Predictig Persimmo Puree Colour as a Result of Puree Stregth Maipulatio. Adrew R. East a, Xiu Hua Ta b, Jataa Sutudprom a a Istitute of Food, Nutritio ad Huma Health, Massey Uiversity, Private Bag 11 222, Palmersto North, New Zealad (a.r.east@massey.ac.z) b Massey Uiversity (Sigapore), Block T1A25, 500 Dover Road, Sigapore ABSTRACT Colour is a importat attribute of ay food product. Persimmo are oted for their itese orage to red colourig ad hece products marketed as made from persimmo are expected to iherit this attribute. As the persimmo fruit ripes the etire flesh of the fruit chages colour from a pale yellow-orage to a dark itese orage-red. This chage is colour imparts chages i the colour qualities of a puree product. The objective of this research was to develop a model to eable colour predictio of a persimmo puree as a fuctio of the fruit maturity ad puree stregth. As a result a mathematical tool is created that eables desired colour properties of a puree product to be achieved by maipulatig the stregth of the puree. Colour data for modellig was created with 3 sets of persimmo fruit of differet maturities (firm, medium ad soft) which were maipulated to 10 differet solids cocetratios ragig betwee half ad double stregth solids. Purees greater tha sigle stregth were created with the additio of freeze dried persimmo solids (source from the same persimmos), while lower tha sigle stregth purees were created with the dilutio of water. Tristimulus CIELAB colour data was collected with a Miolta Spectrophotometer. Replicate 3 sets of persimmo puree were created ad measured for model validatio. The model was created to allow predictio the chage i colour (L*, a*, b*, C or h) as a fuctio of the chage i stregth of the puree. As such the raw data was coverted to fractio chage for both the solids (stregth) ad colour parameters. Coverted data was fitted with empirical secod order polyomial models that were applicable irrespective of the maturity of the fruit used to make the puree. The validatio of the model revealed mea absolute errors of predictio of less tha 1.2 for all CIELAB parameters. The resultig model ca be used to assess the impact of puree stregth maipulatio o fial product colour. Keywords: Diospyros kaki, CIELAB, colour model, fruit processig. INTRODUCTION Colour is a importat quality defiig attribute for ay food product [1] as it is the first perceived judgemet of the quality of the food by the cosumer. Quatitative measuremet of the colour of foods is routiely reported i the Commisio Iteratioale de L Eclairage (CIE) 3-dimesioal model. The CIE L*C*h system allows descriptio of the colour i terms of lightess (L*), ad the polar co-ordiates chroma (C*) ad hue (h) while the same colour ca be described by the CIE L*a*b* system where the Cartesia coordiates a* (gree-red) ad b* (yellow-blue) are used. The persimmo (Diospyros kaki) is a ative plat of Chia i which fruit are produced durig the autumwiter seaso. Numerous cultivars are i commercial productio, with major differeces betwee cultivars iclude the presece of astrigecy ad seeds. Astriget cultivars (e.g. Hachiya ) require ripeig (util they are soft) before beig cosumed, whereas o-astriget cultivars (e.g. Fuyu ) ca be cosumed whe either a crisp apple like texture (firmess = 9 kg f ) or a soft peach like texture (firmess = 1 kg f ). The persimmo is cosidered as a climacteric fruit, meaig that it has the ability to cotiue to ripe subsequet to harvest. Fruit processig provides a meas of utilisig fruit that are ot saleable as fresh product ad preserves the uique qualities ad taste of seasoal fruits throughout the remaider of the year. Global trade i processed fruit product is estimated at US$26 billio [2]. Fruit ca be coverted ito juice, puree, paste ad syrup that are either sold directly as cosumer products (e.g. sauces) or combied with bakig ad dairy products to produce characteristic fruit flavoured combiatios. The defiitio ad differeces of the liquid ad semiliquid food systems of paste, syrup or juice are largely idicators of the textural characteristic of the processed fruit products. Oe meas of maipulated these textural characteristics is to alter the cocetratio of solids withi the product.
