pro-vided, including irrigation, fertilization, wind-, weed-, pest- and diseasecontrol, and winter pruning. These growing conditions were consideredoptimal and were therefore not a factor in the experiment. Ten treesfrom each combination “training system-cultivar” were randomly selectedfor measurements.

From each tree, all leaves and fruits of six 2-year-old spurs were har-vested. Each spur was harvested when more than 80% of the fruits onthat spur attained commercial color (no. 3 on the color chart from theCentre Technique Interprofessionnel des Fruits et Legumes, CTIFL).Each spur leaf was measured with a Hewlett Packard Scan Jet 4C to thenearest 0.1 cm2, using the “Image Tool 3.0” (UTHSCSA, 2002). Allfruits per spur were weighed and counted to calculate MFW andFNLAR.

All leaves on each experimental tree were counted at fruit harvestover the maturation period (7 days for each cultivar determined basedon the fruit color). Random samples of 60 leaves per tree were collectedand the mean area per leaf was determined as described at spur level.LA per tree was calculated from the number of leaves and its mean areaand used to calculate LAI for each tree based on the tree density. Totalyield per tree was measured and MFW determined from a random sam-ple of 50 fruits per tree. Number of fruits per tree and FNLAR were cal-culated from yield per tree and MFW. The sampled fruits were also usedto analyze the relationship between fruit size (mm of equatorial diame-ter) and MFW for the three cultivars.

Cittadini et al. 7

Interception of photosynthetically active radiation (PAR) was calcu-lated as:

Ia = (1 � �c) × (1 � exp (�K × CLF × LAI)) × 100 (3)

where Ia is intercepted PAR (%); �c a reflection coefficient set to 8%(Goudriaan and van Laar, 1994); K the light extinction coefficient, setto 0.6 (Jackson, 1980) and CLF a clustering factor (between 0 and 1),that was calculated with a sub-model (J. Goudriaan, Group Plant Pro-duction Systems, Wageningen University, personal communication), tocorrect for the row structure of the crop. Inputs for this sub-model areLAI, height of the trees (H), and width of the crown (W) and the path (P).

At both spur and whole-tree level, for each combination “trainingsystem-cultivar,” simple linear regression analyses were performedwith GenStat 6.1 (Payne, 2002), using FNLAR as independent variableand MFW as dependent variable. Equations were compared, in order todetect differences (P < 0.05) in slopes and Y-intercepts. When eithertraining system or cultivar had no significant effect on the relationship,that factor was removed and a new analysis was performed.

Experiment 2

This experiment was performed during the 2004/05 season to study therelationship between fruit quality and FNLAR on trees of near-optimalLAI (Cittadini et al., 2005a). As during the period of reproductivegrowth, both, growth of individual fruits and vegetative growth canbe stimulated at low FNLAR values (Grossman and DeJong, 1994;Marcelis and Heuvelink, 1999), the effect of FNLAR on mean shootgrowth (MSG) and trunk cross sectional area increment (TCSAI) wasalso analyzed.

The study was conducted in the same commercial orchard of Experi-ment 1, with ‘Bing’ trees grafted on ‘Mahaleb’ rootstock (966 tree/ha)and trained as a vase. Based on data from the same orchard that showeda positive relationship between LAI at harvest and TCSA (cm2) atbud-break (data not shown), a lower threshold for TCSA = 78 cm2 wasdefined. Therefore, 18 trees with TCSA $ 78 cm2 at bud-break were se-lected. Routine horticultural care, similar to that in Experiment 1 wasprovided, and therefore growing conditions can be considered optimal.

Fruits were harvested with color no. 4 on the CTIFL color chart, be-cause no. 3, as used in Experiment 1, was considered sub-optimal withregard to SSC and TA, and in this experiment, firmness was high


enough to allow the 3-day delay. Weight may change over that periodand therefore, the equations of the two experiments to estimate MFWhave to be compared with caution. All leaves on each of the experimen-tal trees were counted, and 1% was sampled at random. MFW was esti-mated from a random sample of 100 fruits per tree. Mean leaf area, LAper tree, LAI, PAR interception, total yield per tree, and FNLAR werecalculated as described for Experiment 1 at whole-tree level.

