Clarence E. Watson

Evaluation of genetic variances, heritabilities, and correlations for yield and fiber traits among cotton F2 hybrid populations

SummaryF2 hybrid cultivars continue to occupy a small portion of the cotton (Gossypium hirsutum L.) production are in the United States, but occupy a larger proportion of the production area in some other countries. Sixty-four F2 hybrids resulting from crosses of four commercial cultivars and 16 pest-resistant germplasm lines were evaluated for five fiber and four yield traits in four environments at Mississippi State, MS. An additive-dominance genetic model was employed for these traits. The minimum norm quadratic unbiased estimation (MINQUE) method was used with a mixed model approach for estimating genetic variance and covariance components and for predicting genetic correlations. This study investigated genetic variances, heritabilities, and genetic and phenotypic correlations between agronomic and fiber traits among these 64 F2 hybrid populations and discussed the usefulness of these populations for use as hybrids or for selections for pure lines.Dominance variance accounted for the major proportion of the phenotypic variances for lint yield, lint percentage, and boll size indicating that hybrids should have an advantage for these traits compared to pure lines. A low proportion of additive variance for fiber traits and the significant additive x environment variance components indicated a lack of substantial useful additive genetic variability for fiber traits. This suggests that selections for pure lines within these F2 populations would have limited success in improving fiber traits. Genetic and phenotypic correlation coefficients were of comparable magnitude for most pairs of characters. Fiber strength showed a positive additive genetic correlation with boll weight. Dominance genetic correlations of fiber strength with elongation and 2.5% span length were also significant and positive; however, the additive genetic correlation of length and strength was zero.

Establishment of molecular markers and linkage groups in two F2 populations of Upland cotton

Two F2 populations of cotton (Gossypium hirsutum L.) from the crosses of HS46 x MARCABUCAG8US-1-88 (MAR) and HS46 x Pee Dee 5363 (PD5363) were characterized for restriction fragment length polymorphisms (RFLPs) using DNA probes. Seventy-three probe/enzyme combinations were used in the HS46 x MAR population analysis, which resulted in 42 informative polymorphic fragments. These 42 moleclar markers represented 26 polymorphic loci, which consisted of 15 codominant and 11 dominant (+/-) genotypes. Chi-square analyses of these loci fit expected genotypic ratios of 1∶2∶1 and 3∶1, respectively An analysis of these loci with the MAPMAKER program resulted in the establishment of four linkage groups A, B, C, and D with 4,2,2, and 2 loci, respectively, as well as 16 unlinked loci. Six probe-enzyme combinations were assayed on the HS46 x PD5363 population, which resulted in 11 informative polymorphic fragments. These 11 fragments represented 6 polymorphic loci, 1 dominant (+/-) and 5 codominant genotypes. The MAPMAKER analysis of these loci yielded 2 linked loci. Thus, a total of 53 polymorphic fragments and 32 polymorphic loci, representing five linkage groups, were identified among the two families.

Comparisons of quantitative trait locus mapping properties between two methods of recombinant inbred line development

Theoretical comparisons for quantitativetrait loci (QTL) mapping properties wereconducted among simulated recombinantinbred (RI) populations developed bysingle-hill (SH), complete bulk, and singleseed descent (SSD) procedures by MonteCarlo simulations based on variouspopulation sizes, heritabilities, and QTLeffects. Our simulations includedestimation of QTL effects, QTL positions,and statistical testing power in the RIpopulations by comparing the estimates withpreset values. The simulation resultsshowed that the single hill (SH) bulk andsingle seed descent RI populations weregenerally not significantly different withrespect to quality of estimated QTL effectsand positions. Furthermore, when each RIpopulation had 150 lines, each couldprovide desirable properties for QTLmapping. The results implied that a SH RIpopulation consisting of 75 or moreF2-derived families with two lines perfamily (corresponding population size of150 or above) was appropriate for QTLmapping and was not significantly differentthan a SSD RI population of 150. Thus, theSH method could be used to develop largenumbers of RI lines for achieving betterresults in QTL mapping. Simulations alsoshowed that there was no significantdifference between means using SH methodswith 10 and 100 fruits per family. However, RI populations developed by thecomplete bulk method where F2identities are lost were not suitable forQTL mapping.