As most processed fruit products covert solid foods to liquid or semi-liquid products, the colour of the resultig product is a importat attribute i allowig the cosumer to idetify ad establish autheticity that the product is ideed made from the fresh fruit. Cosumers expect tomato products to be red, avocado gree, baaa yellow, ad persimmo orage. Durig fruit ripeig, sythesis ad degradatio of pigmets (caroteoids, chlorophyll ad athocyais) occurs resultig i dramatic colour chages. This atural chage i colour due to ripeig, creates challeges i producig cosistet colour properties of fruit based products. Pilado et al. [3] previously foud that strawberry maturity played a importat role i ifluecig the colour characteristics of strawberry wie. As solids cocetratio is likely to be maipulated ad fruit maturity variable for ay potetial persimmo fruit product, this work aimed to create a model that would eable predictio of the effect of solids cocetratio maipulatio o persimmo puree colour irrespective of fruit maturity. MATERIALS & METHODS Persimmo (Diospyros kaki, cv. Fuyu) were supplied from a commercial producer located ear Whagarei, New Zealad. Persimmo were stored at 0 ± 1 C, 90-95% RH before use. Fruit maturity was gauged based o fruit firmess. Flesh firmess was evaluated usig a electroic peetrometer (Willowbak Electroics, New Zealad) equipped with a 8 mm roud (Effegi) probe. Approximately 0.6 mm of fruit ski was removed at two opposite positios o the fruit equator prior to measuremet. Flesh firmess was cosidered as the average peak force (kg f ) required to pucture the flesh at a speed of 20 mm.s -1 to a depth of 8 mm. Fruit were ripeed at 20 C ad destructively measured i order to idetify approximately 7 kg of fruit withi each maturity class. Three maturity classes represetig firm (6.5-8 kg f ), medium (3-6.5 kg f ) ad soft (0.1-3 kg f ) were used. A source of persimmo solids was required to create higher tha ative stregth purees. A sigle batch of dried persimmo powder was created by freeze dryig firm persimmos acquired from the same source at the start of the experimet. This freeze dried powder was recostituted with de-ioised water to obtai a 45% solids paste, that was used to mix with the ative state puree to obtai high stregth (solids cocetrated) purees. Persimmo purees were created by first preparig a large batch of the ative state puree. Whole fruit were peeled ad the calyx was removed before pureeig i a food processor (Ultima, Black ad Decker Ic) for 8 miutes. This ative state puree was mixed i proportio with either de-ioised water or the 45% solids puree paste ad mixed for 1 miute with a had held bleder to create 10 purees i the rage of half to double stregth. Actual (rather tha desig) solids cotet was verified by later dryig the puree sample i a vacuum ove. Colour of the persimmo purees were measured immediately after preparatio i order to avoid the ifluece of ezymatic ad o-eymatic browig effects. Measuremet was coducted with a Miolta Spectrophotometer (CM-2000d, Koica Miolta Sesig, Japa) usig reflectace mode with illumiat D65 ad a 10 observatio agle. Purees were preseted to the spectrophotometer i a petri dish ad hece calibratio with a white calibratio tile was also coducted through a petri dish. Samples were preseted by fillig a petri dish to a height of 19 mm ad coverig the sample with a ca with a black iterior. The experimet was coducted twice i order to create two sets of data, oe beig used for model creatio ad the secod for model validatio. The aim of the project was to create a sigle model that was fuctioal idepedet of the fruit maturity used to develop the puree. As such all solids cocetratio ad colour parameter data (x i ) was coverted to a fractioal chage scale ( x) relative to the values measured for the ative state puree (x, Eq. 1). As such the developed model aims to predict the chage i colour based o a fractioal chage i solids cotet from the ative state puree.