The fruit sample was also used to estimate F (Durofel index: 0 to100), TA (ml NaOH), and SSC (%). F was measured non-destructivelywith a durometer (Durofel 25®, Copa-technology S.S., France). Durofelis an instrument that applies a very weak pressure to the surface ofthe fruit and measure the resistance to deformation to the surface, andthe readings are relative. TA was measured by adding 10 ml of samplejuice to 90 ml of distilled water and titrating with 0.1 N sodium hydrox-ide (NaOH) to the final point of pH 8.2, SSC was measured on fruitjuice with a refractometer (Atago).

Simple linear regression analyses were performed with GenStat 6.1(Payne, 2002), using FNLAR (fruit/m2 LA) as independent variable andMFW (g/fruit), TA (ml NaOH), SSC (%), SSC/TA, F (Durofel index),MSG (cm), and TCSAI (cm2) as dependent variables, to detect signifi-cant relationships (P < 0.05).

RESULTS AND DISCUSSION

Experiment 1

MFW as a function of FNLAR did not show interactions (P > 0.05)between training system and cultivar, neither at spur nor whole-treelevel. Significant differences (P > 0.05) were neither found betweenvase and tatura-trellis training system at either level (data not shown).Therefore, training system was not considered in the further analyses.

In contrast, at spur level, ‘Lapins’ showed the highest and ‘Van’ thelowest Y-intercept value, but the rate of decrease in MFW with increas-ing FNLAR was the same for the three cultivars (Table 1).

At whole-tree level, mean LAI of individual trees was 2.5 (standarddeviation: 0.9), with an estimated PAR interception (Equation 3) of 69%.Multiple regression analysis, using LAI of individual trees and FNLARas explanatory variables, showed no significant effect of LAI on mean fruitweight (P = 0.863, data not shown). For all three cultivars, MFW decreasedwith increasing FNLAR (Figure 1). Coefficients of determination for the

Cittadini et al. 9

relationship between MFW and FNLAR were higher at this level thanat spur level (Table 1). ‘Lapins’ had the highest Y-intercept and ‘Van’the lowest, as at spur level. In contrast to the results at spur level, theslope of the MFW-FNLAR relation was smaller for ‘Van’ than for


TABLE 1. Statistical analysis of the models describing mean fruit weight(MFW) as a function of the ratio of number of fruits per leaf area (FNLAR) atspur and whole-tree level, for the cultivars ‘Lapins’, ‘Bing’, and ‘Van’.

Cultivar Y-intercept(MFW, g/fruit)

Slope (MFW/FNLAR,(g/fruit)/(fruit/m2 LA))

R2

Spur level

Lapins 10.12a (0.21) �0.0060a (0.0008) 0.25

Bing 8.30b (0.24) �0.0041a (0.0010) 0.27

Van 6.72c (0.21) �0.0028a (0.0006) 0.26

Whole-tree level

Lapins 10.43a (0.23) �0.0192a (0.0020) 0.77

Bing 9.20b (0.36) �0.0213a (0.0042) 0.89

Van 7.72c (0.35) �0.0108b (0.0022) 0.66

Note: Different letters within a single column indicate significant differences (P < 0.05) between cultivars.Values between parentheses are standard errors of estimates.

00

2

4

6

8

10

12

14

50 100 150 200

FNLAR (fruit/m2 LA)

MF

W (

g/fr

uit)

250 300 350 400

FIGURE 1. Relationship between MFW and FNLAR for sweet cherry cultivars‘Lapins’ (�), ‘Van’ (�) and ‘Bing’ (�) at whole-tree level.

‘Lapins’ and ‘Bing’. The relationship between fruit size (equatorial dia-meter in mm) and MFW did not vary among cultivars:

Size = 15.94 � 1.18 MFW (R2 = 0.96) (4)

As under similar growing conditions, cultivar was the main factoraffecting the relationship between MFW and FNLAR, cultivar differenti-ation is needed in predicting fruit weight (Proebsting, 1990). In contrast,the training system did not affect this relationship. However, the resultsof this study refer to trees of relatively low LAI (average = 2.5) and in-formation at higher LAI values is needed to validate the functions foundhere. The lack of differences between training systems at relatively lowLAI is explainable, since in that situation training system has little effecton light distribution. Supplementary observations and analyses in thecurrent experiment also showed no effects of training system or cultivaron other quality parameters (Cittadini et al., 2004a,b). However, firmnesswas significantly correlated to MFW (Firmness = 79.5 � 1.43 � MFW;P = 0.005) (Cittadini et al., 2004a), titratable acidity was correlatedto both MFW and FNLAR (Titratable acidity = 16.64 � 0.00993 3

