Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


A multi-environmental trial was conducted in Rwanda, located in the highlands of Central Africa. The trial consisted of 26 cultivars of sweet potatoes repeated over 11 environments. Genotype X environment (G X E) interactions were highly significant for vine yield, fresh root yield, number of roots, and dry matter yield. Average root weight was not significant for G X E interactions. The G X E variance component was generally more important than the variance component for cultivars. From pathway analysis it was concluded that the G X E effects of dry matter yield were little determined by the G X E effects of the yield components except for number of roots. Specific and relatively independent adaptation mechanisms were postulated for each of the yield components. Genotypic correlations between yield components were neutral indicating that stress and competition between the yield components were low in the sweet potato. The parallel pathway diagram was found to be more adequate to explain the causal relationships between yield and its components, than the allometric (sequential) model. The main yield contributor was number of roots, followed in decreasing order by, average root weight, vine yield, and % dry matter. Three progeny trials were established in Rubona (Institut des Sciences Agronomiques du Rwanda), consisting of 18 half-sib progenies of local origin, 26 from IITA germplasm (International Institute for Tropical Agriculture, Nigeria), and 25 from special seed gardens. The latter trial was planted in a swampy area. The narrow-sense heritability for yield and yield components was generally low. The heritabilities for flesh color, skin color and anthocyanin pigment were generally high. The genetic associations for root yield, number of roots, and vine yield were strong and generally positive. In the swamp trial there was a negative genetic correlation between vine yield and number of roots probably due to the hydromorphic growing conditions. Because of relatively high co-heritability estimates, correlated genetic advance for yield was generally high when selecting for number of roots and vine yield.