developing mechanical harvesting for california black ripe

Ferguson et al. Final Report 2009, California Olive Committee

3

Developing Mechanical Harvesting for California Black Ripe Processed Table Olives: 2007-2010: Final Report

Project Leaders: Louise Ferguson, Extension Specialist, Department of Plant Sciences, 2037 Wickson Hall, Mail Stop II, UC Davis, 1 Shields Ave., Davis CA 95616, (530) 752-0507 [Office], (559) 737-3061 [Mobile], [email protected] Jackie Burns, Professor, Department of Horticulture, University of Florida, [email protected] Jean-Xavier Guinard, Professor, Department of Food Science and Technology, UC Davis, [email protected] Uriel Rosa, Associate Professor, Department of Bioagricultural and Mechanical Engineering, UC Davis, [email protected] Cooperating Personnel: Sergio Castro Garcia, Visiting Scientist from University of Cordoba, Spain Kitren Glozer, Associate Project Scientist, UC Davis William H. Krueger, UCCE Farm Advisor, Glenn County Neil O’Connell, UCCE Farm Advisor, Tulare County Elizabeth J. Ficthner, UCCE Farm Advisor, Tulare County John Henry Ferguson, Volunteer Maria Paz Suarez Garcia, Visiting Scientist from University of Seville, Spain Industry Cooperators Ranch Cooperators Harvester Cooperators Processor Cooperators Rocky Hill Ranch Agright Bell Carter Olives Erick Nielsen Ranch Erick Nielsen Inc. Musco Family Olive Company Reporting Period: 15 April – 31 December 31, 2009 (year 3/4)

ABSTRACT

This is the third of a four-year project to develop economically feasible mechanical harvesting for California black ripe Manzanillo table olives. The project has simultaneously focused on three factors: 1) harvesting technology; 2) adapting current orchards with mechanical pruning and developing new hedgerow orchards; and 3) screening for an abscission agent to increase harvester efficiency by decreasing fruit removal force (FRF).

The limiting factors to successful mechanical harvesting are, in order of importance, processed fruit quality, harvester removal efficiency, and economically feasible harvester operating parameters (harvest time and cost per ton) that produce a positive net return. Research from 2006 through 2009 demonstrated that both canopy contact heads and trunk shakers can produce olives with canning percentages and adjusted values per ton equal to those of hand

mailto:[email protected]�





4

harvested olives. In this 2009 report, the data produced by Dr. JX Guinard and MS. Soh Min Lee, Ph D candidate, definitively demonstrated that neither trained sensory nor consumer panels could distinguish 2008 canopy contact head mechanically harvested olives from hand harvested olives. However, while 2009 trunk shaker harvested olives had canning percentages and adjusted values per ton equal to those of the olives harvested by the canopy contact shaker in 2008 sensory and consumer panels could distinguish between hand and mechanically harvested olives. And the consumers preferred the hand-harvested fruit. However, we suspect the over ripeness of the 2009 olive harvest was somewhat responsible for this result. Therefore the most limiting factor, competitive fruit quality, has not been totally eliminated as a research objective. In 2010 we will have the ability to test both a canopy contact and a trunk-shaking harvester in the same two orchards, along with a hand harvest control.

The next most limiting factor is achieving the needed 80% final harvester removal efficiency. Thus far, the final efficiencies achieved with the different canopy contact harvesters are as follows: DSE, 58%, Agright Olivia, 67% and 97% for the MaqTec Colossus. However, the latter produced virtually 100% fruit damage. The trunk shaking harvesters have produced the following final harvest efficiencies, ENE Inc, 65%, Spanish Noli, 70%, Coe Machinery, 68% and Spanish wraparound, 62%. Therefore, none of the harvesters have achieved the needed 08% final harvest efficiency. Research in 2010 will again focus on improving the harvesters and harvester operating parameters. The harvesters evaluated will be the ENE Inc. trunk shaker and the Coe canopy contact head harvester. Both were selected based on prior performance, and have catch frames. The Coe does not have the high harvester weight and tree height limitations of the Colossus and Agright canopy contact over the row harvesters.

However, improving the harvester is not the only way to improve harvester efficiency. As can be seen in the successful example of California’s high-density olive oil orchards, the tree can be adapted to make fruit more accessible to the harvester with pruning and training into hedgerows. Thus far, our long term mechanical hedging and topping trial, in the second of 6 years is not demonstrating significant cumulative yield losses. However, in 2009, when yields were very low, the mechanically pruned trees did have significantly lower yields than the hand pruned control trees. The long term training trial at Nickels Soils Laboratory is demonstrating 9-year-old trees trained to a trellis can yield as well as conventionally pruned trees.

Our third objective, screening for abscission chemicals has not produced any viable candidates. This experimental approach for this objective will be reexamined and a new approach developed.

INTRODUCTION

Why mechanical harvesting of California black ripe table olives is needed

The California table olive industry will need to develop mechanical harvesting for economic survival. If hand harvesting costs $350.00/ton and growers receive $875.00/ton harvest costs are consuming 40% of the gross returns. Using these figures growers must produce 5 tons/acre to net $81.00/acre after all costs; see page 15 of the online cost study at http://coststudies.ucdavis.edu/files/olivesv09.pdf.

Further, current economic conditions and immigration enforcement efforts suggest manual labor pools will decrease. Even if they do not, the logistics and administration of seasonal manual labor are difficult. Thus, developing mechanical harvesting is both an economic and logistical necessity for the California black ripe processed table olive industry.

http://coststudies.ucdavis.edu/files/olivesv09.pdf�


5

Why mechanical harvesting is difficult to develop Olives destined for California black ripe table processing are harvested physiologically

immature; the abscission zone between the fruit and stem is undeveloped. The fruit is borne on pendulous, flexible, vertical shoots at the outer periphery of a thick canopy that can significantly damp much of the force applied to it. To remove an immature individual 5-10 gram olive fruit can require a kilogram of pull force. Generation of this much force can damage the olive and the tree branches or trunk. How we are approaching developing mechanical harvesting

We are approaching developing mechanical harvesting by simultaneously evaluating the components of mechanical harvesting: 1) Picking technology and associated harvester; 2) Olive tree training and orchard spacing; and 3) Screening abscission compounds for decreasing FRF. Research progress from 2006 through 2009

The three areas of research identified above must all overcome the same limiting factors for successful mechanical harvesting of table olives. These are, in order of importance; 1) processed fruit quality; 2) harvester removal efficiency; and 3) economically feasible harvester operating parameters; harvest times and costs per ton and acre, which produce a positive net return. Our progress from 2006 through 2009 is summarized below.

Picking technology and associated harvester

The two major methods of picking technology investigated thus far are canopy contact heads and trunk shakers. Our observations are that canopy contact heads remove exterior fruit more efficiently. Trunk shakers remove the fruit closer to the tree trunk more efficiently. The major limiting factor of fruit damage with the canopy contact harvester has been overcome. However, while the 2009 sensory and consumer evaluations indicate we need to focus on eliminating fruit damage from the trunk-shaking harvester this quality decrease may have been more a function of low crop and overripe fruits than machine damage. Tree branch damage from the canopy contact head and trunk damage from the shaker have also been largely eliminated. However, both picking technologies are marginally efficient in terms of final % fruit removal. Both are achieving less than the 80% final removal efficiency Dr. Klonsky, Agricultural Economist, has demonstrated is needed.

Reliable operating parameters of acres, tree, tons per hour, or cost to harvest the same, have not been determined. Most machines evaluated thus far can operate between 0.5 and 1 mph. Our results thus far also indicate modified prune and pistachio harvesters can serve as efficient catch frames.

Young tree training and orchard spacing

Harvester evaluations thus far strongly suggest that altering the tree to make fruit more accessible to the harvester would be the best method of increasing harvest efficiency. Yield records of a 9-year-old orchard spaced at 12 x 18 feet have demonstrated trees trained to a hedgerow yield as well as conventionally trained trees thus far. The results will be conclusive when the orchard yields level off for at least three years. Mechanical pruning and hedging to shape current orchards did significantly decrease yield in the second of the first two years of this 6-year trial. However, the yields in this trial thus far, due to poor fruit set, have been so low as to render this data unreliable. In both trials it is hoped that the decreases in harvest costs will offset the losses in yield due to pruning to generate a better net return.


6

Abscission compounds for decreasing FRF Logically, it seems that the primary objective of any mechanical harvesting project should be

developing an effective abscission agent to accelerate the development of the fruit abscission zone and decrease the required FRF and therefore require less forceful trunk shaking or canopy contact head picking force. The hoped for result would be higher final harvest efficiencies, a higher percentage of cannable fruit with higher adjusted values per ton, and less tree damage. Ethylene releasing compounds, (ERCs), primarily ethylene, have demonstrated the best potential thus far. However, decades of abscission research with ethylene, and our recent results, have demonstrated that the abscission zone of olive leaves is more sensitive to ERCs than the abscission zone of olive fruits. The usual result is unacceptable leaf losses (over 25%). Further, ERCs have performed erratically in both multiple trials in the literature as well as our trials. It appears that a new research approach, one more developed than screening potential abscission compounds, is needed.

Finally, even if an effective abscission agent is developed it will require at least 5 years and considerable funding to generate the efficacy, residue, and environmental impact research (EIR) required for registration. In this time the California table olive industry could decline below critical volume. The most effective approach at this time is to pursue mechanical harvesting as if an abscission compound will not be available in the near future. And to develop a more considered research approach to olive fruit abscission.

In summary, when initiated in 2006 the mechanical harvesting research program focused on simultaneously developing a specific canopy contact head harvester, and identifying an abscission agent to increase harvester efficiency and decrease harvester fruit and tree damage. However, results through 2009 strongly suggest that the final year of this project should focus on two priorities. First, research should focus on adapting trees for mechanical harvesting through pruning of existing orchards and training new hedgerow orchards. Second, all available commercial harvesters with potential for table olives should be adapted and evaluated on these reconfigured orchards. Trials thus far have demonstrated that olives harvested by both canopy contact heads and trunk-shaking harvesters have fruit quality and values that are statistically equal to those of hand harvested fruit. Trained sensory and consumer panels could not distinguish the canopy contact harvested processed olives were from hand harvested olives though they could distinguish the trunk shaker harvested processed olives from the hand harvested olives. However, we think this latter result may have been more a function of over ripeness than harvester damage. However, the final efficiencies of both types of harvesters remain below the needed 80%. Finally, development of an abscission compound for decreasing FRF should not be pursued until a more effective approach can be developed. The following 2009 research report supports these conclusions.

OBJECTIVES

The 2009 research had four major objectives; outlined below. Each objective will be reported in an individual section of this report as listed below.

I. Sensory Characteristics and Consumer Acceptance of Mechanically-Harvested California Black Olives A. 2008 harvest with Canopy Contact Shaker: DSE 008 B. 2009 Harvest with ENE Inc. Trunk Shaker: Terry II


7

II. Evaluation of Mechanical Harvester(s) Efficiency and Effects on Fruit Quality and

Value A. ENE Inc. trunk shaker B. Noli trunk shaker C. Agright canopy contact head harvester

III. Evaluation of Pruning and Training Methods on Tree Yield and Fruit Quality

A. Evaluation of Mechanical Topping and Hedging on Existing Orchards B. Evaluation of Developing New Olive Hedgerow Orchards

IV. Screening for Potential Abscission Compounds

OBJECTIVE I

Objective 1.A.

Sensory Characteristics and Consumer Acceptance of

DSE 008 Canopy Contact Mechanically-Harvested California Black Ripe Olives 2008 Harvest Season

Soh Min Lee, Louise Ferguson and Jean-Xavier Guinard

Fruit harvested by the improved DSE 008 canopy contact harvester in 2008 was either processed fresh, or stored and processed in early 2009 by both Bell Carter and Musco Olive Company. The field procedures for the 2008 harvesting the olives are detailed below. Field Procedures: Location: Block 17W Rocky Hill Ranch, Exeter CA

• Planted 1998 • 6 rows, 83 trees per row, ‘Manzanillo’ olives with ‘Sevillano’ pollinators • Spaced @ 12 X 26 feet, 139 trees per acre

29 – 20 September 2008 The six tree rows were divided into five 14-tree replications with 1 buffer tree at each row end, and two buffer trees between each 14-tree replication.