FNLAR � 0.644 � MFW; P = 0.004) and no significant relationship(P > 0.05) was found for explaining soluble solids content (Cittadiniet al., 2004b). The variance in fruit weight (FWV) was positively corre-lated to MFW with:

FWV = MFW � 0.236 � 0.716 (R2 = 0.30) (5)

The results at both spur and whole-tree level showed a negative linearrelationship between MFW and FNLAR, indicating increasing sourcelimitation as FNLAR increased, supporting results reported by Roperand Loescher (1987) and Whiting and Lang (2004). Most studies on theeffect of FNLAR on fruit weight refer to isolated spurs or branches(Facteau et al., 1983; Loescher et al., 1985; Roper and Loescher, 1987).Facteau et al. (1983) observed strong effects at spur level when they wereisolated by girdling, but no (or much weaker) effects on non-girdledspurs. Loescher et al. (1985) reported that on isolated spurs with 3 to 5fruits per spur, the fruits were smaller, matured later, and had less colorthan on adjacent non-isolated spurs. Such differences were not ob-served with less than 3 fruits per spur. These results suggest that whenfruit set is high, at least some of the carbohydrates for fruit growth mustcome from other sources, because the photosynthetic capacity of spurleaves is insufficient to support potential sweet cherry growth (Roper

Cittadini et al. 11

and Loescher, 1987). In our study, results at whole-tree level were moreconsistent than at the spur level. At the spur level, the regression analy-sis between MFW and FNLAR only explained between 25 and 27% ofthe variability. As spurs were not isolated, these results strongly suggestthat fruits of a spur were not only supplied by leaves on that spur, butalso from less fruit-loaded spurs, non-fruiting shoots or reserves.

Whereas FNLAR appears the main determinant of fruit weight(Figure 1), reserves could also play a role in the data variability at wholetree level, although Whiting and Lang (2004) attributed a minor influ-ence of crop management history on fruit weight variation. At anthesis,sweet cherry has very little leaf area, and vegetative buds usually openand develop simultaneously (Flore et al., 1996; Keller and Loescher,1989; Loescher et al., 1985). However, at bud-break, reserves stored inbark, wood and roots (Loescher et al., 1985) provide the carbohydratesfor growth, until assimilated supply from the leaf area of the tree is suf-ficient to satisfy sink demand.

According to Proebsting (1990), fruit size of different cultivarsshould be compared at the same FNLAR. Moreover, cultivars may dif-fer in the sensitivity of their fruit weights to changes in FNLAR. Thus,‘Van’ had the lowest MFW under low or moderate FNLAR, but at highfruit densities its MFW would be less affected by FNLAR (Table 1). Inour study, comparison was based on the regression parameters for thethree cultivars. The Y-intercepts can be considered to represent poten-tial fruit weights (MFW when FNLAR approaches zero).

In the study of Proebsting (1990), fruit weight of the cultivar ‘Rain-ier’ was 2 g higher than that of ‘Bing’ at a similar total yield per tree.Thinning the crop, to reduce yield by half, resulted in increased fruitsize for both cultivars, but MFW of ‘Rainier’ still exceeded that of‘Bing’ by 2 g. This was analogous to the difference between ‘Lapins’and ‘Bing’ in the present study at the whole-tree level (Table 1). How-ever, ‘Van’, having the lowest potential fruit weight, was less sensitiveto increasing FNLAR, suggesting that at high FNLAR it had a highercapacity to mobilize reserves or transport carbohydrates from lessfruit-loaded spurs or non-fruiting shoots.

Experiment 2

Mean LAI of individual trees at harvest was 3.6 (standard deviation:0.74), resulting in an estimated PAR interception of almost 79% (Equa-tion 3). While MSG and TCSAI were not significantly correlated toFNLAR (P = 0.369 and P = 0.092, respectively), MFW, TA, and SSC


decreased linearly with increasing FNLAR (P < 0.05) at rates of 0.029g, 0.027 ml NaOH and 0.065%, respectively, per unit FNLAR (Figures2A to 2D). Firmness showed the same tendency, at a rate of 0.1016Durofel index units per unit FNLAR, but the relationship was not sig-nificant (P = 0.08); neither was the relationship (P = 0.40) with SSC/TA. The coefficients of determination were rather low, that is, 0.47,0.19 and 0.34 for MFW, TA, and SSC, respectively.

