Genotype-Environment Interaction and Stability in Ten-Year
Height Growth of Norway Spruce Clones (Picea abies Karst.)
·By J. B. ST. CLAIR and J. Ki.EINSCHMIT
Niedersachsische Forstliche Versuchsanstalt, Abteilung Forstpflanzenziichtung, 35 13 Staufenberg 6, Ortsteil Escherode (Received 27th June 1985)
Abstract
Norway spruce cuttings of 40 clones were tested on seven
contrasting sites in northern Germany. Analysis of variance
for ten-year height growth indicate a highly significant
clone X site interaction. This interaction may be reduced
Silvae Genetica 35, 5-6
(1986
by selection of stable clones. Several measures of stability
were calculated and discussed. Characterization of sites by
the method of genetic correlation indicate that most of the
interaction is being generated between sites of high and
low elevation. Stratification of the area into two planting
177
zones based on elevation would also reduce the interaction.
of d tecting genotype-environment interactions and eva­
Whatever method is used, the costs involved must be com­
!uating genotypic stability.
pared with the increase in genetic gain:
Key words: genotype
X
Norway
spruce
has
proved
to
be
a
good
species
for
large-scale clonal tree improvement programs. Cuttings
site interaction, stability, Norway spruce
clones, height growth.
from
young
trees
root
easily
and. show
good
growth
and fox:zn KLEINSCHMIT et al., 1973; RouLUND, 1973, 1979).
Zusammenfassung
In addition, serial propagation appears to
in
Stecklinge von 40 Fichtenklonen wurden auf 7 unter­
maintaining juvenility
(ST.
CLAIR
be
et al.,
effective
1985). As
schiedlichen Standorten in Norddeutschland geprii.ft. Die
a result several European countries have initiated ope­
eine hochsignifikante Klan X Anbaustandort Interaktion
lection and propagation of clones from provenance . and
V.arianzanalyse fiir das Hohenwachstum im Alter 10 weist
rational clonal tree improvement programs using early se­
aus. Verschiedene Stabilitiitsma5e wurden errechnet und
progeny trials (KLEINSCHMIT and SCHMIDT, 1977; LEPISTti, 1977;
verglichen. Beschreibt man die Standorte mit Hilfe der ge­
RouwND, 1977;
netischen Korrelationen, so zeigt sich, daB die stiirkste In­
WERNER, 1977;
BENTZER, 1981;
MoNCHAUX,
teraktion zwischen Standorten unterschiedlicher Hohen­
1982).
zonen aufgrund der Hohenlage, wi.irde die Interaktion ganz
Lower Saxony Forest Research Institute, 40 clones were
As part of the clonal tree improvement program of the
lage auftritt. Eine Gliederung des Gebietes in zwei Anbau­
tested on seven contrasting sites in northern Federal Re­
wesentlich reduzieren. Ob man eine Gliederung des Anbau­
gebietes vornimmt oder Klone mit hoher StabiliUit aus­
public of Germany. This material provides an excellent
wiihlt, muB aufgrund der Kosten imVergleich zum Zuwachs
opportunity to assess genotype-environment interaction
im genetischen Gewinn entschieden werden.
and genotypic stability among clones used in an actual, on­
going tree improvement program. The objectives of this
Introduction
Genotype-environment
interaction
study are:
is
the
(1) to examine the magnitude of genotype-environment in­
differential
teraction in ten-year height growth in Norway spruce
response of genotypes to changing environmental condi­
clones;
tions. Such interactions complicate·tes:ting and selection in
(2) t.-J characterize these clones for stability;
(3) . · characterize sites in their contribution to the inter­
tree improvement programs, and result in reduced overall
genetic gains. The literature on genotype- environment in­
o
teractions is extensive. General reviews include CoMsTocK
and MoLL (1963), ALLARD and BRADSHAW (1974), and FREEMAN
environments;
(1973). Reviews within forest tree breeding include SQUIL­
(1970), SHELBOURNE (1972), SHELBOURNE and CAMPBELL
(5) to provide estimates of clonal repeatabilities and gene­
LACE
tic gains for a Norway spruce tree improvement pro­
(1976), MORGENSTERN (1982), and SKR0PPA (1984). MATHESON
gram.
and RAYMOND (1984) evaluated the importance of genotype­
environment interactions on Pinus radiata breeding programs in Australia using a criteria of the resulting loss of
potential gain.
