Skip to main content
ARS Home » Research » Publications at this Location » Publication #76874
Title: MULTIPLE REGRESSION AND MIXED MODEL METHODS TO MEASURE EFFECTS OF MOLECULAR MARKERS FROM INBRED POPULATIONS ON QUANTITATIVE TRAITS

Author
item VAN ZYL, C. - UNIV. OF NEBRASKA-LINCOLN
item Van Vleck, Lloyd
item JOHNSON, B. - UNIV. OF NEBRASKA-LINCOLN
item SMITH, H. - PIONEER HI-BRED INTER.

Submitted to: American Society of Animal Science
Publication Type: Abstract Only
Publication Acceptance Date: 2/21/1997
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: The objective of this pilot study with corn as a model was to determine if markers of inbred parental lines are associated with genes that affect quantitative traits. Phenotypic data were from F1 progeny resulting from matings between stiff stalk synthetic (SSS) and non-stiff stalk (NSS) heterotic groups of inbred lines for a total of 4032 observations. Marker information was from 28 parental lines per heterotic group with 121 marker spread over the 10 chromosomes. Data for nine traits on related sets of single crosses were evaluated with derivative-free REML. Four models were used. In the basic model, lines in SSS were considered fixed and variance of lines in NSS was estimated with relationships among lines considered. The model for comparison also included marker alleles from NSS lines as covariates. Corresponding models considered lines in NSS as fixed and variance of lines in SSS was estimated with relationships considered. This procedure was an attempt to avoid confounding of marker alleles of one heterotic group and line effects of the other heterotic group. Covariates used for multiple regression of performance on marker alleles were entered on a sequential basis. Chromosome one markers were included in the first analysis, then markers for one additional chromosome at a time were added. Markers on chromosomes other than chromosome one confounded with chromosome one markers were subsequently excluded from the model. The log likelihood increased for 8 of 9 traits when the 20 remaining independent sets of markers were included in the analysis when either SSS or NSS lines were considered as random effects. With marker information included, estimates of variance among SSS lines increased for 6 of 9 traits and estimates of variance among the NSS lines increased for all traits.