In cherry production, about 75% of PAR interception at harvest hasallowed maximum yields (Balkhoven-Baart and Groot, 2005; Floreand Layne, 1990). Favorable fruit quality also requires a good light dis-tribution throughout the canopy (Lang, 2005). Estimated PAR intercep-tion in this experiment was thus near-optimal for sustainable fruitproduction. Under these conditions, a linear negative relationship be-tween FNLAR and MFW was found, in agreement with the results ofearlier studies (Facteau et al., 1983; Flore, 1985; Roper and Loescher,1987; Whiting and Lang, 2004), indicating increasing source limitation

Cittadini et al. 13

6

15

17

19

21

23

25

27

0 20 40

FNLAR (fruit/m2 LA)

Mea

n fr

uit w

eigh

t (g)

Sol

uble

sol

ids

cont

ent (

%)

Titr

atab

le a

cidi

ty(m

l Nao

H)

Firm

ness

(Dur

ofel

inde

x: 0

-100

)

FNLAR (fruit/m2 LA)

FNLAR (fruit/m2 LA) FNLAR (fruit/m2 LA)

60 80 100

0 20 40 60 80 100

7

8

9

10

11A B

D

100 20 40 60 80 100

11

12

13

14

15

16

45

50

55

60

65

70

75

0 20 40 60 80 100

C

FIGURE 2. Effect of fruit number to leaf area ratio (FNLAR) on (A) mean fruitweight (MFW; g/fruit), (B) titratable acidity (TA), (C) soluble solids content(SSC) and (D) firmness (F). A: y = 9.734 � 0.02881x, R2 = 0.47, P < 0.001. B: y =14.784 � 0.0272x, R2 = 0.19, P < 0.044. C: y = 25.15 � 0.0649x, R2 = 0.34, P <0.008. D: y = 64.07 � 0.1016x, R2 = 0.13, P < 0.082.

as FNLAR increases. The relationship at high LAI was consistent withthe one at lower LAI (2.5) (and at an earlier stage of fruit maturity)in Experiment 1, in which MFW (g) was predicted as 9.20 � 0.0213FNLAR.

The SSC/TA ratio has been found to be correlated to the perceptionof sweetness, sourness or cherry flavor by trained judges (Crisosto et al.,2002). High TA negatively affected consumer acceptance in the Americanmarket if SSC was below 16 and 13% for ‘Brooks’ and ‘Bing’, respec-tively, while consumer acceptance was high for both cultivars with SSCexceeding 16%, irrespective of TA (Crisosto et al., 2003). As in previ-ous studies (Flore, 1985; Roper and Loescher, 1987), in this experimentSSC decreased as FNLAR increased, but it was higher than 16% even atthe highest recorded value of FNLAR.

The negative relationship between firmness and FNLAR was not sig-nificant (P = 0.08), but other authors have observed the same pattern(Facteau et al., 1983; Whiting and Lang, 2004). Quantification of thisrelationship would be even more important in other cultivars, in whichlack of firmness is more characteristic than in ‘Bing’.

The lack of significant effects of FNLAR on vegetative growth (TCSAIand MSG) suggested that, although some competition between vegeta-tive and reproductive growth may occur, fruits are the main sink organs.Thus, low FNLAR stimulates growth of the remaining fruits, rather thanvegetative growth, as found in earlier studies (Chalmers and van denEnde, 1974; Grossman and DeJong, 1994; Marcelis and Heuvelink, 1999).

“Price-Fruit Quality” Relationship

Fruit quality and price are closely related and differentiated by mar-kets: farm gate prices are much lower in the domestic than in the inter-national market. Exportable fruit can be transported from Patagonia toEurope by ship or by air. Transport costs are approximately 0.4 and2 US$/kg by ship and by air, respectively. The difference of 1.6 US$ isusually partitioned as: 0.2 US$ to the grower, 0.2 to the packing house,0.6 to the importer in Europe, whereas the remaining 0.6 US$ isreflected in the consumer price (A. B. Pugh, INTA–EEA Chubut, per-sonal communication).