Genotype-environment interactions can be diminished in
two ways: (1) by creating groups of essentially homoge­
neous environments and selecting cultivars suited to each
environment, and (2) by developing stable cultivars which
perform dependably over a range of environments. Stabi­
lity may be achieved by population buffering and by indi­
vidual buffering (AL
RD and BRADSHAW, 1964). Population
buffering involves creating varieties composed of different
genotypes adapted to a range of environments. It depends
upon the elimination of less fit genotypes from the stand
through intergenotypic competition. Individual buffering
implies stable performance of individual genotypes and
typically depends on heterozygosity, whic:Q implies a con­
tribution of non-additive gene effects (ALLARD, 1961; RowE
and ANDREW, 1964; ScoTT, 1967).
Evidence from crop plants indicate that selection for
stability may be effective (ScoTT, 1967). Non-additive gene
effects, however, will be lost in a cycle of sexual reproduc­
tion, but may be captured clonal selection and propagation.
Clonal tree improvement programs may be designed to
take full advantage of both individual and population buf­
fering. Besides capturing both the additive and non-addi­
tive gene effects associated with stability, the composition
and number of clones within clonal varieties can be chosen
to provide maximum population buffering. In addition,
clonal programs enable the development of varieties desig­
ned for smaller environmental units, something that is
prohibitively expensive for seed-orchard tree improvement
programs. Clones can also provide a more sensitive means
178
tion;·
(4) tv compare the merits· of these sites as general testing
Materials and Assessment
·
Propagation procedures and the breeding scheme of the
Norway spruce program of the Lower Saxony Forest Re­
search Institute have been described earlier in KLEINSCHMJT
et al. (1973); KLEINSCHMIT (1974), and KLEINSCHMIT and SCHMIDT
(1977). As part of this program, rooted cuttings are serially
propagated on a three-year· cycle. Selection based on nur­
sery and field performance occurs at each repropagation.
Cuttings of the 40 clones used in this study were tertiary
cuttmgs (third cycle of vegetative propagation) rooted in
spring 1974 and grown for three years in the nursery. The
clones originated from different provenances of outstand­
ing performance. The top provenance was Westerhof, thet'e­
fore se dlings of a tested stand of this provenance serve as
a base for comparison.. The clones were selected at age 4
due to growth potential and entered clonal tests. According
to the results of these clonal tests selection occured be­
tween each repropagation. Therefor these 40 clones are the
result of truncation selection. Clones are assumed not to
differ in maturation. During spring 1977, cuttings were
planted at seven contrasting sites in northern Germany
(Figure 1). Heights to the 1983 whorl were measured with
a height pole during summer 1984. Thus, height growth was
after ten growing seasons, ie. seven years after planting at
the test site.
The seven sites were chosen to represent the range of
sites in which Norway spruce clones from this program are
expected to be planted. Three sites are located in the low
coastal plains of northern Germany, while the other four
sites are located in country of higher elevation further
south (Table 1). The three low-elevation sites are warmer
and drier than the heigher-elevation sites. Medingen is the
...
c
Gl
I
0
f
._
...
....
:::>
c
u
...
..
c
"0
0
0
a>
...
...
.&:
....
i...
..
-
.&:
"'
... e
...
-
I
::ll
c. ...
.,..,
-
c
u
•
...
'i
"'
., ...
...
c
-
a
"'
:::>
"'
...
...
...
...
"'
e
...u
c
GIO
_...,
...
... ):
::>o
c:
.,
c
Gl
c
0
Q.
...
"0
c:
...
.,
Cll
a.
Q.
::>
...
Gl
c:
...
...
8
u
"0
c
...
"' 1...0
0
0
c. .r::.'
...._
- ...
:at..
"0
-
-'-
sg
-
-
=
0
0
.,
.,.
a
"'
>."'
"0 ...
c c:
... .,
vt C:
"0
"' c:
vt ...
"0
0
0
Q
t...
0
0
c.
c:
...
c:
...
g
t...
c:
"'
0
N
..,
0
..,
11'1
>.c.
s
a
_..
State of Lower Saxony
Federal Republic of Germany
Fig.
1. - Map of northern Federal Republic of Germany showing
location of tests sites as Indicated by •·
warmest and driest site, being further from the ocean than
Syke and Binnen. Lautenthal is located in the Harz moun­
tains and is the coldest and wettest site. Soil and nutrient
conditions were similar in the four higher-elevation sites,
c
0
"' but vary from a nutrient-rich, loamy soil at Syke, to a less
fertile site with a somewhat sandy soil at Medingen, to a
nutrient-poor, sandy site at Binnen.