• 1 replication per row (six row total) was hand harvested as a control • 1 replication per row (six row total) were harvested with the DSE 008 canopy contact

head harvester o Dropped fruit were collected and weighed; but not combined with harvested fruit o Each tree was hand gleaned, and fruit weighed; but not combined with harvested

fruit • The six hand harvested replications and six machine harvested replications were

maintained in separate bins


8

• The separate bins were reweighed at Musco receiving station to confirm field weight • A COC sample grade was done for each bin/replication

o A 40 pound sample of extra large/large fruit was collected for each replication by running the fruit over the sizer

o The 40 pound sample was separated into two 20 pound samples each for Bell Carter and Musco One sample was processed fresh One sample was processed stored The samples were sent to the two processors that night

2 (harvest methods) X 6 (14 tree replications) X 2 (processors) X 2 (processing methods) = 48

samples total Table 1.A.1 gives the effect of harvest method on the percent cannable fruit and adjusted value per ton. As can be seen in this table, mechanically harvesting with a canopy contact head significantly lowered the percentage of cannable olives and the adjusted value per ton. However, the major factors causing these decreases were significant increases in the percentages of trash and culls in mechanically harvested fruit (data not shown). The high percentage of trash was the result of an inoperative blower. The higher percentage of culls was the result of overripe fruit. Table 1.A.1. Average, and statistically analyzed, receiving station grades for the percentage of cannable fruit and adjusted value per ton for the hand and machine harvested olives.

Effect of Harvest Method on Olive Grade and Value: 2008

Harvest Method Percentage Cannablea Adjusted price/ton ($) a

Mechanical 88.0*** 1013.80***

Hand 96.2 1137.80 a Means separation within columns were performed with PROC TTEST procedure of SAS (SAS Institute Inc., Cary, NC); *, **, *** = 0.05, 0.01 and 0.001 level of significance

These were not factors generated by the harvester. Also, 88% cannable fruit valued at $1,113.80 per ton is well within acceptable ranges for processing California black ripe table olives.

The above data demonstrates that the canopy picking head can produce commercially acceptable quality fruit if receiving station grade and value are the final criterion. However, it is the sensory and consumer evaluations detailed in the report below that confirm these mechanically harvested olives can produce processed olives with sensory quality equal to that of hand harvested product and are acceptable to consumers.

Considering that these olives (Table 1.A.1) delivered to Lee and Guinard for evaluation had a significantly lower canning percentage and adjusted value per ton, it is even more impressive that their evaluation results definitively demonstrated there is little distinguishable difference between hand and mechanically harvested olives.


9

Sensory Characteristics and Consumer Acceptance of Canopy Contact Mechanically Harvested California Black Olives: 2008 Crop Year

OBJECTIVES of OBJECTIVE I.A.

The main objective of the study was to compare the sensory properties and acceptability of

hand-harvested and mechanically harvested table olives. Another objective was to examine the effects of storage before processing on the sensory quality of the olives. Also, two different commercial processors processed the experimental samples, and they were compared to commercial offerings from these processors.

PROCEDURES of OBJECTIVE I.A.

The study examined the sensory properties and acceptability of 10 California black table olive samples that were produced according to the experimental design described below. The variables in the design were harvesting method—hand vs. mechanical; commercial processor—Musco vs. Bell-Carter; and processing method—olives processed fresh vs. olives processed after storage. We also added two commercial products to the design, one from each processor. The samples in the design and the two commercial products are shown in Table 1.A.2 below.

Table 1.A.2. Table olive samples.

Sample abbreviation Processor Commercial Harvesting method

Processing method

A_Comm A Commercial - - A_Hand_F A - Hand Fresh olives A_Hand_S A - Hand Stored olives A_Mach_F A - Machine Fresh olives A_Mach_S A - Machine Stored olives

B_Comm B Commercial - - B_Hand_F B - Hand Fresh olives B_Hand_S B - Hand Stored olives B_Mach_F B - Machine Fresh olives B_Mach_S B - Machine Stored olives

Descriptive analysis The sensory properties of the olives were measured by descriptive analysis with a trained

panel of eight judges. In descriptive analysis, the panel rates the intensity of the sensory attributes of the products. Using a method that combined elements of the Quantitative Descriptive Analysis and the Spectrum Method, the panel rated the intensities of 31 attributes of appearance; flavor (taste and smell), texture and mouth feel across the 10 samples (Table 1.A.3). Reference standards for the flavor attributes in the scorecard were prepared to ensure concept alignment among the judges. Group and individual training sessions were held until the panel was deemed ready to proceed with the actual descriptive analysis. All products were evaluated


10

in triplicate following a randomized complete block design. Olives were presented sliced in half, at room temperature, in a spherical glass covered with a plastic lid (three olives per glass). The intensity of the attributes was rated on a category line scale anchored with ‘low’ and ‘high’ labels (except for the attribute ‘glossy’, which used the labels ‘dull’ and ‘glossy’). Table 1.A.3. Sensory attributes in the descriptive analysis scorecard. Attribute Reference Attribute Reference

Smel

l (A

rom

a)

Painty Correction fluid

Flav

or

Sweetness Sucrose solution Briny Black olive brine Saltiness NaCl solution Ocean-like Green seaweed +

anchovy* Umami MSG +Brine

Fermented Sauerkraut* Bitterness Caffeine solution Canny Keys, cans Roasted Roasted sunflower

seeds Earthy Potting soil* Buttery Melted butter

+olive brine * Sautéed mushroom

Sautéed Mushroom* Ripeness Unripe --- Ripe

Dried fruit Dried Prune Te

xtur

e Firmness

Floral Chrysanthemum tea Juicy/ Moist release

App

eara

nce

Size Small --- Large Crumbly Oval Round --- Oval Fibrous Surface roughness Smooth --- Rough

(wrinkles, cracks)

Afte

r tas

te/

mou

thfe

el Mouth coating

Glossy Dull --- Glossy Briny after-taste Skin brownness Black --- Brown Lasting flavor Flesh Brownness Black --- Brown Astringent Flesh greenness Black --- Green

* mixed with olives Consumer testing

The olive samples were also evaluated by 100 consumers, on Picnic Day (UC Davis’ annual open campus event on 18 April 2009) or during the summer of 2009 (between 25 June and 3 July). Consumers were recruited among Picnic Day visitors and Davis Farmer’s Market customers. To qualify for the study, consumers had to be US Residents and users and likers of black table olives.

Each consumer was presented with 11 samples (the first one was a primer, used to eliminate the first-order effect typically encountered in consumer tests—the first sample receives a higher hedonic score than the subsequent samples in the serving order). The order of presentation of the samples was randomized across consumers. Consumers rated overall degree of liking of the samples on the 9-point hedonic scale, from 1=’dislike extremely’ to 9=’like extremely’, with 5=’neither like nor dislike’. They also rated degree of liking of the appearance, flavor, and texture of the samples. Olives (two olives per cup, sliced in half) were served in plastic cups


11

covered with lids at room temperature. Crackers and water were provided for rinsing and palate cleansing. Upon completion of the tasting, consumers filled an exit survey with demographic and olive usage information. Data analysis

The descriptive analysis data were analyzed using a combination of univariate and multivariate statistics. Analysis of variance was used to examine the effects of the variables in the design. Principal component analysis (PCA) was then applied to the matrix of mean intensity ratings across the samples to examine the similarities and differences among the products in the design.

Hedonic ratings by the consumers were analyzed by analysis of variance and the matrix of hedonic ratings across consumers was analyzed by preference mapping — a combination of factor analysis and classification methods designed to assess preference-based market segmentation and to identify drivers of liking for each uncovered segment.

Partial least square (PLS) regression was then used to examine the relation between the hedonic ratings by consumers and the sensory attributes measured by the trained panel.

RESULTS OF OBJECTIVE I.A.

Descriptive analysis

There were no significant differences between mechanically- and hand-harvested olives for any of the sensory attributes in the descriptive profile except surface roughness (Table 1.A.4). There were, however, many significant differences across a range of appearance, flavor, and texture attributes between fresh-processed olives and olives that had been stored before processing. A number of attributes also differed between processors and even more between the two commercial products and the experimental samples in the design.

Because it is nearly impossible to visualize differences among products across so many sensory dimensions, we used principal component analysis to show the relationships among the sensory attributes and the products in a two-dimensional ‘sensory map’ of the products. The principal component biplot shows the relationships among the sensory attributes — attributes, which are positively correlated, tend to form small angles with each other or to be clustered together on the plot, whereas attributes, which are negatively correlated, are found at opposite ends of the plot. It also shows the main sensory features of each table olive sample — attributes located close to a given sample tend to be higher for that sample, whereas attributes, which are found away from that sample, tend to be lower. The biplot of PC2 vs. PC1 is shown in Fig. 1.A.1 below.

Fig. 1.A.1 shows how close to each other the hand- and mechanically-harvested versions of each olive product are located. By contrast, the location of the fresh- and stored-processed olives is different and so is that of samples processed by processors A and B.


12

Table 1.A.4. F-values for partitioned product source of variation.

Commercial vs. non-

commercial

Harvesting method (hand vs. machine)

Processing method (Fresh vs. Stored)

Processors (A vs.

B)

Harvesting method

* Processing

method

Harvesting method

* Processor

Processing method

* Processor

Painty 4.84 2.03 2.80 0.14 0.16 0.14 0.14 Briny 1.46 0.28 0.98 1.11 1.36 0.01 0.01 Ocean-like 7.21 2.65 1.64 16.17 4.38 0.18 1.71 Fermented 6.48 0.02 3.22 5.03 0.92 0.18 0.18 Canny 2.34 0.02 2.10 0.02 0.39 0.03 0.03 Earthy 0.02 0.00 7.77 0.04 1.20 0.00 0.40 Sautéed mushroom

4.88 1.54 3.94 5.71 0.27 2.42 0.06

Dried fruit 3.89 0.31 6.36 1.25 0.02 0.09 0.09 Floral 0.41 0.83 0.10 0.08 0.28 0.08 0.20 Size 36.87 0.16 6.89 1.78 0.13 2.26 0.02 Oval 4.12 0.46 0.01 3.88 0.16 0.01 1.91 Surface roughness

1.72 4.75 2.06 3.10 0.00 0.37 4.29

Glossy 8.05 0.96 0.46 87.50 0.58 0.14 8.14 Skin brownness

2.85 1.67 0.10 57.50 0.43 7.62 7.62

Flesh Brownness

9.16 0.00 2.73 115.09 0.20 1.74 1.74

Flesh greenness

13.31 0.02 7.66 37.09 0.10 2.94 2.94

Sweetness 6.63 0.29 2.66 0.55 1.57 0.44 0.12 Saltiness 17.07 0.65 89.39 3.69 2.33 4.83 4.83 Umami 8.96 0.08 38.54 8.67 1.14 0.88 1.02 Bitterness 10.35 0.73 1.24 1.61 0.08 2.09 2.09 Roasted 1.21 0.02 6.01 3.41 2.27 0.87 0.01 Buttery 9.05 0.43 25.73 3.94 0.85 0.04 0.69 Ripeness 11.25 0.00 34.05 30.50 0.05 0.03 9.07 Firmness 4.46 1.62 9.31 23.47 0.12 0.12 17.12 Juicy/ Moist release

67.49 0.37 73.49 14.20 0.22 0.17 3.16

Crumbly 0.06 0.18 1.81 0.32 0.00 0.15 0.15 Fibrous 2.44 0.36 12.69 23.39 0.21 1.65 14.47 Mouth coating

3.85 0.00 23.08 1.63 0.06 0.96 0.02

Briny after-taste

16.48 0.03 140.08 22.45 0.86 27.75 27.75

Lasting flavor

1.10 0.22 73.17 18.25 0.39 4.14 12.66

Astringent 0.63 0.12 2.90 0.15 0.20 0.96 0.96

*Bold means significant effect (P<0.05)


13


14

Fig 1.A.1. Principal component analysis of the descriptive analysis data showing the attributes and products. Consumer testing

There was no significant difference in acceptability between mechanically- and hand-harvested olives. This was true not only for overall degree of liking but also for degree of liking of appearance, flavor and texture of the olives (Fig. 1.A.2 and Tables 1.A.5 & 1.A.6). There were, however, significant differences in liking between fresh-processed and stored-processed olives. Consumers liked the fresh-processed olives significantly more than the stored-processed olives, and that was true not only for overall degree of liking, but also for liking of flavor and to a lesser extent liking of texture (Fig. 1.A.4 & 1.A.5, and Tables 1.A.4 & 1.A.5). On average, the consumer population that tested the olives liked the four fresh-processed olive samples best. Fig. 1.A.4 below shows the partitioning of the variance in the hedonic data. It is clear that the main source of variation in the data was whether the olives were processed fresh or after storage.

