ESTIMATION OF OPTIMAL SENSORY CONDITIONS FOR

ROASTING SHELLED MACADAMIA INTEGRIFOLIA

VARIETY ‘YONIK’₁

Reprint From CMS Yearbook 1986

With a gradual increase in the quantity available of Macadamia integrifolia variety ‘Yonik’ (Kadman and Slor, 1982, 1983), interest was expressed in determining the condition for dry roasting such that optimal sensory quality would be attained. In earlier trials, the nuts were roasted in-shell (Rosenthal et al, 1984): this course afforded the kernels some protection against minor fluctuations in the temperature-time treatment, and against exposure to atmospheric oxygen; roasting the nuts after shelling would normally eliminate these defenses, but is more energy-efficient.

As in the earlier trial (ibid.), preliminary testing was required to delineate an approximate range for appropriate treatment. Instead of the round-table procedure then adopted for this purpose, the relatively large selection of roasting conditions was now screened by individual presentation to preliminary-assessors. A taste panel larger by an order of magnitude was then employed to give more precise estimates of the optimum treatments.

Nuts were collected and prepared as previously described (ibid.), except that shelling now preceded roasting: this was performed at 1 000, 1100, 1200 and 1300 for 10, 20, 30 and 40 mm. On the next day, samples from each of these 16 treatments, together with an unroasted sample, were coded with 2-digit random numbers and presented simultaneously to the preliminary panel. The 11 assessors were requested to (a) arrange the samples in order of visual color preference, (b) score the color hedonically on a scale from zero (poorest possible) to 10 units, and (c) now taste the samples, and score this independent parameter similarly to (b). The assessors were all wells acquainted with both Macadamia nuts and taste panel procedures, but no other criterion was used in their selection. The data for the unroasted samples are not considered in the analysis below.

A relationship was sought between the mean hedonic scores assigned to the visual color, and the temperature-time treatment variables. A linear multiple regression equation was obtained:

with standard error of the estimate = 0.7 units. A similar but quadratic multiple regression equation was found to have a correlation coefficient of r = 0.87, which is not significantly greater than that for Equation [1], which latter is therefore preferred, and is shown graphically in Figure 1. From the Figure, a clear preference is seen for the milder roasting treatment.

The hedonic taste scores were considered similarly. A quadratic multiple regression gave the correlation equation:

with standard error of the estimate = 0.4 units. A similar but linear regressions equation was found to have a correlation coefficient of r = 0.61, which is significantly less than that for Equation [2]. This latter is therefore preferred, and is shown graphically in Figure 2. A hypothetical optimal score was extrapolated outside the experimental range, at 96 C for 11 min.

Based on the results from this preliminary test, shelled nuts were roasted for the larger-scale panel at 90 and 100 C for 10 and 25 min., and at 110 C for 5 and 15 mm. These six products were presented in random order to 104 unpreselected assessors, principally researchers, technicians and clerks of the Agricultural Research Organization with general experience in taste panel procedures on a wide variety of products. The assessors were requested to taste the products and (a) state their order of preference among them (first preference 1, second preference = 2, etc.); (b) rate their taste quality on a 15-point non parametric pictorial scale, labeled now for report purposes only as from "A" (highest possible quality) to "0"; and (c) score the taste quality on a zero-to-10 scale as above. A few replies were unclear or illegible, and were therefore not considered. Lower values of the mean rank total imply higher preference.

From the results now obtained, the mean rank totals gave the equation:

= 15.4 - 0.221 (temperature) + 0.00106 (temperature) ² - 0.068 (minutes) + 0.00179 (minutes)² [3]

(N = 6, r = 0.9972, d.f = 1, 0.01 < p 0.05)

with standard error of the estimate = 0.02 units. An optimal value was interpolated at 104 C and 19 min.: see Figure 3.

The median non parametric rating over all six products was that labeled "F’. The proportions of rating "F"-or-better gave the equation:

with standard error of the estimate = 0.05. An optimal value was interpolated at 104 C and 16 min.: see Figure 4.

The quadratic multiple correlation of mean taste quality score and temperature-time treatment conditions was not found to he significant (N = 6, r = 0.75, d.f. = 1).

From the larger-scale panel results, it is concluded that optimal roasting of shelled ‘Yonik’ Macadamia nuts, after preparation as described (Rosenthal et al, 1984), will be obtained at approximately 104 C for approximately 16 min.

Kadman, A. and Slor, F. (1982). ‘Yonik’ macadamia. HortScience 17:991,

Kadman, A. and Slor, F. (1983)- ‘Yonik’ - a new selected Macadamia cultivar. Yb. Ca. Macadamia Sac. 29:138-141.

Rosenthal, 1., Merin, U., Basker, D. and Kadman, A. (1984). A study of Macadamia nuts of ‘Yonik’ variety. J. Ed. Qual, 7:67-73.