Degree Type


Date of Award


Degree Name

Doctor of Philosophy



First Advisor

Kenneth J. Frey


Mass selection is frequently used for genetically improving pearl millet (Pennisetum glaucum R. Br.). To ascertain the ability of mass selection to modify traits of agronomic importance to pearl millet, this study determined (a) the heritability and interrelationships of those traits and (b) the realized gains obtained from selection. S[subscript]0 and S[subscript]1 populations of three pearl millet composites were evaluated for an array of agronomic and developmental traits. Parent-offspring heritability values ranged from 0.46 to 0.64 for panicle size and seed traits, from 0.27 to 0.58 for productivity traits, and from 0.16 to 0.32 for partitioning traits, when averaged over three pearl millet composites. Interrelationships among traits were identified by factor analyses and found to be similar in the three composites. Unique groups of traits were associated with biological yield, panicle size, and seed factors. Certain traits, however, were associated with both the biological yield and the partitioning factors. The orientations of S[subscript]0 plants along the biological yield, panicle size, and seed parameters factors were significantly related to orientations of their S[subscript]1 progenies along the corresponding factors.;Forty-six experimental populations were created via bidirectional selection for grain yield in three pearl millet composites. Selection criteria used were grain yield per se and indices of yield component or developmental traits. Upward selected populations from the EC, NELC, and D[subscript]2C composites outyielded their downward selected counterparts by 635**, 312**, and 155* kg ha[superscript]-1, respectively, when evaluated under high fertility at the location of selection (Patancheru). Yield differences between upward and downward selected populations were greatly reduced, however, when the populations were evaluated at sites distant from Patancheru. Selection for developmental trait indices gave the greatest yield responses at Patancheru, whereas selection for grain yield per se or yield component traits gave the largest yield responses at locations distant from Patancheru. Selection indices constructed via regression analyses or via intuition were equally effective. The composites differed for symmetry of response to selection when compared at Patancheru. Yields of upward and downward selected EC populations differed from the selected EC bulk by +13% dn -16%, respectively, whereas NELC populations deviated by -3% and -13%, respectively, from the NELC bulk.



Digital Repository @ Iowa State University,

Copyright Owner

Henry Fredrick Rattunde



Proquest ID


File Format


File Size

113 pages