Table 1A.5. F-values for partitioned product source of variation Commercial

vs. non-commercial

Harvesting method

(hand vs. machine)

Processing method

(Fresh vs. Stored)

Processors (A vs. B)

Harvesting method *

Processing method

Harvesting method * Processor

Processing method * Processor

Overall degree of liking

62.68 2.04 58.72 0.01 0.00 0.61 0.48

Appearance liking

1.75 0.07 0.05 0.13 0.38 1.17 3.44

Flavor liking 86.02 0.90 79.38 0.31 0.06 0.41 0.00 Texture liking 2.44 0.06 8.53 5.38 0.07 0.07 3.50 *Bold means significant effect (P<0.05)

Astringent

LastflavorBrinyAF

Mouthcoating

Fibrous

Crumbly Juicy

Firmness

Ripeness

ButterRoasted

Bitterness

Umami

Saltiness

Sweetness

FleshGrn

FleshbrownSkinbrown

Glossy

Surrough

Oval

Size

Floral

DriedFruit

Sauteed

Earthy

Canny

Fermented

OceanBriny

Painty

-1

-0.75

-0.5

-0.25

0

0.25

0.5

0.75

1

-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1D1 (42.56 %)

D2

(37.

72 %

)

BMachS

BMachF

BHandS

BHandFBComm

AMachS

AMachF

AHandS

AHandF

AComm

-2

-1.5

-1

-0.5

0

0.5

1

1.5

-3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

D1 (42.56 %)

D2

(37.

72 %

)


15

Table 1.A.6. Mean hedonic ratings of the 10 olive samples for overall degree of liking, appearance liking, flavor liking and texture liking (N=100 consumers)

Products Overall degree of liking

Appearance liking Flavor liking Texture liking

Acomm 4.57 c 6.07 ab 4.38 c 5.57 c AHandF 6.01a 5.91 ab 6.15 a 5.83 bc AMachF 6.05 a 6.08 ab 6.13 a 5.66 bc AHandS 4.98 bc 6.10 ab 4.90 b 5.55 c AMachS 5.45 b 6.20a 5.29 b 5.71 bc Bcomm 4.60 c 5.74 b 4.35 c 5.67 bc BHandF 5.96 a 6.13 ab 5.99 a 6.15 ab BMachF 6.24 a 6.14 ab 6.17 a 6.31 a BHandS 5.19 b 6.02 ab 5.08 b 5.70 bc BMachS 5.06 bc 5.85 ab 4.97 b 5.67 bc

*Duncan’s multiple range comparison test (alpha= 0.05) was used. Unshared same alphabets in the superscripts mean significant difference. Legend: A and B = processors

Hand = hand-harvested; Mach = machine-harvested S = stored then processed; F = fresh-processed

Fig 1.A.2. a. and b. Mean hedonic ratings of the 10 olive samples for overall degree of liking and appearance liking (N=100 consumers).

a. Overall degree of liking

1

3

5

7

9

Acomm

AHandF

AMachF

AHandS

AMachS

Bcomm

BHandF

BMachF

BHandS

BMachS

scor

es

b. Appearance liking

1

3

5

7

9

Acomm

AHandF

AMachF

AHandS

AMachS

Bcomm

BHandF

BMachF

BHandS

BMachS

scor

es


16

Correlation analysis of hedonic ratings

It is interesting to note that when we examine how liking for the various sensory modalities correlated with overall degree of liking, we find that there was a highly significant correlation between liking for flavor and overall degree of liking (Fig. 1.A.4 and Table 1.A.6). This suggests that even in a texturally-relevant product like table olives, flavor characteristics appear to be driving liking for the product overall. Table 1.A.6. Pearson’s correlation coefficients among hedonic ratings by consumers

Variables Overall degree of liking


Overall 1 0.415 0.995 0.727 *Values in bold are significantly different from 0 with a significance level alpha=0.05

Fig 1.A.3.c. and d. Mean hedonic ratings of the 10 olive samples for overall degree of flavor liking and texture liking (N=100 consumers).

c. Flavor liking

1

3

5

7

9

Acomm

AHandF

AMachF

AHandS

AMachS

Bcomm

BHandF

BMachF

BHandS

BMachS

scor

es

d. Texture liking

1

3

5

7

9

Acomm

AHandF

AMachF

AHandS

AMachS

Bcomm

BHandF

BMachF

BHandS

BMachS

scor

es


17

Fig 1.A.4. Partitioning of product source of variation for overall degree of liking, flavor liking and texture liking (appearance liking is not included since there was no significant effect).

a. Overall degree of liking

Commercial vs. others

Processing method

(Fresh vs. Stored)

Harvesting method (Hand vs. Machine)

Commercial vsothersHarvesting method

Processing method

Processor

Harvest method xprocessing method

Harvest method xprocessorProcessing method xprocessor

b. Flavor liking


Processing method

(Fresh vs. Stored)


Commercial vsothersHarvesting method

Processing method

Processor



c. Texture liking

Processing method x

processor


Processing method

(Fresh vs. Stored)


Processor(A vs. B)

Commercial vs others

Harvesting method

Processing method

Processor




18

A B C

Fig 1.A.5. Scatter plots showing the relation between overall liking and liking for the specific sensory modalities of appearance (A), flavor (B), and texture (C) in the olives. Another way to visualize the slight differences in liking patterns between the two segments is to plot the consumers on the PCA biplot (Fig. 1.A.8) according to their cluster affiliation (Fig. 1.A.7). It can be seen that the small cluster liked the olives that had been stored before processing best, likely because of familiarity with the profile that process generated. It is important to realize, however, that this cluster was very small, with only 13 consumers in it. The main cluster that includes the majority (86) of the consumers liked the fresh-processed samples best, and they are located on the right side of the biplot. The other way the two clusters differed was in the distribution of the sources of variation in the data (Fig. 1.A.9).

A B

Fig. 1.A.6. Internal preference mapping showing the consumers (A) and the olive samples (B) (N=100).


19

Preference mapping

Fig. 1.A.6 shows the results of the preference mapping analysis as a biplot of the first two principal components, showing the consumer (Fig. 1.A.6.A) and the olive samples (Fig. 1.A.6.B). The preference map shows that most of the consumers are located on the right side of the plot (Fig. 1.A.6.B), where the fresh-processed samples are located. This confirms the average data (for all consumers) presented above. It was also surprising to find that most consumers were concentrated away from the two commercial samples in the design, indicating most consumers liked those olives the least. Even though the consumers were fairly homogeneous in their liking patterns, with most of them concentrated on the right side of the biplot, close to the fresh-processed samples, there was some market segmentation that translated into two groups of consumers with slightly different preferences. The results of the preference clustering analysis are shown in Fig. 1.A.7, with a two-segment resolution. However, it should be noted that one cluster includes most of the consumers (n=86) and the other is rather small (n-13).

Fig 1.A.7. Cluster analysis dendrogram of the 100 consumers

Fig. 1.A.8. Internal preference mapping with segmentation (G1=86, G2=13, 1 outlier

with no group)


20


21

A. Partitioning of product source of variation on overall degree of liking for GP 1 (N=86)

B. Partitioning of product source of variation on overall degree of liking for GP 2 (N=13) Fig 1.A.9. A. and B. Partitioning of product source of variation on overall degree of liking for GP 1 (A) and GP 2 (B).


22

We then examined the sensory drivers of liking for the whole consumer population using PLS regression (Figs. 1.A.10, 1.A.11 & 1.A.12). This analysis shows which sensory attributes are associated with overall degree of liking by consumers (Fig. 1.A.10), and more specifically, which flavor attributes are associated with liking for flavor of the olives (Fig. 1.A.11) and which texture attributes are associated with liking for texture of the olives (Fig. 1.A.12).

Finally, we examined how the two preference clusters differed in their drivers of liking (Fig. 1.A.13). It is quite clear that they liked very different attributes in the olives, as indicated by the different set of attributes associated with each cluster’s main direction of preference.

Fig. 1.A.10. PLS-Regression of the consumer hedonic ratings onto the sensory attributes from the descriptive analysis; — overall degree of liking vs. all sensory attributes. - X data; DA - Y data; DOL, appearance liking, flavor liking, texture liking variables (N=100)


23

Fig. 1.A.11. PLS-Regression of the consumer hedonic ratings onto the sensory attributes from the descriptive analysis — degree of liking of flavor vs. flavor attributes. - X data; DA – flavor (aroma + flavor + aftertaste) - Y data; flavor liking (N=100)

Fig. 1.A.12. PLS-Regression of the consumer hedonic ratings onto the sensory attributes from the descriptive analysis — degree of liking of texture vs. texture attributes. - X data; DA – texture and mouth feel - Y data; texture liking (N=100)


24

Fig. 1.A.13. PLS-Regression of the consumer hedonic ratings of the entire consumer population and of the two preference clusters (Group 1 and Group 2) onto the sensory attributes from the descriptive analysis — overall degree of liking vs. all sensory attributes. - X data; DA - Y data; DOL - total (N=100), GP 1 (N=86), GP2 (N=13)

Finally, we examined how the two preference clusters differed in their demographics and usage of table olives. Table 1.A.7 below highlights the few significant differences that were found between the two segments. Table 1.A.7. Exit survey results for all consumers, Group 1 (main cluster) and Group 2 (small cluster). Significant differences are highlighted in red. - Chi-square test; checked whether the response pattern of GP 1 and 2 are significantly

different - Significant Qs; Age, Origin, ‘organic product’ influence on food & beverages (P<0.05) <Demographic SAQs> Q. Age

TOTAL GP 1 GP 2 -29 29.0 29.1 23.1

30-39 9.0 5.8 30.8 40-49 12.0 11.6 15.4 50-59 30.0 32.6 15.4 60-69 13.0 12.8 15.4

70- 5.0 5.8 0.0


25

Q. Origin TOTAL GP 1 GP 2

CA 56.0 59.3 30.8 OTHER STATES IN US 29.0 24.4 61.5

ABROAD 11.0 11.6 7.7 Q. Ethnicity

TOTAL GP 1 GP 2 African 0.0 0.0 0.0 Asian 13.0 14.0 7.7

Caucasian 75.0 75.6 69.2 Hispanic/ Latino 6.0 4.7 15.4 Native American 0.0 0.0 0.0

Pacific islands 0.0 0.0 0.0 Mixed 4.0 3.5 7.7

Q. Gender

TOTAL GP 1 GP 2 MALE 39.0 41.9 23.1

FEMALE 59.0 55.8 76.9 Q. Self reported food neo-phobicity

TOTAL GP 1 GP 2 CONSERVATIVE (1) 2.0 2.3 0.0

2 2.0 2.3 0.0 NEITHER NOR 17.0 17.4 7.7

4 33.0 34.9 23.1 ADVENTUROUS (5) 43.0 39.5 69.2

Q. Education

TOTAL GP 1 GP 2 High school diploma/

GED 18.0 16.3 23.1 Bachelor's degree 39.0 39.5 38.5 Master's degree 18.0 16.3 30.8

PhD 14.0 15.1 7.7 Professional degree 6.0 7.0 0.0

Q. Family income TOTAL GP 1 GP 2

Under $50,000 20.0 19.8 15.4 $50,000~$100,000 31.0 29.1 46.2 more than $100,000 30.0 31.4 23.1

Not to report 16.0 16.3 15.4 <Table olive usage SAQs> Q. How often do you eat olives?

