Statistically designed two step response surface optimization of enzymatic prepress treatment to increase juice yield and lower turbidity of elderberry juice

Abstract

This study examined the effects of different, statistically designed, enzymatic maceration treatments on juice yield, turbidity, and phenol yield (total phenols and total anthocyanins) in experimentally produced elderberry juice. Increased pectinolytic enzyme dose, longer maceration time, and elevated reaction temperature all had significantly positive effects on the juice yield. Increased enzyme dose and maceration temperature also increased the yields of anthocyanins in the elderberry juice, while none of the reaction parameters affected the juice turbidity. The juice yield was optimized further in a new experimental template, made by using the statistical steepest ascent optimization method. In the new response surface design template an optimal maceration treatment giving maximal juice yield and anthocyanin yields and low turbidity was identified. With the optimal treatment with a pectinolytic enzyme preparation, Pectinex BE 3L, produced by a cloned Aspergillus strain, a maximal juice yield of 77% w/w of the berry mash, an anthocyanin yield of 2380 mg/kg fresh berry mash, and a turbidity level of 128 formazin nephelometric units (FNU) were obtained. Enzymatic prepress treatment generally decreased turbidity levels by 30% as compared to pressing without prior enzymatic treatment. A comparison of the responses obtained after the optimal enzymatic treatment with five different pectinolytic enzyme preparations showed that the Aspergillus niger preparation Pectinex BE Color gave slightly better juice and phenol yields, and lower turbidity levels than the other enzyme preparations tested. In conclusion, the results demonstrated that juice yields and phenolic yields in elderberry juice could be improved with enzyme treatment and that the optimal reaction conditions for obtaining the best juice yield, highest phenolics, and lowest turbidity levels could be rationally identified via statistical factor level optimization.