Potentielle forklarende variabler for udbytte i forskellige miljøer Hans Pinnschmidt Danmarks...
28
Potentielle forklarende variabler for udbytte i forskellige miljøer Hans Pinnschmidt Danmarks JorgbrugsForskning Afdeling for Plantebeskyttelse Forskning Center Flakkebjerg 4200 Slagelse [email protected]
Potentielle forklarende variabler for udbytte i forskellige miljøer Hans Pinnschmidt Danmarks JorgbrugsForskning Afdeling for Plantebeskyttelse Forskning
Potentielle forklarende variabler for udbytte i forskellige
miljer Hans Pinnschmidt Danmarks JorgbrugsForskning Afdeling for
Plantebeskyttelse Forskning Center Flakkebjerg 4200 Slagelse
[email protected]
Slide 2
Background BAROF WP1 data: multivariate measurements on 86
spring barley genotypes in 10 environments (2 years: 2002 &
2003, 3 sites: Flakkebjerg, Foulum, Jyndevad, 2 production systems:
ecological & conventional). [email protected]
Slide 3
variables: yield 1000 grain weight grain protein contents culm
length date of emergence growth duration mildew severity rust
severity scald severity net blotch severity disease diversity weed
cover broken panicles & culms lodging parameters: raw data
mean/median/max./min. rank/relative values main effects interaction
slopes raw data adjusted for E/G main effects/slopes (residuals)
IPCA scores SD/variance factors: genotypeenvironment G 1 E 1....E
j. G i variables: X 1(i,j)... X m(i,j) parameters X m(i)1... X
m(i)p X m(j)1... X m(j)p } derive information on general
properties, specificity, stability/variability
[email protected]
Slide 4
Objectives Multivariate characterisation of genotypes with
emphasis on yield-related properties.
[email protected]
Slide 5
Statistical methods Non-linear Canonical Correlation Analysis
(NCCA): an optimal scaling procedure suited for handling
multivariate data of any kind of scaling (numerical/quantitative,
ordinal, nominal). Multiple Regression Analysis (MRA)
[email protected]
Slide 6
Non-linear Canonical Correlation Analysis (NCCA) data
treatment: quantitative variables (v m ) were converted into
ordinal variables with n categories (v 11... v 1n,..., v m1... v mn
). [email protected]
Slide 7
Characterisation of environments based on data adjusted for G
main effects (= residuals) [email protected]
Slide 8
Flakkebjerg 2003: high yield, net blotch & panicle
breakage; low mildew & lodging Flakkebjerg 2002: high rust
& 1000 grain weight; late sowing Foulum 2002 conventional &
Jyndevad 2003 ecological: high mildew & lodging; low yield %
net blotch Jyndevad 2002 ecological: low yield, 1000 grain weight,
weed infestation, protein content [email protected]
Slide 9
Characterisation of genotypes based on data adjusted for E main
effects (= residuals) [email protected]
Slide 10
dimension 1 (sq. root) dimension 5 (sq. root) high yield &
1000 grain weight; low protein content & lodging low yield
& 1000 grain weight high mildew; low net blotch & disease
diversity low mildew [email protected]
Slide 11
Characterisation of genotypes in individual environments based
on: actual yield data disease main effects (ME) of Gs environmental
disease variability (SD) of Gs (= standard deviation of E adjusted
data) [email protected]
Slide 12
Flakkebjerg 2003: high yield, net blotch & panicle
breakage; low mildew & lodging [email protected]
Slide 13
dimension 4 (sq. root) dimension 6 (sq. root) high yield; low
net blotch ME & SD short straw high rust ME & SD long straw
low yield; high net blotch ME & SD Flakkebjerg 2003: high
yield, net blotch & panicle breakage; low mildew &
lodging
Slide 14
Flakkebjerg 2003: high yield, net blotch & panicle
breakage; low mildew & lodging Foulum 2002 conventional &
Jyndevad 2003 ecological: high mildew & lodging; low yield
& net blotch Jyndevad 2002 ecological: low yield, 1000 grain
weight, weed infestation, protein content
[email protected]
Slide 15
dimension 1 (sq. root) dimension 5 (sq. root) low yield; high
mildew & net blotch ME & SD low mildew ME & SD high
yield Jyndevad 2003 ecological: high mildew & lodging; low
yield & net blotch
Slide 16
Multiple Regression Analysis (MRA) dependent variables: yield
(actual, E-adj. G mean & SD) independent variables: E-adj. G
mean & SD of disease severity, weed infestation, growth
duration, culm length criteria: Pin/out = 0.05/0.10; Fin/out =
3,84/2.71; tolerance = 0.0001 [email protected]
Slide 17
Variables must pass both tolerance and minimum tolerance tests
in order to enter and remain in a regression equation. Tolerance is
the proportion of the variance of a variable in the equation that
is not accounted for by other independent variables in the
equation. The minimum tolerance of a variable not in the equation
is the smallest tolerance any variable already in the equation
would have if the variable being considered were included in the
analysis. If a variable passes the tolerance criteria, it is
eligible for inclusion based on the method in effect.
Slide 18
Mean versus standard deviation of environment-adjusted yield of
spring barley genotypes; BAROF 2002-2003
[email protected]
Slide 19
Slide 20
Observed versus estimated mean environment-adjusted yield of
spring barley genotypes; BAROF 2002-2003
[email protected]
Slide 21
Slide 22
Observed versus estimated standard deviation of environment-
adjusted yield of spring barley genotypes; BAROF 2002-2003
[email protected]
Slide 23
Yield of spring barley genotypes versus main effect yield of
the environment; BAROF 2002-2003 [email protected]
Slide 24
Slide 25
Yield of spring barley genotypes estimated based on yield main
effect of environment and E-adjusted mean & standard deviation
of genotype property variables (disease severity, weed infestation,
culm length, growth duration); analysis across environments; BAROF
2002-2003 [email protected]
Slide 26
Slide 27
Yield of spring barley genotypes estimated based on E-adjusted
mean & standard deviation of genotype property variables
(disease severity, weed infestation, culm length, growth duration);
analysis by environment; BAROF 2002-2003
[email protected]
Slide 28
Conclusions & outlook NCCA: intuitive method good for
visualising the main features in multivariate data of various
scales useful for obtaining an overall synoptic orientation of G
properties and E characteristics soft systems approach MRA: hard
systems approach synoptic view neglected Mildew & net blotch
had highest yield-related effect, although not always functional
(especially in MRA!) [email protected]