The experimental design aLeach site was a randomized
20 blocks in single-tree plots. In addition to the 40 clones,
each block contained nine 3-0 seedlings from the provenance
Westerhof. Westerhof is considered an excellent prove­
I
c.
...
!
complete block design with 40 clones replicated in each of
I
..
.,_
c
.. ...
Cll
"'
\0
en
N
Q
t...
a> c
..
II
nance for planting in northern Germany (DIETRJcHsON et al.,
1976), and may be used to evaluate gains achieved so far
....
.....
\0
a
toO
....
-
11'1
.... ..,
.....
-
0
0
Cl)
0
0
t::n
....
....
..,
-
from clonal selection as compared to seedlings.
c:
0
"'
...
Statistical Analysis and Concepts
Analysis of variance was done for all sites combined·
using the statistical model
y 1 Jk 8 IJ + C)
where
Y
l
iJk
th
C
+
$J
+
C$ i j +
€. I jk
th effect of the i
S;
offoct of tho j
CS
• interaction bttween· the 1
•J
ftJk
th
ramet of thr
clone at s1te j
•
th
...
...
t
Q
c
-
"'
.
....
-
N
-
!
t...
Q
c
i
...
...
Cll
>.
11'1
aO
,....
0
...
....
....
""
,....
,....
0
.,.,
clone
sito
th
•
..
...
:::>
..
Gl
"'
...
• ten• year height _growth of the k
• avera 1 T mean .
(1)
u
0
clone and tl'lt j
tn
s1t
trror term.
0
11'1
Block effects within sites were not considered in this part
of the study, owing to the unbalanced nature resulting
from dead or missing trees. The form for the analysis of
variance is given in Table 2. All effects were treated as
random. Estimated variance components were used to cal­
culate phenotypic variance, broad-sense heritability, re­
.,
l
r.;;
..
...
c:
.,
"'
c:
-
i
c:
..
c
c:
;;:;
0
....
..,
..,.
..
....
...
....
...
..c
...
c
.,
...
::>
...
peatability of clonal means, and genetic gain. Formulas
USed include (see SHELBOURNE and THUUN, 1974):
179
Table 2. -Analysis of variance format for ten-year height of o::lones
wlth sites combined (assuming fully random model).
ource
expected ••en
dr
c-1
lone'
t t es
a;
a;
s-1
l on e x s1 te
(c-1 )(s-1)
rror
cs(t-1)
t
t .
•
t s
•
toZ
cs
and PERKINS, 1971).
squa r es
t st
0
FINLAY and WILKINSON used the estimated regression coef­
ficient b; (where b; estimates 1 + {3;) to measure stability
+ ct
and relative adaptability. A variety with a value near one
was considered to be of average stability and
adapted to good and poor sites. A variety With
Note: c - number of clones; s - number of blocks; t - effective
number ramets per clone per site; "'
among sltes; a•
sites; o1
A
••
A
qually
a
value
greater than unity was of low stability and better adapted
oZ•
differences among clones; "'
dom, and the sum of squares due to deviations from the
regressions with (c - 1) (t - 2) degrees of freedom (F'REEMAN
- variance due to
- varian e 'due to differences
- variance due to Interaction of clones and
- within-site variance.
to good sites. A variety with a value less than unity was of
high stability and better adapted to poor sites. Ea.unHARn
and RussELL (1966) proposed the use of an additional stabili­
ty parameter, the mean squares deviations from the :regres­
sion for each genotype, as measured by s di; a stable variety
was defined as one with bi
(2)
=
1.0 and St,Ji
=
0.
The use of the mean of all genotypes as the environmen­
tal index has been criticized on statistical grounds (FREE­
MAN and PERKINs, 1971; HARDW IC K and Wooo, 1972). However,
provided that the numbers of genotypes and environments
are reasonably large and the environmental range is suf­
ficient, linear regression using the mean of all genotypes
should be biologically valid (FRIPP and C.nEN, 1971; FIIIPP,
1972; HOHN, 1980; SK R0PPA 1983).
,
where
a:, 0:.
and
a:
o-:
a;.
h!
are estimated variance components
• phenotypic variance
The technique of FINLAY and WILKINSON (1963) and EilER­
(Table 2)
HART and RussELL (1966) is used in this study to characterize
genotypic stability. In addition, WRICKE's (1962) ecovalence
• phenotypic variance of clonal means
and Sn from HCnN (1979) are calculated. Ecovalence is the
• broad-sense heritabi 1 ity
tt
R
•
repeatabil fty of clonal. means
•
number of sites.
