Premium
Classification of Geographical Origin by PNN Analysis of Fatty Acid Data and Level of Contaminants in Oils From Peruvian Anchovy
Author(s) -
Standal Inger B.,
Rainuzzo José,
Axelson David E.,
Valdersnes Stig,
Julshamn Kåre,
Aursand Marit
Publication year - 2012
Publication title -
journal of the american oil chemists' society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.512
H-Index - 117
eISSN - 1558-9331
pISSN - 0003-021X
DOI - 10.1007/s11746-012-2031-0
Subject(s) - anchovy , principal component analysis , docosahexaenoic acid , fatty acid , stock (firearms) , fishery , geography , environmental science , polyunsaturated fatty acid , biology , mathematics , statistics , archaeology , biochemistry , fish <actinopterygii>
Abstract The aim of this study was to examine Peruvian anchovy oil fatty acid (FA) compositions, and to test the possibility of using the FA data to classify the oils according to geographical origin along the Peruvian coast. The levels of contaminants in a representative set of samples were determined to examine the general levels and investigate if such measurements could aid in future discrimination between oils. The FA results showed that the two known stocks of Peruvian anchovy displayed different levels of docosahexaenoic acid (DHA, 22:6n‐3) (southern stock; 14.4 ± 0.8% versus central‐northern stock; 9.9 ± 1.2%). However, principal component analysis (PCA) of the FA data indicated clusters according to three regions; North, Center and South. Using a data set of 57 anchovy samples and 21 FA as input, a probabilistic neural network (PNN) was constructed. For the validation data sets, “North” oils was predicted accurately 100% of the time, “Center” oils 100% and “South” oils 83% of the time. The levels of contaminants in the oils determined were low in all but one sample.