TOTAL GP 1 GP 2 2-3/ wk or more 24.0 23.3 23.1

1/wk 25.0 25.6 23.1 1/ 2wk 27.0 24.4 46.2

1/ month 14.0 15.1 7.7 less than 1/month 9.0 10.5 0.0

never 0.0 0.0 0.0 Q. Table olive in what food or beverages do you eat?


26

TOTAL GP 1 GP 2 canapé 20.0 18.6 23.1 pasta 60.0 58.1 76.9 pizza 78.0 75.6 92.3 salad 82.0 81.4 84.6

sandwich 37.0 40.7 15.4 cocktail 9.0 7.0 15.4

themselves 87.0 88.4 76.9 Q. From what source do you get your olives?

TOTAL GP 1 GP 2 can 89.4 87.8 91.7 deli 62.8 62.2 58.3

makes own 10.6 12.2 0.0 Q. How often do you buy olives?

TOTAL GP 1 GP 2 2-3/ wk or more 0.0 0.0 0.0

1/wk 10.6 9.8 8.3 1/ 2wk 17.0 17.1 16.7

1/ month 34.0 30.5 58.3 1/ 3 months 24.5 26.8 8.3

less than 1/ 3 months 11.7 13.4 0.0 never 2.1 1.2 8.3

Q. What type of olives do you buy?

TOTAL GP 1 GP 2 black whole 84.0 84.1 83.3 black sliced 44.7 41.5 58.3 green whole 73.4 69.5 91.7 green sliced 5.3 3.7 16.7

spiced black whole 39.4 40.2 25.0 spiced black sliced 5.3 6.1 0.0 spiced green whole 63.8 63.4 58.3 spiced green sliced 4.3 3.7 8.3 stuffed black whole 20.2 22.0 8.3 stuffed green whole 71.3 70.7 66.7

Q. How long do you store your olives?

TOTAL GP 1 GP 2 less than 1 wk 15.4 18.2 0.0 1wk ~ 1 month 46.2 42.4 57.1

1~3 months 20.5 18.2 28.6 3~6 months 17.9 18.2 14.3

more than 6 months 0.0 0.0 0.0 Q. What affects your olive purchase?

TOTAL GP 1 GP 2 type of olive 96.8 96.3 91.7

brand 31.9 31.7 25.0 country 25.5 25.6 25.0 package 17.0 18.3 8.3

price 79.8 80.5 66.7 nutrition 14.9 15.9 8.3

<Foods and Beverages in general>


27

Q. How much influenced by each factor? Total (N=100)

NOT AT ALL SOMEWHAT FAIRLY HIGHLY price 3.0 28.0 46.0 21.0

package 29.0 56.0 10.0 2.0 brand 22.0 51.0 19.0 5.0

nutrition 10.0 21.0 36.0 31.0 availability 9.0 27.0 35.0 27.0

organic 26.0 33.0 23.0 16.0 GP 1 (N=86)

NOT AT ALL SOMEWHAT FAIRLY HIGHLY price 2.3 25.6 47.7 22.1

package 31.4 51.2 11.6 2.3 brand 22.1 52.3 17.4 4.7

nutrition 9.3 20.9 38.4 29.1 availability 10.5 26.7 36.0 24.4

organic 25.6 36.0 24.4 11.6 GP 2 (N=13)

NOT AT ALL SOMEWHAT FAIRLY HIGHLY Price 7.7 46.2 30.8 15.4

Package 15.4 84.6 0.0 0.0 Brand 23.1 38.5 30.8 7.7

Nutrition 15.4 15.4 23.1 46.2 Availability 0.0 30.8 30.8 38.5

Organic 30.8 7.7 15.4 46.2

CONCLUSIONS of OBJECTIVE I.A

CANOPY CONTACT SHAKER HARVESTED OLIVES: 2008 SEASON

The main conclusion of this year’s research is that there was no difference in sensory quality and acceptability between the hand- and mechanically-harvested olives. There were, however, differences in sensory quality and acceptability between the fresh- and stored-processed olives, with the fresh-processed olives showing better sensory quality and receiving higher liking scores. We found some segmentation among consumers in terms of their preferences, but the two-cluster solution was one with one very large cluster (86% of the consumers) and another with a very small cluster. So we can confidently conclude that the consumer population is fairly homogeneous in terms of their table olive preferences.


28

Objective 1.B

Sensory Characteristics and Consumer Acceptance of

ENE Inc. Trunk Shaker Mechanically-Harvested California Black Ripe Olives 2009 Harvest Season

Soh Min Lee, Aurora Gomez Rico Rodrigueaz Barbero Louise Ferguson and Jean-Xavier Guinard

Fruit harvested by the ENE Inc. Terry II harvester in 2009 was either processed fresh, or

stored and processed in early 2010 by both Bell Carter and Musco Olive Company. The field procedures for the 2009 harvesting of the olives are detailed below. Location: Nickels Estate: Greenway Ave, Arbuckle CA Planted 7-8-01. Tree spacing = 12'x18' or 202 trees/ac ‘Manzanillo’ cultivar with Sevillano (S) pollinators; center row budded to Sevillano 07-03

May 2009: Trees were trained and chemically thinned.

30 September – 7 October 2009: Harvest trials were conducted with an ENE Inc. trunk shaking, ‘Terry II” harvester. Harvest Procedure: The trunk shaking harvester shook one replication in each of four rows. - catch frame was cleaned after each replication - fruit in bin was weighed in field using a bin scale - fruit on ground under tree was collected and weighed in the field using baby scale*

o held in extra bin for the entire row - fruit remaining on tree was hand harvested and weighed in the field with baby scale*

o held in extra bin for the entire row - mechanically harvested fruit in the bin was sent to Orland Musco grading station for weight

and COC grade and value - two 40 pound samples were drawn from each of the replications and each divided into two

20 lb samples and shipped out the same day o 2, 20 pound samples each were sent to Musco Olive Company and Bell Carter for

Fresh processing Processing after storage

o These samples were sent to Dr. Guinard for: Sensory panel evaluation after March 2010 Consumer panel evaluations after March 2010

2 (harvest methods) X 4 (8 tree replications) X 2 (processors) X 2 (processing methods) = 32

samples total Table 1.B.1 gives the effect of harvest method on the percent cannable fruit and adjusted value per ton. As can be seen in this table, mechanically harvesting with a trunk-shaking harvester did not significantly lower the percentage of cannable olives and the adjusted value per ton.


29

However, this was a very light crop year with a high percentage of large fruit and high values per ton. When the fruit was downgraded it was for over ripeness. Table 1.B.1. Average, and statistically analyzed, receiving station grades for the percentage of cannable fruit and adjusted value per ton for the hand and machine harvested olives.

Effect of Harvest Method on Olive Grade and Value: 2009

Harvest Method Percentage Cannablea Adjusted price/ton ($) a

Mechanical 95.6 1147.00

Hand 97.0 (NSD) 1178.60 (NSD) a Means separation within columns were performed with PROC TTEST procedure of SAS (SAS Institute Inc., Cary, NC); *, **, *** = 0.05, 0.01 and 0.001 level of significance

OBJECTIVES of OBJECTIVE I.B

The main objective of the study was to compare the sensory properties and acceptability of

hand-harvested and mechanically harvested table olives. Another objective was to examine the effects of storage before processing on the sensory quality of the olives. Also, two different commercial processors processed the experimental samples, and they were compared to commercial offerings from these processors.

PROCEDURES of OBJECTIVE I.B.

The study examined the sensory properties and acceptability of 10 California black table olive samples that were produced according to the experimental design described below.

The variables in the design were harvesting method—hand vs. mechanical; commercial processor—Musco vs. Bell-Carter; and processing method—olives processed fresh vs. olives processed after storage. We also added two commercial products to the design, one from each processor. The samples in the design and the two commercial products are shown in Table 1.B.1 below. Materials and Methods The study examined the sensory properties and consumer acceptability of 10 California black table olives that were produced according to the experimental design shown in Figure 1.B.1 below, and harvested either manually or with the trunk shaker method of mechanical harvesting. The labels used for the 10 samples throughout the report are shown in Table 1.B.2 Our two industrial partners in this project (Musco and Bell-Carter) received identical samples that were then processed with two different methods. Thus, 8 treatments, with the olive fruits being harvested either by hand or mechanically [i.e. 2 harvesting methods], then shipped to processor A and B [i.e. 2 processors], and then processed fresh or after being held in storage tanks [i.e. 2 processing methods], and 2 commercial products, one from each processor, were used in the study (Figure 1.B.1 and Table 1.B.2).


30

Figure 1.B.1. – Experimental design for olive production and harvest

Table 1.B.2. – Table olive samples Sample Abbreviations Processors Commercial Harvesting

methods Processing methods

A_Comm A Commercial - - A_Hand_F A - Hand Fresh olives A_Hand_S A - Hand Stored olives A_Mach_F A - Machine Fresh olives A_Mach_S A - Machine Stored olives B_Comm B Commercial - - B_Hand_F B - Hand Fresh olives B_Hand_S B - Hand Stored olives B_Mach_F B - Machine Fresh olives B_Mach_S B - Machine Stored olives

Descriptive analysis The sensory properties of the olives were measured by descriptive analysis with a trained panel of 8 judges (6 female, 2 male) all graduate students at UC Davis.

This year, the panel developed a scorecard with 30 attributes of appearance, flavor (taste and smell), texture, mouthfeel and after-taste (Table I.B.3). After the panel training, all the products were evaluated in triplicate, following a randomized complete block design. Olives were


31

presented at room temperature (20 °C), in a spherical glass covered with a plastic lid (3 whole olives and one sliced in half). The intensity of the attributes was rated on a category line scale labeled with “low” and “high” at the ends of the scale, except for the appearance attributes and the lasting flavor, which used the labels shown in Table 1.B.3 below.

Table 1.B.3. – Sensory attributes evaluated in the descriptive analysis

Attribute Reference Attribute Reference

SME

LL

(AR

OM

A)

Briny (aroma) Black olive brine

APP

EA

RA

NC

E

Glossy Matte --- Glossy

Ocean-like Green seaweed + olive brine

Surface roughness Smooth --- Rough (wrinkles, cracks)

Rancid Rancid oil* Size Small --- Large

Green/Grassy Parsley*, cut grass* Skin brownness Black --- Brown

Dried fruit Dried prune* Skin color unevenness

Even --- Uneven

Floral/Citrus Dried apricot* Brightness-gray (flesh)

Light gray --- Black

Fermented fruit Sauerkraut* Brightness-brown (flesh)

Light brown --- Dark brown

Earthy Potting soil*

TE

XT

UR

E/M

OU

TH

FEE

L

Firmness Sautéed mushroom

Sautéed mushroom* Crunchiness

Buttery Melted butter + olive brine*

Fibrous

Roasted Roasted almonds Juicy/Moist release

Painty Correction fluid* Oily, Silky

TA

STE

/FL

AV

OR

Overall flavor

AFT

ER

-T

AST

E

Lasting flavor Short --- Long Saltiness NaCl solution

Umami MSG solution Briny after-taste Black olive brine

Metallic Iron tablet solution *mixed with olives

Consumer testing 109 consumers also evaluated the olive samples during Picnic Day (UC Davis’ annual open campus event on April 17th, 2010) or during the spring of 2010 (between May 17th and June 4th, 2010). The participants were recruited among Picnic Day visitors, Davis Farmer’s Market customers, and members of Master food preservers of Sacramento County. The screening criteria for participation were to be black table olives users and likers and US Residents.

