Critical survey of interrelationships of nitrogen and amino acids in seeds

The present paper is intended for removing inconsistencies from available information concerning the interrelationhips of nitrogen (N) and amino acids in seeds (or grains). It consists of a critical survey of data reported in a series of papers by Mossé and collaborators (see Mossé and Huet, Sci. Aliments 1990, 10, 151 for review). These authors contended for perfectly linear relationships (A(i) = a(i)N + b(i)) to hold between N and amino acid i (g/100 g seed dry matter) for seeds of a given species on the basis of correlation coefficient (r) found to be close to 1. Three linear regressions are called upon to analyze variations in the amino acid compositions of 33 samples of pea seeds. The regression C(i) (g amino acid i/16gN) = 16a(i) + 16b(i)N(-)(1), not used customarily, is the more amenable to accommodate all the facts in pointing out values significantly lower than 1 for r(i)(2), and parameters a(i) and b(i) similar to those found with the regression A(i) = NC(i)/16 = a(i)N + b(i) whose r(i)(2) values, very close to 1, are overestimated on account of the presence of N in both sides of equation and sampling limited in number. Examination of variations in lysine content of maize grains as a function of N shows that the phenotypic variations for a given genotype are a poor fit to the linear relationship A(lys) = aN + b available in the literature, as determined from 153 commercial samples. It is concluded that linear N-amino acid relationships hold only approximately for predicting amino acid composition of seed of a given species from its N content.