Fruit color is the main indicator of maturity (Mozeti� et al., 2004;Usenik et al., 2005). To be marketable, the fruit needs an appropriatecolor, according to the market and the distance to its final destination(no. 3 or 4 on the CTIFL color chart). The stem of the fruit has to be


present and in good condition, and the fruit should not show any rot, pit-ting, brushing, bruising, deformity and/or damage of any kind. Fruit thatdoes not fulfill these requirements is used for processing and brings thelowest price. Moreover, minimum threshold values are required forsize, SSC and F and is set depending on the market (industry, domestic,and export) and means of transport (truck, ship, and air) (Table 2). Prac-tically all fruit sold in the domestic market is transported by cooled-truck to the Central Market of Buenos Aires (MCBA) (at 1500 to 2100km from the production areas considered in the present study), fromwhere it is distributed to other cities. Firmness of fruit on the Durofel in-dex should be >40 for the domestic market and for export to Europe atleast 60 for ship transport and >50 by air. There is no restriction on solu-ble solids levels for the domestic market, but for export a minimum of14% is required. Minimum size (equatorial diameter) is 22 and 24 mmfor the domestic and the export market, respectively. When all thresholdvalues are met, price is mainly defined by fruit size, set in 2 mm classes(Table 3).

Cittadini et al. 15

TABLE 3. Farm-gate price (US$/kg) based on fruit size (mm), differentiated bymarket and transport possibilities (A. B. Pugh, INTA–EEA Chubut, personalcommunication).

Market 22-24 24-26 26-28 28-30 30

Domestic 0.55 0.69 0.90 1.03 1.17

Export (transport by ship) NM 1.70 1.80 1.95 2.05

Export (transport by air) NM 1.50 1.60 1.75 1.85

NM: Not marketable.

TABLE 2. Threshold values for quality parameters according to the differentmarket and transport possibilities (A. B. Pugh, INTA–EEA Chubut, personalcommunication).

Quality parameter Industry Domestic Export

(truck) Ship Air

Size (mm equatorial diameter) NR 22 24 24

SSC (%) NR NR 14 14

Firmness (0 to 100 Durofel® 25) NR 40 60 50

NR: No restrictions; SSC: Soluble solids content.

Fruit price is not only dependent on quality, but also on timing of mar-keting. Highest prices in Europe are obtained around Christmas, betweenDecember 20 and 30 (G. Laborde, CTIFL, personal communication). Tosell the fruit at that specific moment, it should arrive between December10 and 15. Consequently, taking into account the time needed for pro-cessing and transporting, harvesting in Patagonia should be, at the lat-est, 5 days earlier (between December 5 and 10) for air transport andabout 30 days earlier (between November 10 and 15) for transport byship. In practice, however, harvesting has to be spread over a longer pe-riod to more efficiently use packing facilities and labor.

Parameters Estimation for a “Target-Tree”

As defined earlier, “target-trees” are designed to maximize GVP atfarm level. Therefore, a conceptual model to estimate GVP was de-fined. The model assumes 5% of the fruit for industry and 15% for thedomestic market (minimum values found in practice, A. B. Pugh,INTA–EEA Chubut, personal communication), due to various defects(lack of stem, inadequate color, pitting, brushing, bruising, doublefruits, deformities, etc.). For the remaining fruit, FNLAR determinesSSC, F and MFW (actually size). These three parameters define meanfruit price (Figure 3), taking into account weight distribution against


2.0

1.5

1.0

Pf (

US

$/kg

)

0.5

0.00 50 100

FNLAR (fruit/m2 LA)

150 200

FIGURE 3. Mean fruit price at farm gate (Pf) as a function of fruit number to leafarea ratio (FNLAR). FNLAR influences Pf indirectly by affecting soluble solidscontent, firmness and mean fruit weight. Fruit price is not a continuous functionof size, but becomes continuous when size distribution is taken into account.

MFW obtained with Equation 5. FNLAR, in combination with MFWand LAI, also determine yield (see Equation 2). Finally, GVP is calcu-lated as the product of yield and fruit price (see Equation 1; Figure 4).

The combination ‘Bing’/‘Mahaleb’ on vase was used to illustrate para-meters estimation (LAI and FNLAR) for a “target-tree” in Patagonianorchards. Under this training system, a LAI of 3.07 is required to at-tain the near-optimal situation of 75% interception of PAR at harvest(Equation 4).