Each consumer was presented with 11 samples, with the first sample serving as a primer, for the purpose of eliminating the first-order effect typically encountered in consumer tests – the first sample receives a higher hedonic score than the subsequent samples in the serving order. The presentation order of the other 10 samples was randomized across consumers. Consumers were


32

instructed to rate overall degree of liking of each sample, followed by degree of liking of appearance, flavor, and texture on the 9-point hedonic scale, from 1=‘dislike extremely’ to 9=‘like extremely,’ and with 5=‘neither like nor dislike’. Two whole olives were served in plastic cups covered with lids at room temperature (20 °C). Crackers and water were provided for rinsing and palate cleansing. Upon completion of the tasting, consumers filled an exit survey with demographic, attitude and olive usage information. Data analysis The descriptive analysis data was analyzed using a combination of univariate and multivariate statistics. Analysis of variance (ANOVA) was used to examine the effect of each source of variations in the design. Principal component analysis (PCA) was then applied to the matrix of mean intensity ratings across the samples to visually summarize the similarities and differences among the products in the design.

The consumer hedonic ratings were also analyzed using a combination of univariate and multivariate statistics. ANOVA was first performed to observe the effect of each source of variation in the design. The matrix of hedonic ratings of samples across consumers was then analyzed by preference mapping – a combination of factor analysis and classification methods designed to assess preference market segmentation and drivers of liking identification for product optimization purposes. Partial least square (PLS) regression was performed to examine the relation between the hedonic ratings by consumers and the sensory attributes measured by the descriptive analysis panel.

RESULTS OF OBJECTIVE I.B. Results and discussion: Descriptive analysis There were significant differences between mechanically- and hand-harvested olives for several of the sensory attributes rated by the panel, as listed below (Table 1.B.4):

Aroma - rancid, fermented fruit, earthy, sautéed mushroom, buttery Appearance – surface roughness, skin brownness, brightness of the flesh (gray and brown) Flavor and Taste – overall flavor, saltiness, umami Texture – firmness, crunchiness, fibrous, oily/silky After-taste – lasting flavor, briny after-taste

This is in contrast to last year, when there were no significant differences between mechanically- (i.e. canopy contact harvester) and hand-harvested olives for any of the sensory attributes except surface roughness.

There were many significant differences across a range of appearance, flavor, texture, mouthfeel and aftertaste attributes between fresh-processed olives and olives stored before processing, and also, between processors and even more between the 2 commercial olives and the other samples.


33

Table 1.B.4. F-values for partitioned product source of variation.

Commercial

vs. non-commercial

Harvesting method

(hand vs. machine)

Processing method

(fresh vs. stored)


Harvesting method *

Processing method


Processor *

Processing method

Briny (aroma)

2.79 0.90 0.00 0.99 1.06 0.00 4.23

Ocean-like 4.53 0.91 0.01 0.31 2.72 1.82 2.93

Rancid 2.37 4.25 0.46 0.50 0.80 3.32 2.03

Green/Grassy 1.23 0.02 13.91 5.70 0.94 3.85 8.03

Dried fruit 0.07 1.47 5.82 3.10 3.06 0.40 0.05

Floral/Citrus 2.62 1.12 4.16 7.84 0.72 2.06 1.12

Fermented fruit

0.39 0.02 3.78 3.20 4.00 0.03 3.62

Earthy 1.49 13.84 33.02 16.12 12.84 26.88 3.35

Sautéed mushroom

4.34 3.83 5.96 2.51 4.06 4.06 0.39

Buttery 13.65 4.69 30.35 18.49 1.95 0.36 1.95

Roasted 6.48 1.03 0.95 1.44 3.87 1.35 0.24

Painty 0.56 1.27 0.02 6.83 2.77 0.26 1.05

Glossy 13.46 0.17 3.85 7.97 0.28 0.75 2.51

Surface roughness

11.04 25.81 31.42 60.00 4.55 10.59 40.50

Size 70.39 0.23 0.00 3.27 0.10 0.07 0.00

Skin brownness

5.93 31.01 1.44 36.90 14.54 10.97 10.56

Skin color unevenness

0.04 0.04 7.67 1.25 0.87 1.33 1.79

Brightness-gray (flesh)

4.43 23.74 0.03 10.37 14.38 2.91 20.98

Brightness-brown (flesh)

0.52 17.77 16.08 11.54 10.13 0.03 0.03


34

Overall flavor

29.40 2.64 0.08 0.35 0.28 4.28 1.10

Saltiness 24.71 15.75 3.25 3.94 0.45 0.16 0.04

Umami 41.43 0.11 2.41 26.38 2.44 17.34 5.22

Metallic 10.54 0.50 40.34 10.04 1.41 3.47 14.32

Firmness 32.27 6.24 42.73 4.67 4.34 1.30 0.21

Crunchiness 46.46 7.69 66.28 7.53 1.41 2.20 0.44

Fibrous 8.43 0.30 9.91 13.95 4.18 3.25 3.34

Juicy/Moist release

84.09 0.39 4.98 0.42 2.26 2.75 0.19

Oily, Silky 44.70 3.92 0.28 5.97 0.01 8.23 1.17

Lasting flavor

20.42 9.89 0.03 2.05 0.44 0.69 1.01

Briny after-taste

23.14 19.45 0.17 4.93 1.22 1.60 0.83

*Bold means significant source of variation (P<0.05)

In order to visually summarize the relationships among the sensory attributes and the products in a 2-dimensional ‘sensory map’ of the products, PCA was applied. The principal component (PC) biplot shows the main sensory features of each table olive sample – attributes located close to a given sample tend to be higher for that sample, whereas attributes, which are found away from that sample, tend to be lower. It also depicts the relationships among the sensory attributes – attributes, which are positively correlated, tend to form small angles with each other or to be clustered together on the plot, whereas attributes, which are negatively correlated, are found at opposite ends of the plot.

The biplot of PC1 vs. PC2 is shown in Figure 1.B.2 below. PC1 and PC2 explained 68.52% of the variance.


35

Figure 1.B.2. - Principal component analysis of the descriptive analysis data showing the sensory attributes (right) and the products (left). • The largest difference among products was observed between commercial vs. non-commercial products, primarily along PC 1 (39.6%). Commercial olives were characterized by higher crunchiness, firmness, fibrous texture and ocean-like flavor, whereas most of the non-commercial olives had a higher overall flavor, umami taste, saltiness, oily, juicy texture, lasting flavor, briny after-taste, and glossy appearance, and were bigger in size. • The next largest difference was observed between processing methods; fresh processed vs. stored processed, with interaction effects between harvesting method (hand vs. mechanically) and processor (A and B); primarily along PC 2 (28.9%). Olives processed after being held in storage tanks had a higher level of earthy, green, fermented fruit, metallic flavor, bright flesh color (gray and brown), and surface roughness (mainly the olive samples from Processor A), while fresh processed olives were characterized by a higher buttery, sautéed mushroom, crunchiness and firmness (mainly the olive samples harvested by hand). • The difference between harvesting methods was not large, but was higher than the previous year. Mechanically (trunk shaker) harvested olives were positioned on the negative direction of the PC 2 compared to hand harvested olives when the other sources of factors were identical (i.e. same processor and processing method). Mechanically harvested olives had relatively less firmness, crunchiness, buttery, and sautéed mushroom flavor, but stronger earthy, metallic, fermented fruit, green, surface roughness, and brighter flesh colors (gray and brown) compared to hand harvested olives. Consumer test There was a significant difference in consumer acceptance between mechanically- and hand-harvested olives. For consumer hedonic ratings (degree of liking), ‘processing method’ (fresh vs. stored) proved to be the biggest sources of variation, and the next most important source of variation was ‘harvesting method’ (hand vs. machine). This was true not only for


36

overall degree of liking but also for degree of liking of appearance, flavor and texture of the olives (Table 1.B.5 and Figure 1.B.3). Table 1.B.5. – F-values for partitioned product source of variation

Commercial vs. non-

commercial

Harvesting method

(hand vs. machine)

Processing method

(fresh vs. stored)


Harvesting method *

Processing method


Processor * Processing

method

Overall degree of liking 1.04 29.08 123.73 24.31 8.30 3.34 1.16

Appearance liking 4.14 7.28 45.83 1.02 9.77 4.40 2.21

Flavor liking 0.03 25.89 117.21 27.45 6.19 6.08 3.15

Texture/Mouthfeel liking 3.27 22.30 79.32 1.61 0.58 0.45 0.18

*Bold means significant source of variation (P<0.05).


37

Figure 1.B.3. – Phi-plot for F-values of partitioned product source of variation on (a) overall degree of liking, (b) appearance liking, (c) flavor liking, and (d) texture/mouthfeel liking

*Indicated source of variations on the phi-plot has significance on the dependent variables (P<0.05). An examination of the mean hedonic ratings confirms the observations above (Figure 1.B.4.a). There was a significant difference in liking between fresh-processed and stored-processed olives. On average, the four fresh-processed olive samples were liked best by the consumer population that tested the olives. Also, degree of liking measures for the mechanically-harvested olives were all slightly lower than those for hand-harvested olives (Figure 4a). Same result was observed for flavor liking and texture liking ratings, and to a lesser extent for appearance liking (Figure I.B.4.b). In conclusion, the main source of variation in the consumer acceptance data was ‘processing method’ followed by ‘harvesting method.’


38

123456789

1

3

5

7

9

Overall degree of liking

Appearance

(a)

(b) Figure 1.B.4.a and b. LS means for hedonic ratings of the 10 olive samples for (a) overall degree of liking and (b) including appearance liking, flavor liking, and texture liking (N=109 consumers). The overall degree of liking showed the highest correlation to flavor liking, and then to texture liking and appearance liking (Table 1.B.6, Figure 1.B.4.b), similar to what we observed last year. This suggests that flavor characteristics are the most important determinants of consume liking for table olives. Table 1.B.6. Pearson’s correlation coefficients among hedonic ratings by consumers

Variables Overall degree of liking


Overall degree of liking 1 0.916 0.990 0.900

*Values in bold are significantly different from 0 with a significance level alpha=0.05


39

Preference mapping and consumer segmentation Figure 1.B.5 shows the results of the internal preference mapping analysis as bi-plot of the first two principal components; showing the main direction (as vectors) of each individual consumer’s preferences for the 10 olives tested (i.e. each dot represents each individual consumer’s main preference direction). The preference map shows that most of the consumers are located on the right side of the plot, where the fresh-processed samples are located, while the stored-processed samples are on the opposite side of the plot. This observation is comparable to the previous year’s result (i.e. fresh-processed olives were liked the most). However, unlike last year, the two “hand-harvested” olives among the four fresh-processed samples are located at the very far end of plot where most of the consumers are located. Also among the four stored-processed olives, the two “mechanically-harvested” olives are located away from the consumers. This implies that harvesting method had an influence on consumers’ acceptance of the olives, which mechanically harvested olives not faring as well as hand-harvested olives. This confirms the conclusions stated above (Table 1.B.5, Figures 1.B.3 and 1.B.4). In addition, it is worth noting that the commercial olives were not located away from the consumers, which was the case in last year’s research. Figure 1.B.5. Internal preference map generated based on overall degree of liking, showing individual consumers and the olive samples (N=109) Consumer preference for Californian-style black olives was shown to be quite homogeneous, which is the same conclusion as the one we drew for last year’s research. In order to check possible market segmentation, cluster analysis (Pearson’s dissimilarity proximity matrix; Average-linkage agglomerative method) was performed and the resulting diagrams are shown in Figures 1.B.6, 1.B.7 and 1.B.8 below. The small group consists of consumers who preferred stored-processed olives (Figure 1.B.8).


