Analysis of Moisture, Oil, and Fatty Acid Composition of Olives By Near-Inferred Spectroscopy: Development and Validation Calibration Models.

Olive is a new crop in Georgia with its acreage expanding. The ideal harvest time of olives is determined by the oil content of the fruits. With progression of maturity, the oil content of olives steadily increases but eventually plateaus at the maximum oil content. Once the oil reaches this threshold, the olive should be promptly harvested before over-ripening occurs and the quality of begins to degrade. Thus, the producers require testing their olives every few days to identify the best time to harvest for maximum yield and quality. However, the traditional wet-chemistry techniques for determining moisture, oil, and fatty acid composition in olives is very laborious, time consuming, and costly. If successfully developed and validated, the Near Infrared Reflectance Spectroscopy (NIRS), a rapid and low cost analytical method which is also precise and accurate, could be very appropriate to effectively address this situation. We developed and validated NIRS calibration models for analysis of moisture, fresh-matter-oil (oil-FMO), dry-matter-oil (oil-DMO), and nine different fatty acids namely palmitic, palmitoleic, stearic, oleic, linoleic, linolenic, arachidic, gadoleic, and behenic acids contents of fresh olives. A total of 128 samples, 98 samples from 2015 and 30 samples from 2016 growing seasons were utilized for this work. Fresh olives were blended with a coffee grinder until homogenous and pits were minced, packed in a circular NIR cell, and then scanned on a FOSS XDS NIR system covering both visible and NIR regions in the wavelength range from 400 to 2498 nm at 2 nm intervals to give a total of 1050 data points per sample. Calibration models were developed with 83-88 randomly chosen samples for various parameters using modified partial least squares regression with internal cross validation. Out of the12 models developed, the models for the 9 constituents such as moisture, oil-FMO, oil-DMO, and four major fatty acids namely palmitic, palmitoleic, oleic and linoleic acids (representing 88-97% of the total fatty acids) had low standard error of both calibration (SEC) and cross validation (SECV) with high coefficient of determination in both calibration (R² = 0.81-0.98) and cross validation (1 ̶ VR = 0.74-0.86). For these 9 constituents, prediction of an independent validation set of 35-40 samples yielded excellent agreement between the NIRS predicted values and the reference values based on the low standard error of prediction (SEP), low bias, high coefficient of determination (r² = 0.80-0.93), and high ratios of performance to deviation (RPD = SD/SEP; 2.21-3.85), indicating the models had good quantitative information. Thus, the results suggest that precise, accurate, and rapid analysis of fresh olives for moisture, oil, and major fatty acid composition can be done using NIRS at a low cost, meeting the analytical need of the industry.