To estimate optimum FNLAR, the relationships between fruit qualityparameters and FNLAR found in Experiment 2 were used, that is, SSCbelow 14% was observed with FNLAR exceeding 172 fruit/m2 LA(SSC = 25.15 � 0.065 FNLAR) (Figure 2C), with MFW equal to 4.78 g(9.73 � 0.029 FNLAR) (Figure 2A). A size of 21.6 mm (Equation 4) islimiting market possibilities as fresh fruit, even for the domestic market(Table 2). Similarly, when FNLAR exceeds 40 fruit/m2 LA, F limits

Cittadini et al. 17

Fruit number toleaf area ratio(fruit/ha LA)

Soluble solidscontent (%)

Mean fruit weight(kg fruit–1)

Fraction of fruitwith defects (0-1)

Mean fruit price(US$/kg)

Firmness (1-100Durofel® index)

Gross valueof product(US$/ha)

Yield(kg·ha–1)

Leaf area index(ha LA/ha ground)

FIGURE 4. Factors affecting GVP (gross value of product, US$/ha) at farmlevel. GVP = Yield × Pf (price at farm gate, US$/kg). Yield (kg�ha�1) = MFW ×FNLAR × LAI; where MFW is mean fruit weight (kg/fruit); FNLAR–fruit numberto leaf area ratio (fruit/ha LA) and LAI–leaf area index (ha�ha�1). Price is de-fined based on the information presented in Tables 1 and 3 and integrated inFigure 3.

export possibilities for cherries by ship (F lower than 60; Figure 2D).When F is below 50, associated with more than 138 fruit/m2 LA and aMFW of 5.76 g (22.7 mm), cherries can only be sold in the domesticmarket (Table 2). Although color is an important quality characteristic,it can be optimized through harvesting time. TA can affect consumeracceptance, but only when SSC is below 16% (Crisosto et al., 2003),which was never observed in the experiments described.

At the estimated optimum LAI of 3.07, the “fruit quality-FNLAR”relationships found in Experiment 2, the quality thresholds for differentmarketing possibilities defined in Table 2 and the “price-fruit quality”relationship described in Table 3 and Figure 3, maximum GVP was esti-mated at 21900 US$/ha (Figure 5), with a FNLAR of 80 fruits/m2 LA.Yield was 18.25 t�ha�1 (Figure 6) and MFW was 7.43 g (24.7 mm), with56.2% of the fruit in the category 24 to 26 mm, 27.4% in 22 to 24, 14.4%in 26 to 28, 1.6% below 22 mm and 0.4% between 28 mm and 30 mm, ac-cording to the weight distribution derived from Equation 5 (Figure 7). Fand SSC were 55.9 and 20.0%, respectively. Mean fruit price at farmgate was 1.20 US$/kg.

Changes in the “price-fruit quality” relationships modify optimumFNLAR. For example, if the threshold size for selling fresh fruit in the


25000

20000

15000

10000

5000

00 50 100

FNLAR (fruit/m2 LA)

GV

P (

US

$/h

a)

150 200

FIGURE 5. Gross value of product (GVP) of Patagonian orchards at farm levelas a function of fruit number to leaf area ratio (FNLAR), for ‘Bing’/‘Mahaleb’trees with LAI of 3.07. Maximum GVP (GVPm: 21900 US$/ha) is observed at80 fruit/m2 LA. In the range of 55-103 fruit/m2 LA, GVP exceeds 90% of GVPm(19700 US$/ha).

domestic market moves from 22 to 24 mm and the price for the category26 to 28 mm equals those of fruits exceeding 30 mm, optimum FNLARdecreases from 80 to 70 fruit/m2 LA, resulting in a GVP of 19942 US$/ha, with 16.5 t�ha�1 of fruits of 7.72 g (25.0 mm) on average.

Cittadini et al. 19

30

25

20

Yie

ld (

t·h

a–1)

FNLAR (fruit/m2 LA)

15

10

5

00 50 100 150 200

FIGURE 6. Yield of Patagonian orchards as a function of fruit number to leafarea ratio (FNLAR), for ‘Bing’/‘Mahaleb’ trees with LAI of 3.07. Maximum GVP(GVPm) is obtained at 18.25 t�ha�1 (80 fruit/m2 LA). In the range of 55-103 fruit/m2 LA, corresponding to 13.8 and 21.4 t�ha�1, respectively, GVP exceeds90% of GVPm.

0.0<22

22-2424-26

26-2828-30

>30

0.2

0.4

0.6

Fre

qu

ency

Size category (mm)

FIGURE 7. Distribution of fruit size categories in a ‘Bing’/‘Mahaleb’ “target-tree” trained as vase in South Patagonia, with a mean size of 24.7 mm (meanfruit weight = 7.43 g).