40

84 20 78 66 98 85 99 15 71 69 54 94 35 22 27 43 100

106 7 88 14 39 64 47 93 65 1 77 57 89 6 45 68 52 96 8 104 51 103 55 34 109 49 29 44 76 92 23 46 73 12 105 17 53 63 32 25 42 9 80 97 30 82 3 67 108 21 38 58 86 2 61 95 59 37 60 19 107 70 4 74 10 62 81 28 91 13 90 50 72 5 79 18 11 56 83 101 40 41 16 26 31 36 75 33 87 48 102

0

0.2

0.4

0.6

0.8

1

1.2

Diss

imila

rity

-1.0

-0.5

0.0

0.5

1.0

-1.0 -0.5 0.0 0.5 1.0

C1

C2ACOMM

AHANDF

AHANDSAMACHF

AMACHS

BCOMM

BHANDFBHANDS

BMACHF

BMACHS

-15

-10

-5

0

5

10

15

-25 -20 -15 -10 -5 0 5 10 15

F2 (1

2.80

%)

F1 (30.46 %)

Observations (axes F1 and F2: 43.26 %)

Figure 1.B.6 – Cluster analysis dendrogram of the 108 consumers (1 excluded; model could not be generated)

Figure 1.B.7. Internal preference mapping with segmentation (C1=101, C2=7; 1 excluded- model is unable to be generated)


41

1

3

5

7

9

C1 (N=101)

C2 (N=7)

Figure 1.B.8. Mean overall degree of liking scores of the 10 samples for each clusters (C1=101, C2=7; 1 excluded) Identification of drivers of liking

PLS-regression was performed in order to examine sensory drivers of consumer liking for black table olives (Figures 1.B.9 and 1.B.10). This analysis shows which sensory attributes are associated with overall degree of liking by consumers, and more specifically, which appearance attributes are associated with liking for appearance of the olives (Figure 1.B.10.b), which flavor attributes are associated with liking for flavor of the olives (Figure 1.B.10.c) and which texture attributes are associated with liking for texture of the olives (Figure 1.B.11.d).

Figure. 1.B.9. PLS2-Regression of the consumer hedonic ratings (overall liking, appearance liking, flavor liking and texture liking) onto the sensory attributes from the descriptive analysis


42

a)

b)


43

c)

d)

Figure 1.B.10.a, b, c, d. PLS1-Regression of each consumer hedonic ratings (i.e. a) overall liking, b) appearance liking, c) flavor liking and d) texture liking) onto the sensory attributes from the descriptive analysis.


44

CONCLUSIONS of OBJECTIVE I.B ENE. INC. TRUNK SHAKER HARVESTED OLIVES: 2009 SEASON

The main conclusion of the 2009 research is that there was some difference in sensory quality and consumer liking between the hand and trunk shaker mechanically-harvested olives. In comparison to the research conducted in 2008-2009, when we found no significant differences between hand and canopy-contact - harvested olives in both sensory properties and consumer acceptance, there were some noticeable differences between hand- and mechanically-harvested olives fin 2009-2010. However, we were again able to confirm that fresh-processed olives have strong potential in the Californian-olive market.

OBJECTIVE II

Evaluation of Mechanical Harvester(s) Efficiency and Effects on Fruit Quality and Value A. ENE Inc. “Terry I” and “Terry II” self propelled trunk shakers with pistachio catch frames B. Noli trunk shaker only, without catch frame C. Agright canopy contact harvester; self-propelled with catch frame

INTRODUCTION of OBJECTIVE II

The two picking technologies currently available for evaluation are the canopy contact heads

and trunk shakers. Trunk shaking harvesting of oil olive trees is common in Europe. In the 1960s, the University of California also developed pruning methods, and an ‘inertia head” shaker for mature California table olive trees. However, the technology, never widely adapted, was designed exclusively for larger trees, not younger hedgerow orchards. Because a younger hedgerow orchard, developed by Krueger and Ferguson, (discussed under Objective III.B of this report), now exists, there is an opportunity to examine both a canopy contact head and trunk shaking harvesters in young hedgerow table olives.

For trunk shaker harvester evaluations in 2009 the companies and machines selected were two versions of the ENE Inc. trunk shaker, “Terry I” and “Terry II” (Fig. 2.1), named for the hydraulic engineer who developed the shaker heads, and a Noli Spanish trunk shaker. The ENE Inc. machines had a double-sided modified pistachio harvester catch frame with a bin take out system. The Noli head was front mounted on an almond shaker cab and therefore had no catch frame or take out system. As a result the trials with the Noli were more limited.


45

Fig. 2.1. ENE Inc. Terry II modified pistachio trunk shaking harvester with catch frame harvesting at Nickels Soils Laboratory on 5 October 2009. Mr. Terry Tompkins is operating the harvester.

The canopy contact machine evaluated was a modified Agright over the row pomegranate

harvester, the Olivia (Fig. 2.2). Due to harvester size limitations only a limited, non-replicated trial was conduced on the Agright. The Agright Olivia harvester in the fully lowered position has a top closure height of 1.6 feet, a bottom picking rod height of 3 feet, a top picking rod height of 8.8 feet and a top closure height of 1.6 feet. The Olivia has a clear passage from the ground of 9.3 feet, a left and right clear passage width of 4.75 feet and 2.6 feet of rod penetration into the tree canopy from both sides. The pruning required to make the trees suitable for the Olivia decreased the yield significantly relative to the trees in which the trunk shaking harvesters were evaluated. Therefore, as both the number of replications, and the canopy of the trees harvested by Olivia were different from those harvested by trunk shakers, the final removal efficiencies of the three trunk shaking harvesters can be roughly, but not statistically, compared to the efficiency of the Olivia.

The objective was to evaluate efficiency, and effects on fruit quality of these four machines. As the analyzed data provided below demonstrates, trunk shaking of young hedgerow orchards has potential, but is currently still below the 80% necessary removal efficiency. Similar, but not strictly statistically comparable analyzed results, were obtained for the Agright Olivia.


46

Fig. 2.2. Agright Olivia canopy contact head harvester operating in Nickels Soils Laboratory hedgerow olive orchard on 30 September 2009. Mr. Richard Loquaci of Madera Ag Services is operating the harvester.

PROCEDURES of OBJECTIVE II Location: Nickels Estate: Greenway Ave, Arbuckle CA Planted 7-8-01. Tree spacing = 12'x18' or 202 trees/ac ‘Manzanillo’ cultivar with Sevillano (S) pollinators; center row budded to Sevillano 07-03

May 2009: Trees were trained and chemically thinned.

30 September – 7 October 2009: Harvest trials were conducted with two ENE Inc. trunk shakers, “Terry I” and ‘Terry II” , a Noli Spanish trunk shaker, and a modified pomegranate harvester, Olivia.

Harvest Procedure: Each trunk shaking harvester shook one replication in eight rows. - catch frame was cleaned - fruit in bin was weighed in field using a bin scale* - fruit on ground under tree was collected and weighed in the field using baby scale*

o held in extra bin for the entire row - fruit remaining on tree was hand harvested and weighed in the field with baby scale*

o held in extra bin for the entire row


47

- mechanically harvested fruit in the bin was sent to Orland Musco grading station for weight and COC grade and value

- two 40 pound samples were drawn from each of the replications and each divided into two 20 lb samples

o two 20 pound samples each were sent to Musco Olive Company and Bell Carter for Fresh processing Processing after storage

o These samples will be sent to Dr. Guinard for: Sensory panel evaluation after March 2010 Consumer panel evaluations after March 2010

* These three field weights, and their confirming weights at the receiving station, were used to calculate the calculate harvester removal efficiency and final harvester efficiency as follows: Fruit Removal Efficiency = (fruit in harvest bin + fruit on ground) X

(Fruit in bin + fruit on ground + fruit remaining in tree)

Final Harvest Efficiency = (fruit in harvest bin) X (Fruit in bin + fruit on ground + fruit on tree)

Data Analysis

A method of analysis sometimes referred to as “box and whisker plots” was used to analyze this data (see Fig. 2.3). It is particularly applicable to data in which one wants to see the range of values within treatments.

Fig. 2.3. “Box and whisker plots”. When describing a set of data, without listing all of the values, we can use measures of

location such as the mean and median. It is also possible to get a sense of the data's distribution by examining the (1) minimum, (2) maximum, (3) median (or second quartile), (4) the first quartile, and (5) the third quartile (Fig. 2.3). Such information will show the extent to which the data is located near the median or near the extremes.

http://regentsprep.org/REgents/math/ALGEBRA/AD6/..AD2/measure.htm�


48

The first quartile is the middle (the median) of the lower half of the data (Fig. 2.3). One-fourth of the data lies below the first quartile and three-fourths lies above (i.e., The 25th percentile). The third quartile is the middle (the median) of the upper half of the data. Three-fourths of the data lies below the third quartile and one-fourth lies above (i.e., The 75th

percentile). A quartile is a number; it is not a range of values. A value can be described as "above" or "below" the first quartile, but a value is never "in" the first quartile. A ‘box plot’ or ‘box and whisker plot’ describes (graphically) all these summary statistics, giving a graphic of the distribution of the data. The first and third quartiles are at the ends of the box, the median is indicated with a vertical line in the interior of the box, and the maximum and minimum are at the ends of the whiskers.

All data were analyzed by standard statistical analysis using Statistical Analysis Systems software (SAS Institute, Cary, NC) to perform the means separation (Duncan’s multiple range test and LS means; 5% level of significance) and analysis of variance (PROC GLM); distributions analyzed by PROC UNIVARIATE AND PROC BOXPLOT.

RESULTS of OBJECTIVE II Final Trunk Shaking Harvester Efficiencies and Effect of Harvester on Fruit Quality &Value

Fig. 2.4. Analyzed final harvest efficiencies of the three trunk shaking harvesters (Noli, Terry I, Terry II) relative to the hand harvested control, HHC.

The analyzed final harvest efficiencies of the three trunk shaking harvesters relative to the

hand-harvested control, HHC, are shown in Fig. 2.4. The HHC trees were considered to have had 100% of the fruit removed in all four replicates, and therefore the data display no variability for efficiency. There was no significant difference in the final harvest efficiency between the Noli trunk shaker (had no catch frame) and the Terry II, but both had significantly higher final harvest efficiencies than the Terry I. All were significantly less efficient than the hand harvest.

As discussed earlier, the results with the Noli trunk shaker cannot be strictly compared to the Terry I and II because the lack of a catch frame limited the number of replications that could be


49

done. As a result the data from the Noli had the widest distribution of efficiency among the harvest technologies, as results were highly variable from tree-to-tree.

Only the data from the ENE Inc. Terry I and II can be considered to be statistically comparable to the hand harvest control in that sufficient replications were done to produce comparable data. This data demonstrated that the ENE Inc. Terry II was significantly more efficient than the Terry I. Thus far, the ENE Inc. Terry II is the best performing trunk shaker tested. However, it averaged only 64% efficiency, 16% less than the 80% needed.

Loss from the catch frame was minimal; less than a pound per tree and easily equal to that of hand harvest. As a result the harvester removal force and final harvester efficiency are essentially the same.

The previous problem with trunk shaking trees (i.e, trunk barking) was largely eliminated by decreasing shaker head clamping strength below 1000 psi. However, in the future it would be advisable to develop a method of measuring bark strength as a function of irrigation status to ensure barking is totally eliminated. There was little branch damage caused by the trunk shakers. The shakers also sometimes disturbed the soil around the trunk. For this reason all trunk harvesting was followed by a field capacity irrigation.

