Many studies of QTL locations record several different traits on the same p
opulation, but most analyses look at this information on a trait-by-trait b
asis. In this paper we show how the regression approach to QTL mapping of H
aley & Knott (1992) may be extended to a multi-trait analysis via multivari
ate regression, easily programmed in statistical packages. A procedure for
identifying QTL locations using forward selection and bootstrapping is prop
osed. The method is applied to examine the locations for QTLs for six yield
characters (the number of fertile stems, the grain number of the main stem
, the main stem grain weight, the single plant yield, the plot yield and th
e thousand grain weight) in a doubled haploid population of spring barley.
Several chromosomal locations with effects on more than one trait are found
. The method is also suitable for examining a single trait measured in diff
erent years or environments, and is used here to examine data on heading da
te, a highly heritable trait, and plot yield, a trait with moderate heritab
ility and showing QTL-environment interactions.