The concept of “target-tree” resulted in a single combination ofLAI (3.07) and FNLAR (80 fruit/m2 LA) to maximize GVP. In prac-tice, such a precise combination will be impossible to attain, but thisapproach also allows determining acceptable deviations from optimalvalues. For example, GVP values exceeding 90% of the maximum(19700 US$/ha) can be attained in the range of 55 to 103 fruit/m2 LA,allowing considerable freedom in management practices (Figure 5).

In this paper, a “target-trees” concept was applied in the perennialcrop sweet cherry, that is, an approach that integrates eco-physiologicalinformation (relationship between fruit quality and FNLAR), model es-timates (relationship between light interception and LAI) and expertknowledge (“price-fruit quality” relationships). Although the examplepresented is limited to a single combination of cultivar and trainingsystem in a specific location, the methodology can be applied to othersituations, provided reliable relevant eco-physiological information isavailable. The approach allows a transparent discussion on the relativemerits of producing high yields of poor quality or maximizing quality atthe expense of yield. The analysis suggests that the optimum is locatedat an intermediate situation, which has to be identified on the basis ofdata from local experiments and “price-fruit quality” relationships.

Once the “target-tree” has been identified, it can serve as the basis for aplanning tool for management practices to promote tree vigor (e.g., win-ter pruning, irrigation, N-fertilization), to weaken the tree (e.g., summerpruning) or to affect fruit set and retention (e.g., hormonal treatments,boron application, beehive density and frost protection). Further researchis needed to quantify and predict the effect of such practices.

LITERATURE CITED

Balkhoven-Baart, J.M.T. and M.J. Groot. 2005. Evaluation of “Lapins” sweet cherryon dwarfing rootstocks in high density plantings, with or without plastic covers.Acta Hort. 667:345-351.

Cavallo, P., S. Poni, and A. Rotundo. 2001. Ecophysiology and vine performance ofcv. “Aglianico” under various training systems. Scientia Hort. 87:21-32.

Chalmers, O. and B. van den Ende. 1974. Productivity of peach trees: Factors affectingdry-weight distribution during tree growth. Ann. Bot. 39:423-432.

Cittadini, E.D., A.B. Pugh, and M. Gigena. 2004a. Factores que afectan la firmeza defrutos de cereza. Proc. XXVII Argentinean Congr. of Hort., Merlo, Argentina,September 21-24, CD-rom, Abstr., F-E-8.


Cittadini, E.D., A.B. Pugh, and M. Gigena. 2004b. Factores que afectan el contenidode sólidos solubles y la acidez titulable de frutos de cereza. Proc. XXVII Argentin-ean Congr. of Hort., Merlo, Argentina, September 21-24, CD-rom, Abstr., F-E-7.

Cittadini, E.D., P.L. Peri, N. de Ridder, and H. van Keulen. 2005a. Relationship be-tween mean fruit weight and the ratio of fruit number to leaf area, at spur andwhole-tree level, for three sweet cherry varieties. Proc. 5th Intl. Cherry Symp.Bursa, Turkey, June 6-10. Abstr., p. 122.

Cittadini, E.D., M.B. Vallés, M.J. Rodríguez, H. van Keulen, N. de Ridder, andP.L. Peri. 2005b. Effect of fruit number to leaf area ratio on fruit quality and vegeta-tive growth of “bing” sweet cherry trees at optimal lai sweet cherry production inSouth Patagonia, Argentina. Proc. 5th Intl. Cherry Symp. Bursa, Turkey, June 6-10.Abstr., p. 124.

Crisosto, C.H., G.M. Crisosto, and P. Metheney. 2003. Consumer acceptance of“Brooks” and “Bing” cherries is mainly dependent on fruit SSC and visual skincolor. Postharvest Biol. Technol. 28:159-167.

Crisosto, C.H., G.M. Crisosto, and M.A. Ritenour. 2002. Testing the reliability ofskin color as an indicator of quality for early season “Brooks” (Prunus avium L.)cherry. Postharvest Biol. Technol. 24:147-154.

Donald, C.M. 1968. The breeding of crop ideotypes. Euphytica. 17:385-403.Facteau, T.J., N.E. Chestnut, and K.E. Rowe. 1983. Relationship between fruit weight,

firmness, and leaf/fruit ratio in Lambert and Bing sweet cherries. Can. J. Plant Sci.63:763-765.

Flore, J.A. 1985. The effect of carbohydrate supply on sour cherry fruit size and matu-rity. Hort. Science 20:568.