In summary, the major problem with trunk shakers is the final harvester efficiency of 64%, is not the necessary 80% final removal efficiency our economic analyses demonstrate is needed. In our observations the shakers were more efficient at removing fruit closer to the trunk. Large clumps of olives at the ends of longer branches in the top quarter of the canopy were resistant to trunk shaker harvesting. A possible solution might be to prune these branches off as they are observed during harvest. It also suggests that chemical thinning; both to eliminate these large clumps and increase fruit size as heavier olives harvest more easily, would enhance trunk shaking. Thus, the solution to increasing the efficiency of trunk shakers appears to be to change the tree as well as improve the harvester and harvesting parameters.

Table 2.1 gives the final harvest efficiencies as well as the canning percentages and adjusted price per ton. The data clearly shows there were no significant differences in either canning percentage or adjusted price per ton among the mechanically harvested olives, or between them and the hand harvested olives. Therefore trunk shaking can produce olives with quality and value equal to that of hand harvested olives.

These olives were delivered to Musco and Bell Carter immediately after harvesting for both fresh and stored processing. Lee and Guinard will evaluate the sensory characteristics and consumer acceptability in Spring 2010. Even though these olives were harvested in 2009, versus 2008 for the olives evaluated by Lee and Guinard in Objective I of this report, they have virtually identical canning percentages and adjusted values per ton. Therefore it is expected, that as with olives harvested with the canopy contact head harvester in 2008, these olives, when processed, will be virtually indistinguishable from hand harvested olives. The sensory and consumer evaluations will be available by July 2010.

Final Harvester Efficiency of Agright Olivia Canopy Contact Harvester and Harvester Effects on Fruit Quality & Value The Agright Olivia, an over the row modified pomegranate harvester, was not evaluated in a randomized replicated trial. However, in two 18-tree rows it produced an average of 67% final harvest efficiency with a canning percentage of 93.5% and an average value per ton of $1071.96 per ton. The olives left in the tree appeared to be primarily closer to the trunk; unlike the better interior fruit removal of the trunk shakers. The harvester damaged 23% of the trees including


50

Table 2.1. Final Trunk Shaking Harvester Efficiency and Harvester Effects on Olive Quality and Value

Harvest method Harvest efficiency %Cannable fruit Adjusted value per ton

($)

HHC 100.0 a X 97.0 a 1178.6 a

Noli 69.5 b 94.1 a 1146.9 a

Terry I 41.2 c 94.8 a 1147.3 a

Terry II 64.5 b 95.6 a 1147.0 a

Significance *** NS NS

X Means separation within columns by Duncan’s multiple range test; P = 5%. Where letters are different within columns, means are significantly different; level of significance by ANOVA ***, NS = significant at 0.1% and non-significant, respectively.

pulling one tree from the ground. Had the trees had been properly pruned the harvester would have probably achieved the very high efficiencies observed with the MaqTec Colossus in Argentina in 2007, and inflicted little damage. The Agright Olivia harvester in the fully lowered position has a top closure height of 1.6 feet, a bottom picking rod height of 3 feet, a top picking rod height of 8.8 feet and a top closure height of 1.6 feet. The Olivia has a clear passage from the ground of 9.3 feet, a left and right clear passage width of 4.75 feet and 2.6 feet of rod penetration into the tree canopy from both sides. Successful harvesting with the Olivia would require a tree with no stiff wood above 10 feet high, beyond 4 feet wide and below 4 feet.

As with the olives harvested by both the earlier tested canopy contact harvester and the tree trunk shaking harvesters evaluated this year, the harvested olive’s receiving station grade and value demonstrate these olives would also produce processed olives that are indistinguishable from hand harvested olives.

Finally, as with the trunk shaking harvesters evaluated in 2009, it appears the way to improve the final harvesting efficiency of the Olivia is to train young trees into a high-density hedgerow. The Olivia is unsuitable for California’s existing olive orchards; the trees are too large.

CONCLUSIONS of OBJECTIVE II

The three trunk shaking harvesters and one canopy contact harvester evaluated this year all produced fruit that, based upon canning percentage and adjusted price per ton, would produce processed olives that are indistinguishable from hand harvested olives. What must be improved is the fruit removal percentage. The data thus far strongly indicates this can be done through both machine improvements and tree training and pruning.


51

OBJECTIVES III.A and III.B

Evaluation of Pruning and Training Methods for Mechanical Harvesting: 2009 Season OBJECTIVE IIIA. Evaluation of Mechanical Topping and Hedging: 2008 and 2009 Seasons

INTRODUCTION of OBJECTIVE III.A

Preliminary work done in 2006 demonstrated fruit on the canopy facing the row middles were mechanically harvested with the canopy contact harvester significantly more efficiently than fruit on the canopy surface between the trees. This suggested that a typical orchard topped and side hedged to a hedgerow configuration could be harvested more efficiently with a picking head harvester. However, the effect of long term hedging and topping, at least 6 years, on yield and fruit quality has not been demonstrated. Therefore, this trial tests the hypothesis that a moderate annual mechanical pruning program does not decrease yield or fruit quality over a 6 year period. The results presented here are for the first two seasons.

PROCEDURES of OBJECTIVE III.A

Location: Block 17W: Rocky Hill Ranch, Exeter, CA - planted 1998 - 13 rows, 83 trees per row, ‘Manzanillo’ olives with ‘Sevillano’ pollinators - Spaced @ 12 X 26 feet, 139 trees per acre 12 June 2009 Six rows of 83 trees each were conventionally pruned

• Trees were skirted 3 feet from the ground Six rows of 83 trees each were pruned for mechanical harvesting:

• All large, horizontal limbs extending into the row middle were pruned off • Trees were skirted at 3 feet from the ground • Trees were topped at 12 feet • Trees were hedged on the east side 6 feet from the trunk •

11-12 September 2009 All 12 tree rows were completely hand harvested

• yield was weighed on a tared field bin scale • bin was reweighed at Musco receiving station • COC sample grade pulled from each hedging treatment bin • yield per acre, % cannable, % of sizes and adjusted price/ton analyzed • samples were not sent for processing as this trial was to evaluate mechanical pruning

effects on yield and fruit size


52

RESULTS of OBJECTIVE III.A

As can be seen in Tables 3.1 & 3.2, and Fig. 3.1 below, the second year of moderate mechanical topping and hedging in June 2009 significantly decreased yield, but did not significantly affect the % cannable fruit or adjusted price per ton. However, due to poor fruit set in 2009, the total tree yields were so low the data is unreliable. Trees that were mechanically hedged and topped in 2009 produced 147.3 pounds per acre, significantly less than the 351.3 pounds per acre for hand pruned trees. In 2008, these same pruning treatments produced 2682.7 and 3079.1 pounds per acre, respectively, and were statistically not significantly different. The two-year totals thus far are 2830 and 3423.2 cumulative pounds per acre for mechanically and hand pruned trees, respectively. Therefore, over two years the mechanically pruned trees have produced 593.2 pounds per acre less than hand pruned trees, or 296.2 pounds less annually.

At least 6-8 years of data are required, particularly in an alternate bearing species that produces on one year old wood, to draw conclusions about the effects of a long term pruning program on growth and annual yield. The abnormally low fruit set 2009 results also makes the data less reliable. Therefore, no conclusions can be drawn from this data thus far.

As in 2008 the canning percentages and adjusted values per ton for mechanically pruned and hand-pruned trees were, as expected, not significantly different. Table 3.1. This T-test analysis demonstrates the significant difference in yield produced by the mechanical topping and hedging, and its lack of effect on olive grade or quality. T-Test Analysis of Effect of Mechanical Pruning on Olive Yield, Grade and Value

T-Tests Method Variances DF t value Pr > |t| Significance

Yield (lb) Pooled Equal 10 3.78 0.0036 ** Satterthwaite Unequal 8.38 3.78 0.0050

%Cannable fruit Pooled Equal 10 -1.81 0.1012 NS Satterthwaite Unequal 9.35 -1.81 0.1033

Adjusted value/ton ($) Pooled Equal 10 -1.14 0.2791 NS Satterthwaite Unequal 9.57 -1.14 0.2803 Means separation by Student’s t-test; P = 5%. Significance **, NS = 1%, non-significant, respectively. Variance tests for equal and unequal variances give the same result. Table 3.2. This table demonstrated the effect of two sequential years of topping and hedging on yields per acre, fruit quality and value.

Effect of Topping and Hedging on Yield and Olive Grade and Value: 2008 and 2009 Pruning treatment cumulative 2008 * 2009* Cumulative yield

Hedge + top Yield (lb) 2682.7 147.3 a

2830.00 % Cannable fruit 95.9 96.1 NS Adjusted value per ton ($) 1131.50 1174.3 NS

Hand pruned Yield (lb 3071.0 351.3 b

3423.20 % Cannable fruit 97.4 94.2 NS Adjusted value per ton ($) 1151.40 1157.5 NS

NS NS * Means separation within columns by Duncan’s multiple range test; P = 5%. Where letters are different within columns, means are significantly different; level of significance by ANOVA*, NS = significant at 1% and non-significant, respectively


53

Effect of Mechanical Topping and Hedging on Yield: 2009

Fig. 3.1. This graph demonstrates that mechanical topping and hedging in 2009 significantly decreased yield.

Fig. 3.2. Mechanically topped and hedged treatment in Rocky Hill Ranch, Exeter, Tulare County, CA on 6 June 2009. The Manzanillo tree on the left was hedged 6 feet from the trunk, topped 12 feet from the ground, and skirted 3 feet from the ground. A hand-pruned tree is on the right. This is the second of six planned years of mechanical pruning. Thus far, there are no significant differences in cumulative yield. However, the yields have averaged less than one ton per acre due to poor fruit set in 2008 and 2009.

CONCLUSION of OBJECTIVE III.A

Unlike the 2008 season, mechanical topping at 12 feet and hedging 6 feet from the trunk on one side of the tree in June 2009 significantly decreased yield. Fruit quality and value were unaffected by mechanical pruning both years. However, the cumulative and average yields for


54

2008 and 2009 are so low, less than a ton per acre average over the two years that this data can be considered unreliable. This work will be continued for another 4-6 years to determine the effects of mechanical pruning on normal yields. Then, we hope to demonstrate that the decreased cost of a more efficient harvest will compensate for the yield losses produced by mechanical topping hedging. III.B. Evaluation of Pruning and Training Methods on Tree Yield and Fruit Quality B. Establishing and Training Manzanillo Table Olives for Mechanical Harvest: 2009 Season W.H. Krueger, Louise Ferguson, Stan Cutter, Dorothy LaCaroix, Charles Garcia, and Antonio Isern

Table olives in California are hand harvested. The cost of hand harvest can be as much as 50 percent of the gross. From 1997 to 2000, the California Olive Committee (COC), the table olive marketing order, sponsored the development of a mechanical harvester for table olives. Prototype canopy shaker machines were developed. Although these machines looked promising, they had two major drawbacks: 1) Efficiency of harvest - when the picking head came into close proximity of the fruit, it was removed. However, leading and trailing canopy edges and inside fruit proved to be problematic because it was difficult to get the head close to fruit located in these positions. Fruit removal was often disappointing. 2) Fruit damage - The fruit can be damaged in the removal process. While this damage may appear similar to what may occur with hand harvest, the bruises are generally deeper and more severe. One of the major table olive processors quit accepting mechanically harvested fruit due to concerns related to fruit damage. This temporarily stopped progress toward mechanical harvest with this machinery. A continued and increasing need for mechanical harvest has rekindled interest. The COC resumed funding for mechanical harvest research in 2006 and is continuing to support this research. The focus of the research has been on improvement of the previously developed machinery to increase removal and reduce damage, the development of loosening agents to facilitate mechanical harvest and other types of mechanical harvesters such as trunk shakers.

