ABSTRACT: Triacylglycerol composition of 10 almond culti- vars grown in seven different environments was determined by high-performance liquid chromatography, and multivariate sta- tistical techniques (principal component and cluster analysis) were used to detect and establish associations among almond cultivars. In particular, linear discriminant analysis showed that cultivars Desmayo Largueta, Masbovera, Texas, Non Pareil, Tuono, and Guara are different from each other and can be clas- sified in different groups. This statistical approach also predicted the origin of eight hybrids produced from these cultivars. As a result, it is concluded that triacylglycerol composition in al- monds can distinguish genetic from environmental effects on almond triacylglycerol composition. Paper no. J8877 in JAOCS 76, 267–272 (February 1999). KEY WORDS: Almond, characterization, chemometrics, high- performance liquid chromatography, triacylglycerols. Chemometric techniques may reveal useful information from analytical data, including characterization of natural goods. However, caution must be taken to ensure that these techniques are used in an appropriate manner (1). We have applied some common statistical algorithms to fatty acids (2), free amino acids (3,4), and inorganic element composition of almond ker- nel (5). All these studies showed the existence of cultivar-de- pendent differences in chemical composition that were inde- pendent of growth conditions. To verify this hypothesis we evaluated triacylglycerol composition of the almond kernels in samples of cultivars that were grown under diverse conditions. A recent study on edible oils showed that triacylglycerol com- position could be differentiated using a two-dimensional plot of principal component analysis (PCA) of mass spectrometric data (6). In this work, we determined the triacylglycerol com- position of 10 almond cultivars (Mediterranean and American) from seven different localities and analyzed these data with multivariate techniques [principal component analysis (PCA), cluster analysis, and linear discriminant analysis (LDA)]. MATERIALS AND METHODS The 10 almond cultivars used in this investigation were: Des- mayo Largueta (DL), Marcona (MR), Guara (GU), and Mas- bovera (MS) from Spain; Texas (TE), Non Pareil (NP), and Titan (TI) from the United States of America; Tuono (TU) from Italy; Ferragnes (FE) from France; and Primorskyi (PR) from a Caucasian region. These cultivars were grown in six different geographical areas in Spain: María (Almeria, ALM); La Puebla de Don Fabrique (Granada, GR); Santomera and Cehegín (Murcia, MUR), Castalla and Bacarot (Alicante, ALIC), Mas Valero, Mas Bove, and Mora d’Ebre (Tarragona, TA); and Aula Dei (Zaragoza, ZAR); as well as Avignon (AVIG), in France. These locations differed significantly in climatological conditions. Granada and Zaragoza are in cold regions, while the rest of the localities have mild climates. A total of 107 samples was collected from cultivars at these lo- cations in 1996. Samples were stored at 4°C until analysis. Almond samples were blanched and ground in an electri- cal grinder. Oil was extracted in an all-glass apparatus, simi- lar to the commercial Soxtec, using a mixture of diethyl ether/hexane (1:1 vol/vol). Samples of 0.1 to 0.2 mg of oil were dissolved in 4 mL of acetone and filtered through a 0.45 μm pore size filter. Aliquots were injected into a Waters Mul- tisolvent high-performance liquid chromatography (HPLC) system (Milford, MA) equipped with a double piston pump, thermostatic control of column temperature, and a double re- fractive index detector model Waters 410. The column was a 4.6 × 250 m Symmetry C 18 , supplied by Waters. The injec- tion valve was a Rheodyne model 7125 with a 20 μL loop (Cotati, CA). HPLC-grade acetonitrile and acetone were ob- tained from Lab Scan Ltd. (Dublin, Ireland). Trilinolein, tri- olein, tripalmitin, and tristearin reference standards were from Sigma Chemical Co. (St. Louis, MO). The optimal elution solvent found was acetone/acetonitrile 65:35 (vol/vol). Elution was carried out isocratically at a col- umn temperature of 30°C and a flow rate of 1.5 mL/min for 30 min. The detector was held at 40°C and operated at medium sensitivity. Identification of triacylglycerol molecular species (TGMS) was achieved by comparison with relative retention times ob- tained in a reference soy oil (7). The same response factor was assumed for all TGMS and the content of each TGMS was calculated relative to triolein. Several statistical methods in an SPSS statistical package [SPSS, 1994 (8)] were used for data analysis. PCA was ap- plied to autoscaled data. The number of the components to be retained was selected by using the Scree test and Kaiser crite- Copyright © 1999 by AOCS Press 267 JAOCS, Vol. 76, no. 2 (1999) *To whom correspondence should be addressed at Department of Analytical Chemistry, University of Alicante, P.O. Box 99, 03080-Alicante, Spain. E-mail: maria.prats@ua.es J8877 A Chemometric Study of Genotypic Variation in Triacylglycerol Composition Among Selected Almond Cultivars M.S. Prats-Moya*, N. Grané-Teruel, V. Berenguer-Navarro, and M.L. Martín-Carratalá Department of Analytical Chemistry, University of Alicante, 03080-Alicante, Spain