xi x = 1 (1) x The resultig trasformed data limits the rage of fractioal solids chage ( S) values to betwee approximately -0.5 ad 1, represetig half ad double stregth purees. The resultig data sets spaig across the three maturities created smooth curved relatioships betwee the relative chage i all CIELAB colour parameters ad the relative chage i solids cocetratio ( S). A simple secod order polyomial was used to fit ad describe the patters observed (Eq. 2) for each CIELAB colour parameters idepedetly. This polyomial was forcibly fitted through the ative puree characteristics (i.e. whe S = 0, x = 0) as o chage i colour is possible with o chage i solids cotet. 2 ( S) + z( S) x = v (2) Validatio of the resultig model was coducted by applyig the equatios for predictig the colour chage to the secod idepedet data set. Model accuracy was quatified by calculatig the mea absolute error (MAE) of predictio (Eq. 3) i= MAE = 1 x o x p (3) Where = umber of observatios; x o = observatio value; x p = predicted value. RESULTS & DISCUSSION Irrespective of fruit maturity it was observed that dilutio i the solids cotet of the puree result i reduced L*, a*, b* ad C parameters ad icreases i h, while the iverse was true for purees with icreased solids cocetratio (Figure 1). These chages i the CIELAB colour parameters suggest that dilutio of solids cotet i persimmo puree results i a darker (lower L*), less itese (lower C) ad more yellow (icreased h) product colour. I most cases, the secod order polyomial fitted through the ative puree properties was foud to be a adequate descriptio of the chage i colour parameter as caused by a chage i solids cotet (Table 1, Figure 1). Hue agle was adequately described, with the exceptio of the regio of high solids cotet ( S > 1.5) where the fitted model deviates from the acquired data (Figure 1e). This poor fit may be iflueced by a potetial outlier (at S 0.6, h 0), ad hece was refitted without this poit ad foud to result i a more satisfactory fit (dotted lie, Figure 1e, Table 1). Table 1. Fitted costats for the secod order polyomials to describe the fractioal chage i CIELAB colour parameters as a fuctio of the fractioal chage i puree solids cotet. CIELAB Parameter v z R 2 L* 0.139-0.083 0.91 a* 0.943-0.378 0.96 b* 0.651-0.255 0.98 C 0.689-0.273 0.99 H -0.113 0.122 0.89 H (refitted) -0.125 0.057 0.92
Figure 1. Fractioal chage i measured CIELAB colour parameters (L*, a*, b*, C ad h) as a fuctio of the fractioal chage i solids cocetratio. The curved lies represets the fitted secod order polyomial (Eq. 2) i which the costats are detailed i Table 1. Use of the polyomial predictio models o a idepedetly collected data set foud that the predictio was robust over a large rage of potetial values (Figure 2). For the five CIELAB parameters the mea absolute error of predictio raged betwee 0.55 (for L*) ad 1.17 (for h). The largest deviatio from predictios occurred for purees made from firm fruit ad with icreased ( S > 0.5) solids cocetratios. These deviatios from predictio was a result of low predictio of a* values from those observed (Figure 2b) which also results i predictio of higher h values (Figure 2e). CONCLUSION Colour is a importat attribute of fruit based products that ca be substatially iflueced by solids cocetratio maipulatio. I this work a simple secod order polyomial was used to model the chage i CIELAB parameters with maipulatio of cocetratio of solids i persimmo puree. The developed models were idepedetly validated ad foud to be robust, suggestig that predictio ad hece maipulatio of colour outcomes is achievable through solids cotet maipulatio. REFERENCES [1] Fracis F.J. 1995. Quality as iflueced by color. Food Quality ad Preferece, 6(3), 149-155. [2] UN Comtrade Database. 2009. Fruit, preserved ad fruit preservatios. <http://comtrade.u.org> [3] Pilado L.S., Wrolstad R.E., Heatherbell, D.A. 1985. Ifluece of fruit compositio, maturity ad mold cotamiatio o the color ad appearace of strawberry wie, Joural of Food Sciece, 50, 1121-1125.
(a) (b) (c) Figure 2. Idepedet validatio (predictio vs observed) of use of secod order polyomial (Eq. 2; Table 1) to predict CIELAB colour parameters (L*, a*, b*, C, ad h) of persimmo puree as a fuctio of maipulatio of solids cotet.