Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Plant, Environmental Management and Soil Sciences

First Advisor

Don R. LaBonte


The use of genetic markers for paternity identification was investigated in hexaploid sweetpotato. Traditional sweetpotato breeding relies on the open pollination of numerous parents and the visual selection of progeny. Previous empirical and theoretical evidence shows this type of breeding system is extremely inefficient. Typically, only a few of these maternal parents routinely produce superior progeny. A similar scenario exists for the paternal parent. Identifying specific parental combinations which produce superior progeny would improve breeding efficiency, hence the need to identify paternity in superior progeny. A computer simulation study established the minimum number of alleles and loci required for paternity analysis using DNA molecular markers. PCR-based microsatellite data was used to calibrate the computer system developed. Computer simulation studies were conducted to explore the application of genealogy reconstruction techniques on hexaploid individuals based on co-dominant marker data. The progeny obtained on each female parent is categorically assigned to each male with non zero exclusion probability based on its paternity likelihood. Computer simulations show it is possible to discriminate between putative parents with few mis-assignments. The number of loci scored for a 10 parent population should not be less than 20 in the case of 3 alleles per locus and no more than 10 loci for a five allele model. An increment in the number of alleles or loci increases the discriminatory power, with the number of alleles yielding a more important effect than the number of loci. It is shown for first time that polymorphic microsatellites in the hexaploid sweetpotato are able to disclose the polysomic inheritance pattern. The results obtained stress the necessity of screening sweetpotato sequence libraries for longer repeat sequences. It has been also demonstrated with an experimental population the feasibility of the paternity analysis in polyploids. The most likely parent method proved to be reliable and more powerful than simple exclusion for paternity allocation in breeding populations. Two microsatellite loci with 4 and 3 alleles respectively, allowed the paternity determination of 7 out of 8 offspring in a broad genetic base population and 3 out of 14 in a narrow genetic base population.