Mineral composition and comparison of storage roots of sweet potatoes varieties (Ipomoea batatas (L.) Lam.) cultivated in the island of La Palma, Spain

Na, K, Ca, Mg, Fe, Cu, Zn and Mn were determined by atomic absorption spectrophotometry in roots of 15 varieties of sweet potatoes cultivated in La Palma. (De papa o mani, Parado, Cubana, Cubana de baselo gordo, Cubana de hoja redonda, Blanquita, Saucero o Lanzarotero, Pata de Gallo, Padron de Seda, Alicantina, Amarillo de Ano, De Lanzarote, Matojo, Matojo Fino y Rajada; the 5 first ones of short-cycle crop and the 10 remaining ones of long-cycle crop). Variables data were subjected to analysis of variance, and means were compared by Duncan's multiple range test. There were calculated the coefficients of Pearson's correlation (r) to establish degrees of relation of the minerals among the varieties; Student's t-test to compare classifying to the samples according to crop cycles (Short and Long); analysis factorial to interpret the existing relations to slant factors and discriminant analysis (of stepwise with criterion of selection Wilks's lambda (lambda) and introducing all the variables) validated by means of cross-validation analysis. The mineral element (in mg/kg) present in higher concentration in the totality of the samples was K (2086 +/- 418), followed by Ca (974 +/- 486) and Mg (244 +/- 70), being the Zn the least present (1.12 +/- 0.61). The varieties with higher mineral content were Alicantina (Ca: 1809 +/- 327 and Cu: 2.97 +/- 0.23) and Padron de Seda (Mg: 435 +/- 46 and Mn: 3.08 +/- 0.49) and that of lower contribution was Parado (Cu, Mn, Ca, K and Mg). The Late varieties presented higher concentrations of Cu, Fe, Mn, Ca, K, Mg and lower of Zn and Na in relation to the Early varieties (p < 0.05) there was demonstrated that the concentrations of Cu, Zn and Na presented high number of significant correlations. The analysis factorial allowed to deduce that Mg, K and Fe are the variables that allow to characterize the system losing a minimum of total information. The percentage of original cases correctly classified after applying discriminant analysis to the varieties of short-cycle crop was 98 % of the original cases and 96 % after cross-validation, whereas for the varieties of long-cycle crop was 100 % of the original cases and 99 % after cross-validation.