Grain yield was studied in a collection of 220 Nordic barley lines at
diverse locations in the Nordic countries. Two-row (2r) and six-row (6
r) lines differed very significantly in reaction to the growing condit
ions within and between the two locations, Svalov (in southern Sweden)
and Hojbakkegard (in Denmark). This difference was also highly signif
icant at Viikki (in Finland), but not at As (in Norway) or between Vii
kki and As. Genotype x location (GL) and genotype x year(GY) variance
components were used to estimate phenotypic yield stability by Shukla'
s stability variance (sigma(i)(2)). Only 7 lines did not contribute si
gnificantly to GL- and GY-interactions, and their yield levels were 7-
27% lower than that of the highest yielding line (5057 kg/ha). Estimat
es of GL- and GY-stability parameters were not significantly correlate
d. Neither responsiveness, measured by the regression coefficient (b(i
)), nor phenotypic yield stability, measured by the deviations from re
gression (Tai's lambda(i)) were correlated with yield. Pedigree studie
s showed that both b(i) and sigma(i)(2) can be changed by recombinatio
n and/or induced mutations. Mixing of near isogenic lines with differe
nt resistance genes, and selection within a landrace, also resulted in
changes in responsiveness. Recently released 2r-cultivars were more u
nstable than older 2r-cultivars revealed by positive correlation betwe
en the year of release and lambda(i). Cultivars originating from south
ern Scandinavia were higher yielding than cultivars originating from t
he central or the northern regions of Scandinavia.