Methodology to remove strong outliers of non-climacteric melon fruit aroma at harvest obtained by HS-SPME GC-MS analysis

A methodology for making consistent studies of outliers of non-climacteric melon volatile organic compounds at harvest is reported. The juice was squeezed from the fruit of the ‘Piel de sapo’ cultivar harvested during two consecutive seasons and the aroma volatiles were extracted by headspace solid phase microextraction and measured by gas chromatography coupled to mass-spectrometry. A deconvolution analysis was performed to obtain volatile organic compounds. For multivariate the reliable identification of outliers, compound classes were studied as a percentage of total area counts of the melon compounds identified in the chromatogram by principal component analysis and partial least-squares discriminant analysis, and then verified by correlation analysis, box-whisker plot, and formal tests for univariate outliers. Principal component analysis was the key methodology for selecting outliers in variables that mostly did not follow a normal distribution. The presence of an excess in terms of relative percentage of area and the diversity of minor compounds such as alcohols, terpenes, acids, among others, are usually a sign of anomalous data that can be considered outliers in the aroma of this non-climacteric cultivar. This multivariate approach removed outliers, but kept the variability of aroma among the samples of every cultivar.