Flore, J.A. and D.R. Layne. 1990. The influence of tree shape and spacing on light in-terception and yield in sour cherry (Prunus cerasus cv. Montmorency). Acta Hort.285:91-96.

Flore, J.A. and D.R. Layne. 1999. Photoassimilate production and distribution incherry. Hort. Science 34:1015-1019.

Flore, J.A., C.D. Kesner, and A.D. Webster. 1996. Tree canopy management and theorchard environment: Principles and practices of pruning and training, pp. 259-277.In: A.D. Webster and N.E. Looney (eds.). Cherries: Crop physiology, productionand uses. CAB Intl., Cambridge, UK.

Goudriaan, J. and H.H. van Laar. 1994. Modelling potential crop growth processes.Textbook with exercises. Kluwer Academic Publishers, Dordrecht, The Netherlands.

Grossman, Y.L. and T.M. DeJong. 1994. PEACH: A simulation model of reproductiveand vegetative growth in peach trees. Tree Physiol. 14:329-345.

Jackson, J.E. 1980. Light interception and utilization by orchard systems. Hort. Rev.2:208-267.

Keller, J.D. and W.H. Loescher. 1989. Non-structural carbohydrate partitioning inperennial parts of sweet cherry. J. Amer. Soc. Hort. Sci. 114:969-975.

Lang G.A. 2005. Underlying principles of high density sweet cherry production. ActaHort. 667:325-333.

Lappi, J. and P. Stenberg. 1998. Joint effect of angular distribution of radiation andspatial pattern of trees on radiation interception. Ecol. Model. 112:45-51.

Cittadini et al. 21

Loescher, W.H., T. Roper, and J.D. Keller. 1985. Carbohydrate partitioning in sweetcherries. Proc. Annu. Mtg. Wash. State Hort. Assoc. 81:240-248.

Marcelis, L.F.M. and E. Heuvelink. 1999. Modeling fruit set, fruit growth and drymatter partitioning. Acta Hort. 499:39-49.

Mozeti�, B., P. Trebše, M. Sim�i�, and J. Hribar. 2004. Changes of anthocyanins andhydroxycinnamic acids affecting the skin colour during maturation of sweet cher-ries (Prunus avium L.). Lebensmittel-Wissenschaft und-Technologie 37:123-128.

Patrick, J.W. 1988. Assimilate partitioning in relation to crop productivity. Hort. Sci-ence 23:33-40.

Payne, R.W. 2002. The Guide to GenStat, Release 6.1. Part 1: Syntax and Data Man-agement, Part 2: Statistics. VSN Intl., Oxford, UK.

Proebsting, E.L. 1990. The interaction between fruit size and yield in sweet cherry.Fruit Var. J. 44:169-172.

Proietti, P., A. Palliotti, F. Famiani, E. Antognozzi, F. Ferranti, R. Andreutti, andG. Frenguelli. 2000. Influence of leaf position, fruit and light availability on photo-synthesis of two chestnut genotypes. Scientia Hort. 85:63-73.

Roper, T.R. and W.H. Loescher. 1987. Relationships between leaf area per fruit andquality in “Bing” sweet cherry. Hort. Science 22:1273-1276.

Sedgley, R.H. 1991. An appraisal of the Donald ideotype after 21 years. Field CropsRes. 26:93-112.

Usenik, V., D. Kastelec, and F. Štampar. 2005. Physicochemical changes of sweetcherry fruits related to application of gibberellic acid. Food Chem. 90:663-671.

UTHSCSA. 2002. http://ddsdx.uthscsa.edu/dig/itdesc.html (Access date: 18 March2004).

Wagenmakers, P. 1994. Light relations in orchard systems. Ph.D. Thesis. WageningenAgricultural University, Wageningen, The Netherlands. 151 p.

Whiting, M.D. and G.A. Lang. 2004. “Bing” sweet cherry on the dwarfing rootstock“Gisela 5”: Thinning affects fruit quality and vegetative growth but not net CO2exchange. J. Amer. Soc. Hort. Sci. 129:407-415.

doi:10.1300/J492v06n03_02


http://ddsdx.uthscsa.edu/dig/itdesc.html

Designing a “Target-Tree” for Maximizing Gross Value of Product in Patagonian Sweet Cherry...

Documents

Transcript of Designing a “Target-Tree” for Maximizing Gross Value of Product in Patagonian Sweet Cherry...