If a tree canopy could be developed in which all of the fruit was accessible to the picking head, a much improved harvest efficiency with reduced force and, therefore, reduced fruit damage should be attainable. The ideal tree and orchard configuration would appear to be a close spaced hedgerow system which would present a flat narrow fruiting wall to the harvester with no leading, trailing edge or inside fruit. A fruiting canopy approximately 6 feet in width and approximately 12 to 15 feet high would be appear to be ideal for maximum machine efficiency. With a narrow tree canopy and tree height such as this, narrower row spacing will be necessary to achieve maximum yields. This type of tree architecture should also be more adaptable to other types of mechanical harvesters including existing trunk type shakers and other types of machinery which could be developed.


55

OBJECTIVES of OBJECTIVE III.B

1) Develop a narrow canopy hedgerow to facilitate mechanical harvest; 2) Evaluate and demonstrate the feasibility of a high density hedgerow developed specifically

for mechanical harvest; and 3) Compare different training methods for developing a narrow canopy hedgerow.

PROCEDURES of OBJECTIVE III.B

In the spring of 2000, Manzanillo variety table olives were planted on 2 acres at the Nickel’s Estate in Arbuckle with a north-south row orientation and a tree spacing of 12 feet in the row and 18 feet between rows (202 trees per acre) (Fig. 3.3). The selected training treatments included “conventional” and three narrow canopy hedgerow treatments. The conventional training consists of thinning out fruit wood and opening up the center of the tree. The trees will eventually have 3 to 5 primary scaffolds. With the narrow canopy hedgerow treatments, permanent limbs are being trained parallel to the row in a narrow plane (approximately 1 foot wide) with flexible temporary fruiting wood extending approximately 3 feet out into the row on either side. Large stiff limbs extending into the tree row are positioned into the permanent limb plane or are removed. The narrow canopy hedgerow treatments are: 1) Free Standing - where pruning alone is used to conform the trees to the system; 2) trellised woven - where potentially permanent limbs are woven between three wires spaced at 4, 7 and 10 feet; and 3) trellised tied - where potentially permanent limbs are tied to the wires. In 2007, the tied treatment was not pruned because cropping potential appeared light. In 2008 and 2009 this treatment was pruned, but not tied. The pruning consisted of thinning the tree canopy. This treatment will gradually be brought back to the narrow canopy system through a combination of pruning and tying. The treatments are arranged in a randomized complete block design and consist of blocks of three rows of either seven or eight trees. There are four replications of each treatment. The center row of each three row plot has been harvested by hand, yield weighed and 10 to 12 lb. samples submitted to Musco Family Olives for commercial grading. The sample results were used to assign a value to the production.

Originally 6 trees of the Sevillano variety were strategically placed in the planting to provide for cross pollination for the partially self incompatible Manzanillo. Due to disappointing growth of these trees, cross pollination was inadequate. Even though there was a good bloom, the fruit set for 2003 (third year) was disappointing and did not warrant harvest. During the summer of 2003, the center row of the planting was top worked to Sevillano to provide for adequate cross pollination. During bloom in the spring of 2004 and 2005, the block was artificially cross-pollinated using Sevillano pollen. The grafted pollinators developed well and artificial pollinization was discontinued in 2006. In the spring of 2007, about two weeks after full bloom, all of the plots were chemically thinned with Napthalene Acetic Acid (NAA). In 2008 bloom appeared lighter and less uniform than in 2007 so no chemical thinning was done. In 2009, although the bloom looked good, fruit set was relatively light and variable so no thinning was done.


56

Fig. 3.3. Nine-year-old olive hedgerow training trial at Nickels Soils Laboratory planted in 2002. Trees are spaced at 12X18 with 202 trees to the acre. The four training treatments are conventional training, a freestanding espalier, a woven trellised espalier and a tied trellised espalier. Thus far there are no significant differences in yield or fruit quality among the training treatments.

RESULTS of OBJECTIVE III.B

Yields in 2009 were extremely variable ranging from 0.87 to 7.32 tons per acre for individual

plots. There were no statistically significant differences between any of the treatments for yield, value per ton or value per acre (Table 3.3). The average yield for all treatments was 3.68 tons per acre. Cumulative yields for all 6 years of production are similar for all treatments with no statistically significant differences.

Table. 3.3. Nickel’s Hedgerow olive harvest, 2004-09.

Treatmeant2004

(4th yr.)2005

(5th yr.)2006

(6th yr.)2007

(7th yr.)2008

(8th yr.)2009

(9th yr.)Cum. Yield

(2004-09) Tons/A Tons/A Tons/A Tons/A Tons/A Tons/A $/Ton $/Acre Tons/A

Conventional 4.09 1.75 2.81 6.39 5.96 3.35 1193 3991 24.35Free Standing 3.66 1.51 2.26 6.4 5.04 4.37 1189 5192 23.24Trellised, Woven 4.21 1.68 2.28 6.07 5.88 2.29 1192 2731 22.41Trellised, Tied 3.58 3.45 1.76 7.51 4.52 4.42 1179 5178 25.24

NS* NS NS NS NS NS NS NS NS

*NS (Not Significant at the 5% level using Ficher's Test)


57

DISCUSSION of OBJECTIVE III.B

As has been the case in other years, variability from plot to plot resulted in no significant

differences between treatments despite some rather large numerical differences between treatments. The overall production, though the lowest of the last 3 years, would have to be considered good for this year compared to very low statewide production. Cumulative yields for all treatments are very similar through the first 9 years. Over the years we have often observed that the highest or lowest yielding treatments will trade places the following year due, mostly likely, to the alternate bearing nature of the olive. These results indicate that these trees can be maintained in the narrow canopy configuration with no loss in production or value compared to conventional trees.

A comprehensive project aimed at developing mechanical harvest for table olives is currently underway. This project is being headed by Dr. Louise Ferguson, UCCE Olive Specialist, and includes collaboration with a University of Florida researcher, UC Davis Department of Agricultural Engineering and Plant Sciences, UCCE Farm Advisors, California State University researchers at Chico, farmers and equipment manufacturers and mechanical harvesters and is supported by the COC. Research is being conducted in the southern producing region (San Joaquin Valley) and the northern producing region (Sacramento Valley). The planting at Nickels is playing a central role in this effort. In 2009 we evaluated two trunk shakers and one canopy shaker at the Nickels planting. Our trials were advertised to our clientele and were open to the public. About 40 interested people came to observe the harvest. This year we made significant progress on reducing trunk damage from the trunk shakers. Removal efficiency still needs to be improved. We were also able to test a Wheel Rake harvester which is being developed in collaboration with CSU Chico specifically for narrow canopy hedgerow systems. A complete report and more information on this effort can be fount at http://groups.ucanr.org/olive_harvest/

Even though this planting was planted and developed with a canopy shaker type of harvester in mind, this was the first year we have been able test this type of machine on these trees. This experience along with our experience with the other harvesters has shown the need for more specific pruning practices to improve harvest efficacy of the machines. In the coming years we will be testing pruning practices designed to improve harvest efficiency and collecting data on the effect of these practices on yields and value.

The results from this trial have been extended through grower meetings, newsletters, and an International Horticultural Society meeting. There are currently an estimated 500 acres of table olives following this system planted or scheduled to be planted in 2010 in Glenn and Tehama Counties. It is expected that this acreage will increase dramatically if mechanical harvest becomes practical. This would result in the revitalization of the California table olive industry.

http://groups.ucanr.org/olive_harvest/�


58

OBJECTIVE IV

Screening Abscission Compounds for Black Ripe Table Olives: 2009 Season

INTRODUCTION of OBJECTIVE IV

Due to extremely poor fruit set at Lindcove Field Station olive abscission compound screening trials were conducted at only Nickels Soils Laboratory, Arbuckle. Trials were conducted from 9 – 24 September 2009. The objective of this trial was to continue olive screening with compounds from the Florida citrus fruit abscission agent library. The long-term goal of this project is to adapt table olives to mechanical harvesting. Identification of a suitable abscission agent is viewed as a key to industry adoption of mechanical harvesting, because mechanical harvesting could be performed less aggressively and fruit damage could be minimized.

PROCEDURES of OBJECTIVE IV

Nickels Soils Laboratory trial. A trial was initiated on 9 September 2009 in a block of olive trees located in Nickels Soils Laboratory on Greenbay Ave in Colusa County, CA. Five uniform ‘Manzanillo’ trees with good fruit load were selected, and one replicate branch on each tree was tagged for each treatment. Thus, treatments were replicated five times. Each branch contained at least eight fruit and 25 leaves. Fruit number was recorded. All treatments were randomly assigned to the branches on each tree. Abscission compounds were dissolved in water containing 0.05% Activator-90 and applied between 9:00 AM and 2:30 PM with a hand-held 1.5L pressurized sprayer until run-off. A water control containing adjuvant was included in all trials. Maximum, minimum, and average temperatures on the day of application were 36, 19, and 28ºC, respectively. Fruit detachment force (FRF) in grams-force was measured 10 days after application using an Imada DPS-11 digital force gauge.

RESULTS of OBJECTIVE IV

Maximum, minimum, and average temperatures for the duration of the trial were 33, 15, and

25ºC, respectively. No fruit or leaf drop occurred at the time of measurement (data not shown). The analyzed data in Table 4.1 below demonstrates there were no significant difference among the eight treatments and only 5000 ppm Ethrel and 10ppm Sumagic significantly decreased FRF relative to the control.

CONCLUSIONS of OBJECTIVE IV

1. Eight abscission compounds or combinations were screened. 2. Only 5000 ppm Ethephon and 10 ppm Sumagic significantly decreased FRF when

compared with the control treatment, but were not significantly different from the other six treatments.


59

Table 4.1. Evaluation of four compounds at two different concentrations each on the decrease in FRF, and fruit and leaf drop of Manzanillo table olives.

Treatments #Fruit dropped FRF 10/9 Change in FRF

10/24 #leaves dropped

T1: Ethrel (23% a.i.) 2000 ppm- 0.87 mL 0.6 b 477.0 ab 104.6 ab 1.8 ab T2: Ethrel (23% a.i.) 5000 ppm- 2.2 mL 4 c 445.2 a 136.4 b 2.6 b T3: Profile (21.8% a.i.) 1000 ppm 0.46 mL 0 a 497.9 ab 83.7 ab 1 a T4: Profile (21.8% a.i.) 2000 ppm 0.92 mL 0 a 486.5 ab 95.1 ab 1 a T5: Sumagic 10 ppm - 2 mL 0 a 453.4 a 128.2 b 1 a T6: Sumagic 100 ppm - 20 mL 0 a 483.3 ab 98.4 ab 1 a T7: Embarck 1000 ppm - 0.36 mL 0 a 476.4 ab 105.2 ab 1 a T8: Embarck 2000 ppm - 0.72 mL 0.2 ab 507.7 ab 73.9 ab 1 a T9: Control 0.2 ab 562.0 b 19.6 a 1 a

Four successive years of screening potential abscission compounds have not identified any

potential abscission agents for table olives. It appears a more considered approach to decreasing FRF is needed.

FUNDING

The 2009 harvest research reported here was funded entirely by the California Olive

Committee: $117, 545.00

ACKNOWLEDGEMENTS

We gratefully acknowledge the funding support of the California Olive Committee. We gratefully acknowledge the cooperation of Musco Family Olive Company, particularly Edward Melanesio, Jesus Lopez, Dennis Burreson, Matthew Kobal and Abdul Sigal, and Bell Carter Olives, particularly Jane Yegge and Cody McCoy. These California experiments would not have been possible without the cooperation of Rocky Hill Ranch and Marc Pascoe and Jesse Luna, and Nielsen Ranch and Erick and Gavin Nielsen. We wish to thank Erick Nielsen of ENE Inc., Phil Scott of Agright and Dave Loquaci of Madera Ag Services for their patient and flexible cooperation in the harvester evaluations. Finally, we thank volunteer John Henry Ferguson BS, MBA and visiting Scientist Maria Paz Suarez Garcia, Ph.D. of University of Sevilla, Spain for their experimental field and laboratory assistance.

developing mechanical harvesting for california black ripe

Documents