•
mean number ramets per c 1 one per s 1 te
contribution of each genotype to the interaction sums of
i (Yu - Y;.- Y.i + Y . . . )!, a
squares, and is given by
low value indicating greater stability. Sn is the average
change in rank between all pairs of env-ironments for geno­
Genetic gain was calculated in two ways (FALCONEn, 1981).
First, gain is calculated as
6.
G
inj;- R,::-where i is the in­
=
tensity of selection. In this study, it is assumed to equal
1.596 which corresponds to selection of five clones out of
40 (BEcKER, 1984). Second, the selection differential D is de­
termined from the difference of the mean of the top five
clones and the overall mean. Genetic gain is then calculated
as 6. G
=
D
·
R,;:
As the clone X site variance component increases rela­
tive to clonal variance, the repeatability of clonal means
decreases, and thus, genetic gains are reduced. One solution
is to select clones with high stability. Several statistical
techniques have been proposed to characterize stability.
type i, that is Sli
wh•re
1
•
B ll
J
(3)
.., GlJ
departure of the 1 inear regression coefficient of the
th
i
clone from the overall 1 1near regression coefft:.
cient
environmental index of ttlt:> j
J
th
site
devlations from the regress1on 1 ine of the
clone at the j
th
then
then be rewritten as:
yiJk
..
Ij becomes the site effect SJ and the model can
+ c, + (1
+
13,} SJ
+
i,J +
EiJk
It is one of six
The contribution of environments in generating interac­
(1977). He considers heights in different environmemts as
separate
traits
between
heights
at
selection
at
site
selection
be
to
and
one
calculates
pairs
used
one
to
with
1981).
indicate
another
(in
genetic
correlations
environments.
planting
formula
(F.-\LCOSER,
the
of
at
Gain
another
for
is
de­
for correlated response
Estimated
which
terms
of
correlations
sites
are
showing
most
least
to
can
simi­
inter­
action), and a matrix of expected gains can indicate which
sites are best for testing.
The formula for expected gain at site y based on clonal
selection at site X can be written as:
where
If the environmental index is taken to be the mean yield of all genotypes
in that environment,
.
tions was studied using a method developed by EluRooN
tn
site.
rii• I
sli.
lar
J
-
by Hen:-.. A genotype which ranks similarly in different
then
J
r;i
environments is considered stable and has a low value for
analysis, regresses the yield of each genotype upon some
and RussELL, 1966; PERKINS and JI NKs, 1968; SKROPPA, 1984). In
I
stability parameters based on rank Changes as developed
environmental index (FINLEY and WILKINSON, 1963; EBE R UA IIT
terms of the model presented above:
2i -;: i•
n (n-1)
termined from the
The most common approach, known as joint regression
CS
,,
=
;
.AG
•
Y·"
•
I
rR-"
e...
0"
t'V
r
(4)
•w
intensity of selection
repeatability of clonal means at site
rfcy
r
,,
11
.. clonal component of variance at sitr y
•
correlation between clonal means at sl tes
x
and y.
Variance components and repeatability of clonal means
were estimated from analyses of variance done separately
for each site. The statistical model considered effec:ts due
The interaction sums of squares can then be partitioned
to clone, block and error. Clone X block interaction was
into two components, the sum of squares due to differences
confounded with the error term since single-tree subclas­
in individual regression lines with (c - 1) degrees of free-
ses were used. All twenty blocks were included.
180
C"'lcul ations were
COIT.pl':er
carried out in Gl!ttlngen on the UNIVAC-
Table :i.
1110 of the "Gesellschaft fUr wissenschaftllche Daten-
-
clones at seven sites.
ve:arbeitung" and on the TA 1600/30 of the Lower Saxony Forest
Resea,;:ch Institute. using SPSS, Harvey and own programs.
ource
df
lones
39
Result and Discussion
I tes
Overall means and analysis of variance one
All effects, including clone X
site interaction, where
highly statistically significant (p < 0.001) in the overall
•••
Note:
4524
o.4o
ma9nltude
o.o48
9.o0***
6.52
2o9.91
92 .29•••
40
I
n.o
o.3o5
l.Bo***
o.12
234
vert anee comoonP.nts
F
HS
6
site
x
ttor '
ot
Analysis of variance for ten-year height (meters)
B2.o
o.o19
S.o
0.399
Indicates highly significant, p < .001; variance compo­
nent percentage is from total excluding error variance.
analysis of variance (Table 3). Site means decreased with
increasing elevation and decreasing temperatures (Table 4).
uverall clonal means ranged from 1.66 meters for Clone
11l9 to 2.69 meters for Clone 37. Clone 37 was particularly
Svke was the too-ranked site with a mean ten-vear height
of nearly three meters. 2.3 times better than the bottom­
outstanding, ranking first at five out of seven sites
ranked site, Lautenthal. Differences between the three
low-elevation sites may be related to soil structure and
'l'he significant clone X site interaction indiicates that
,.Jones perform ·differently between sites. Of greater i'l' ·
utrient availability. 'J'aate 4.
