Evaluation of apple quality based on principal component and hierarchical cluster analysis

Abstract: The purpose of this study was to investigate the variations in physical and chemical characteristics of apple fruit from 30 varieties grown in the same place using pattern recognition tools. Twenty quality parameters of apple samples (e.g. weight, volume, density, color, hardness, sugar-acid ratio, Vitamin C, etc.) were analyzed. Interrelationships between the parameters and the apple variety were investigated by descriptive statistics, principal component analysis (PCA) and hierarchical cluster analysis (HCA). PCA is a mathematical tool which performs a reduction in data dimensionality and allows the visualisation of underlying structure in experimental data and relationships between data and samples. In hierarchical cluster analysis, samples are grouped on the basis of similarities, without taking into account the information about the class membership. The results obtained following HCA are shown as a dendrogram in which five well-defined clusters are visible. Samples will be grouped in clusters in terms of their nearness or similarity. Cluster analysis uses less information (distances only) than PCA. It is interesting to observe what kind of classification can be made on the basis of distances only. The results showed that density, fruit shape index and water content of 30 apple varieties were not significantly different. The remaining seventeen measurements were investigated by principal component analysis. The first six components represented 83.56% of the total variability on the base of the total variance explained and screen plot of principal component analysis. The first principal component was related to titratable acidity, sugar-acid ratio and solid-acid ratio attributes, which were the taste quality factor. The second principal component was related to L, a, and b attributes, which were the color factor. Following that were sweetness factor, texture factor, quality factor and size factor. The sample score plots visually displayed the relationship between measurements and apple varieties. The varieties of Liaofu, Lixiang, Golden Delicious and Rome Beauty were located on the first quadrant of the plot for PC1 vs. PC2. They had better quality of among the mid-early maturity apples with high content of sugar-acid and solid-acid ratios which were good characteristics of taste even though their skin color was green. The varieties of Hanfu, Rainier and Fuji were located on the fourth quadrant of the same plot. They were common late maturity apples with red color and good quality for eating. Priam, Baixing and Pvma were located on the second quadrant of the plot, which may be suitable for processing not for eating, and they had a green skin color and sour taste. The varieties located on third quadrant were red but had low sugar-acid and solid-acid ratios, which can be used as breeding materials for red color and good quality of the early maturity apple. HCA classified 30 varieties into five main groups on the basis of the measured parameters, which was consistent with the results of PCA score plots. These results can be used preliminarily to determine whether the 30 apple varieties were appropriate for eating or for processing or for use as a breeding stock for future apple variety development.