Physical Capsule and Seed Characteristics Impact Nutrient Densities in Cuphea Seed.

We quantified interrelationships between nutrient densities and their dependencies on weights and physical characteristics of capsules and seeds of indeterminate Cuphea plants subjected to no (GDD0) or to source-sink manipulation by removing the top 25% of plant foliage at 100 and 200 growing degree days (GDD1 and GDD2, respectively) after flower initiation. Measurements on capsule weight, number and weight of seeds per capsule, and packaging cost (i.e., ratio of capsule tissue weight to seed weight) (G1), and on perimeter, area, height, width, and major and minor axis of capsules (G2), and of individual seeds within each capsule (G3) indicated that canonical r-values were significant (p<0.01) and increased in magnitude between traits in  G1 and G2 (0.77), G1 and G3 (0.80), and G2 and G3 (0.85); whereas, those between nutrient densities and traits in G1 (0.94), G2 (0.82), and G3 (0.73) exhibited a gradual decrease. Carbon and N densities exhibited opposite, but not always significant, correlations with capsule and seed physical characteristics; the same trend was observed for N and S densities. Seeds sampled from plants subjected to GDD0 and GDD2 treatments were 100% correctly classified based on nutrient densities; whereas, those sampled from plants subjected to GDD1 were 88% correctly classified. Discrimination among seed samples based on traits in G1, G2 or G3, singly or in combination, was not as accurate. Regression models differed in predicting nutrient densities in seed as functions of treatments, years and seed physical characteristics, and separated nutrients into a group with R2<=0.25(S and Zn) and two larger groups one with R2>0.25 and <0.50 (Ca, Cu, K, P and Se) and another with R2=>0.50 and <0.70 (B, Ba, C, Fe, Mg, Mn, and N). The results provide selection criteria aimed at improving the inherently non-uniform size of Cuphea seed as a prerequisite to larger seed and oil yields.