-
Mean 10-year heights for 40 clones and seedlings or provenance Westerhof at seven sites (In meters).
Syke
clo:-:e me11n rank
Hedtr: en
mean
rank
Btnnen
mea1 r
k
Pac!erborn
KattenbUhl
me11n rank
mean .rank
l!olzmlnden
me n rank
Lautenthal
mean rank
overall
mean rank
37
G
2.56
1
2.5o
1
2.16
1
1.81
1
2.69
1
3.71
1
3.29
2
2.77
2.37
3
2.46
123
3.52
6
3.36
1
3.o4
2
2
2.12
2
1.55
2
2.63
2
95
3.51
7
2.91
6
3.cl
3
2.22
8
2.29
5
1.66 15
1.44
7
2.43
3
41
3.52
5
3.o2
4
2.87
4
2.11 15
1.89 23
1.97
3
1.44
8
2.4o
4
l.o9 33
2.35
5
3.05'
1
2.2o
9
1.98 16
1.71
lo
1o7
3.59
2
2.82
9
·--------------------------------------------------------------------------------- ----------------------143
lo3
152
188
______
----
3.o2
3.49
3.55
3.2o
! l---
2o
8
3
1o
2.51
2.62
2.86
3.o2
2o
14
a
3
2.83
2.59
2.15
2.22
--- ---- z --! ---
5
8
19
13
2.26
2.o1
2.41
2.28
§ ---l---
125
2.78 27
2•67 13
2.3G
45
3.2o 11
2.46 24
2.44
0:6
3.19 12
2.86
7
2.15
3.lo 14
2.92
5
2.4o
lo1
46
3.37
9
2.61 15
2.18
--- ------------- --------- ------- ----
-
£§
7
23
2
5
___
£
2.44
2.3
2.12
2.1o
!
____
1o
2.o8 16
9
l. Z. 2
18
2.c8 17
1o
1.96 25
16
1. 79 34
----------------
3
4
1
a
1.86
1.75
1.65
1.62
f--f ----! 1z
5
9
11
23
__
1.35
1.33
1.23
1.47
------l
13
16
2.32
2.:lo
2.28
2.27
t5
4
l- !z
-
9o
2.91 21
2.5o 22
2.oo 27
1.98 Z4
1.S8 15
1.84
6
87
3.o9 15
2.68 12
2.o6
23
1.89 28
1.93 14
1.69 12
So
3.15 13
2.57 18
2.21
14
2.o4 Z:�
1.72 34
1.69 13
18
2.9o 22
2.71 11
2.o9
2o
2.o2 21
1.74 33
1.62 2o
8
1.81
1.89 22
15
42
2.o2 :-:2
33
3.o9 17
l.7e
2.58
- - - - - ------------------ --------- ------ - - - - - ------------------------ ------- -----
-
--! -
___
1.47
6
2.o6 1o
1.87
4
1.96 18
1.81
7
l.So
3
IE.
1.31
2.c4 11
1.66 16
1.25 22
1.39 24
1.46 32
1.15 31
2.26
6
1.43 34
------------------ -------- ------ - - - 1.38
1.2o
6
7
a
9
2.19 11
2.Hl 12
2.18 13
2.14 14
2.11
15·
--------
-
-
12
2.o8 16
7.8
2.oa 17
LoS 35
2.c6 18
1.35 14
2.o6 1g
2.o5 2o
1.19 29
---------------------
-
1.23 26
2.o4 21
142
2.83 24
2.38 25
2.3;, 12
2.o7 18
1.75 32
1.63 19
9
2.79 26
2.o3 22
2.31 31
z.o4 24
2.35
4
1.85 27
1.59 25
1.29 19
4
3.o9 1
2.21 34
2.c
21
1.82 32
2.o6
9
1.62 2o
1.26 2o
Z.vZ 23
1.38 11
2.o2 24
145 3.o4 18
2.27 32
1.ee 37
2.11 14
1.94 19
1.7o 11
1.34
15
2.o1 25
15
2.Gg 31
2.5?. 16
1.91 3o
2.o4 19
2.o1 13
1.47 31
------------------------ -----------------------------------------------------------------------
-
------
88
ll!J
115
113
2.54
z. 75
2. 5
z.ro
;;
23
35
25
98
2.75
116
3.o3
lo4
2.48
2.48
196
181___2.69
28
19
37
38
32
2.34
2.47
2.33
2.51
______ 21--- z -- ----
-
--
i73 2:88--2j
2.15
1.1!6
1.91
1.47
1--!2... !!
