Geographical variations in the fatty acids of Zanthoxylum seed oils: A chemometric classification based on the random forest algorithm

Some Zanthoxylum species have the potential to develop oil-oriented wood plants. But confusion of seed oils and low environmental adaptability of fatty acid lead to obstacles to their further development. In this work, a total of 99 seed samples of two widely distributed Zanthoxylum species (Zanthoxylum bungeanum and Zanthoxylum armatum) seeds were collected from 97 plantations located at 32 locations in China spread over a wide geographic area (a range of 17.2 degrees longitude and 12.8 degrees latitude) and fatty acid composition of seed oils extracted by supercritical CO2 extractor apparatus among individuals was determined. The random forest algorithm was employed to conduct a classification model to identity two species' oils and the relationship between variation of fatty acid composition and environmental factors was analyzed. Two species seed oils contained C15-C20 fatty acids, of which palmitoleic acid was the most abundant in Z. armatum and oleic acid was the most abundant in Z. bungeanum. The built classification model showed 100% accuracy and explained proportions of total fatty acid composition's variance by environmental factors between Z. armatum and Z. bungeanum were 83.15% and 67.43% respectively. Among all environmental factors, latitude had a significant contribution to variation of fatty acid composition in Z. armatum seed oil, while annual maximum precipitation greatly contributed to variation of fatty acid composition of Z. bungeanum seed oil.