2.12
2.45
2.13
2.o5
2.34
__
__
23
23
:!o
21
37
25
35
38
29
17
32
28
4o
2.13
1.8o
2.17
2.16
22
26
29
15
35
1.96
1.71
1.84
1.88
1.83
-- f---
2.o6
2.o2
1.91
2.19
1.76
13
33
1o
11
1.93
1.86
2.o3
1.97
2o
26
12
17
!1 .. ! ----! Z -'26
37
Jo
29
31
1.88
1.77
1.92
1.78
1.69
----l
25
31
21
29
37
c 1 'ma 1 means
!Ud11ng
��eans
--!1
1.56
1.33
1.62
1.45
1.47
1.25
1.47
1.24
l.ft3
24
18
27
36
______
26
33
22
33
2
2:j5--2;---i:8;--;i·--i:;;--;;----i:6;--;8·---i:;;-- 34
1.69 36
1.58 38
36
1.68
2.22 33
66
2.60 33
1.78 Jo
1.56 39
1,64 39
2.o1 39
2.57 34
112
1..55 39
1.73 36
2.o3 25
2.16 35
2.19 .¢o
11
4o 1.77 34
1.53
4o
1.3o
2.42 !
1.86 4o
12
----------------------------------- - - - - -------------- - -----------O\lerall
1.61
1.64
1.51
1.37
2.97
2.54
2.18
2.24
2.o9
1.84
1.32
1.35
1.24
!
1.99
1.98
1.96
1.96
2-- !
l.oJ
1.o2
1.17
1.21
___
37
o.86
1.43
1.15
o.92
lo
27
! 21--
---!
4o
9
32
9
-- 1.- !--g ------! g -- !
-
26
27
28
29
31
32
33
34
1.91
1.9o
1.87
1.86
l3
!!£ --
______
39
37
4o
23
5
24
1o
--
1..83
36
1..7
37
38
!!
--1 -
1.7,
1.69
___
39
2.c2
1.95
1.63
1.28
2.o9
1.74
2.22
1.41
l.o8
1.81
181
0.65 t o l .46 (Table 6). The plot of clonal means aga:lnst site
clOnal mean, y;;
(ml
means with the corresponding regression lines for four
clones illustrates the differential reactions of clones to
4.0
changing environments (Figure 2). Clone 18 repre!sents a
.....
18
clone of average stability (b1
=
0.99) as defines by FiNLAY
and WILKINSON (1963). Its performance is relatively equal
3.6
on poor and good sites. Clone 189 represents a clone of high
3.4
stability (b1
=
0.65 ). It performs relatively better ,on poor
sites (although its overall performance is poor). Clone 107
l2
represents a clone of low stability (b1
=
1.46). It p erforms
3.0
relatively better on good sites.
2B
performance. In forest tree breeding this information is
The point is, the regression coefficient measures relative
useful to distinguish genotypes for specific environments,
2.6
but if all environments tested are in one planting zc1ne, and
each represents the same proportion of area to be planted,
2.4
2.2
2.0
····
.... ·
1.8
•
1.6
...·
t4
.. ··
·
.. ··
..··
·
····
....
.. ··
,.... .
. ····
. ·
·
··
then this information is irrelevant. Selection on the overall
·
mean is all that is necessary to assure the largest overall
gains.
·· ·
More important to forest tree breeding is the predictabi­
... ·
lity of yield of
a
genotype in various environments. This
concept of stability may be measured by the mean devi­
ations from the regression line, S%di (BEcKER, 1981). Clone
.
107 performs well overall, but is very unstable as measured
3T
..............._
lt)1
-·
II
·-•
........ . ... . .. . 11!9
..
--•
-
t2
tO
by Stdi (Table 6). Clone 37 also performs well, but is very
stable as measured by stdi· This is apparent from the de­
viations of the individual points from the regressi.on line
in rigure 2.
t6
1.4
Lautenthal
Fig. 2.
-
I
1.8
2.0
2.6
I
2.8
30
i
site mean.9.1(m)
Syke
���fl
Clone 37 rates as very stable, and Clone 107 as unstable.
Rank correlation between ecovalence and S21 is high, and
Paderborn
Plot of clonal me ans for four clones against site means
with corresponding regression lines.
terest than statistical significance is the importance of the
interaction in reducing gains. The magnitude of the inter­
action component was less than half that of the clonal
variance component (Table 3).
Broad sense heritability
The conclusions are the same when the other s;tability
parameters, ecovalence and Sn, are considered (Table 6).
(hZg)
overall for 10-year height
is still good between Sn and S2ui• and Su and ecovalence
(Table 7).
A plot of the stability parameter against the clonal means
is useful as an aid to selection (Figure 3). In each figure,
clones falling in the lower, right-hand side are p:refered.
Clone 37 is the best clone no matter which stability para­
meter is used. Selection of clones based on height s1nd sta­
bility can then proceed using a selection index for multiple­
growth was 0.10, which is relatively low. However, repeata­
trait selection (SoNECYPHER and ARBEz, 1976), but assigning
bility of clonal means (Rc) overall was 0.89. The genetic
weights based on relative economic value may prove dif­
gain based on a selection intensity of i
=
1.596 is 0.33 m, a
ficult. This would require knowledge of the reduction in
gain of 16 percent above the overall mean. If the top five
gain associated with using unstable versus stabl1
clones are selected and genetic gain calculated based on
types. The use of independent culling levels, that is, setting
the observed selection differential, the genetic gain is 0.42
an acceptable value for a stability parameter followed by
m, 20% above the overall mean. The difference between
selection based on height, may prove more practical.
the two gain estimates results from the outstanding height
growth of clones 37 and 123. These gain estimates repre­
sent gains to be achieved from further clonal seiection, and
do not include gains already achieved from past selection.
The F-value for heterogeneity of regressions from the
breakdown of the interaction sums of squares confirms the
statistical significance of the clone X site interaction (Ta­
ble 5). Values for the regression coefficient b1 range from
Table 5.
- Breakdown of interaction sums of squares.
Source
heterogeneity of
regressions
deviations from
regressions
182
•••
p
Correlated gains among sites
The expected gains from various combinations of testing
and planting sites, along with estimates of clonal variance
and repeatability of clonal means are presented in Table B.
Expected gains at planting sites are generally greatest or
Stability
Note:
geno­
< .001: •• < .01.
df
MS
F
39
1.42
3.56***
195
o.58
1.44**
close to the greatest, when selection is done at the same
site. Where this is not true (Paderborn, Kattenbi.:lhl, and
Holzminden), the clonal variance is small relative to that
at Syke, Medingen, and Binnen. Selection based on overall
means results in gains second only to those when selection
is done at the same site as planting. When selection is done
at a single site for planting at all sites, the best sites for
testing, as indicated by the largest gains, were ME!dingen,
Syke and Binnen. These three sites were of higher site in­
dex, and had the largest value for repeatability of clonal
means and the greatest clonal variances.
Medingen and Syke were the best single sites for testing,
largely because of the high correlations with most of the
Table 6.
egress ton
coefficient bi
clone
37
1.16
1.22
1.28
-
rank
Stability statistics for 10-year height growth.
mean square
rank
dev. S
ecovalence
S
u
rank
rank
l
12
8
5
6
o.o1
o.2o
o.28
0.31
1
31
35
37
88
0.92
1.21
1.34
1.14
24
9
3
15
o.36
!).2o
o.12
o.12
38
3o
21
2o
125
45 0.80
o.97
33
2o
1
o.o5
o.14
o.os
23
5
o.JJ
o.ls
o.o9
11
14
7
8.8
11
s
19
31
6
1.19
-9
11
18
o. o 7
0.9o
1o
12
o.13
o.o9
12
5
1o 9
7.9
3o
13
142
9
4
145
o.94
0.86
o.98
0.88
23
3o
19
26
o.lo
o. 17
o.17
o.Jo
16
27
28
36
o.ll
o.2o
o.I7
0.32
lo
21
16
31
7.9
9.4
lo s
11 9
13
24
28
35
88
118
115
113
o.77
o.79
o.7B
o.87
36
34
35
2t.l
o.o8
o.o9
o.l4
o. So
11
14
22
4o
o.IB
o.17
o.24
o.SJ
19
18
26
39
9.1
11.2
ll.o
14.1
22
33
31
39
98
116
lo4
196
o.92
1.21
o.71
o.74
25
1o
39
38
o.o9
o.o2
o.os
o.1o
13
2
6
17
o.1o
o.u
o.2o
o.23
8
9
2o
25
6.8
9.o
7.3
9.1
8
2o
9
22
17 3
66
112
11
189
1.15
o.76
o.81
o.82
o.65
14
37
32
31
4o
o. ol
o.Ui
o.1o
o.l6
0.16
3
o. o7
o.27
o.17
o.22
o.39
3
27
17
24
35
7.3
9.8
4.4
5 .5
8.3
9
26
5
7
16
123
95
1:
i·
------------ --------------!-----143
to3
152
:2!________!
·
--------
------------ ------2 !--------
o' o6
2
1.6
o Jo
3o
1 6
1
36
o 43
4 1
4
.
37
0.44 .
8.2
15
4o
o.8o
11.9
35
2___----------------------------------------
:
:
o.37
34
8.s
17
o.28
29
7.5
11
o.34
32
1o.6
29
o.15
7.9
15
13
0.48
38
12.6
37
--------------------------------------------
7
::
i·
:
!:: ::
_!§________ ! : 2____________!______ !2________ ; t
:
: :: :---------:: : ·----------:---------:: i----- !
:
------------ 2____________ !?______2 l1-------- ---------2 !---------- --------!2 ----·- Z-
1
9
8
5 0
_! --------2;
:
----------- ?______2;1 --------! ---------2;!!__________! ---------2; ----..
-
_2!________2 2!___________g ------2 !!________ 1---------2 t ----------g --------! ;t ____.. §-
!§!________2;2 ----------- !______2;21_________!_________2
25
!5
24
26
other sites (Table 9). Medingen shows high correlations
·
with Syke and Binnen, yet still shows good correlations
with the high-elevation sites. The estimated gain for selec­
tion at Medingen and planting at Paderborn, KattenbUhl,
Holzminden and Lautenthal was in three cases greater
than for selection at either Syke or Binnen. Medingen may
be considered intermediate between the low and high­
elevaticn sites. The least similar sites are Lautentha
and
Binnen as indicated by a low correlation of clonal means
and low gains from selection and planting between sites. Table
7.
-
Rank correlation coefficients between stability para­
v,
bl
s,
ecoval nee
0. 69
s,
ecovaltnce
StJ
0. 2 7
o. Zo
-o. o2
2o
0.14
0.04
0. 88
0. 43
o.
Comparison of clones and seedlings
Estimates of gains from further clonal selection have
been g1ven. Gains achieved from past clonal selE!ction can
be assessed by comparison of seedlings and cle nes. It is
. assumed that seedlings used for comparison are represen­
tative of populations from which clones were initially se­
lected, and that these seedlings would be used au planting stock if clones were not used. The mean ten-year height growth of seedlings was 1.81
meters as compared to a mean for clones of 2.09 meters
(Table 4). This represents a gain of 15 percent fmm clonal
meters.
bI
-----------!_________1;2------ -
0.
43
selection and planting. Clones grew better than seedlings
at all test sites except KattenbUhl. All differen.ces were
highly significant owing to the large number of degrees of
freedom. The reason seedlings were superior at Katten­
bUhl
is unclear.
Clones and seedlings
may have been
treated differently at planting since survival of clones was
much worse than survival of seedlings whereas at all other
sites survival was about equal for clones and seedlings.
183
deviations
from regressions
Conclusion
.
A statistically significant interaction in ten-year height
growth exists between Norway spruce clones and their test
.60
,
.50
sites in northern Germany. This interaction may b1 reduced
-au
clonal-
by dividing the region into two planting zones. This would
result in an increase in gain from clonal selection, within
each planting zone, but also involves an increaSE! in COStS
from having two clonal programs as opposed to one.
'l
'i'
.30
. 20
I.S
••
.
-a ... It
Tt •: 'f2
.
tification of planting zones should proceed b1!lsecl on ele­
moan
deviation
from '-str•ssiona
i ... .";2
..
2D
vation. Whether one or two planting zones are used, the
best sites for testing within each zone may be identified
using the method of genetic correlations, taking into ac­
'"' ""
Ill "'
T•
18
generated between sites of high and low elevation. If the
increase in overall gain justifies the increase in costs, stra­
...
" "'
t6
•
•
•
m.P.
0
..
"
.•,
11!1•
·
.10
Characterization of sites by the method of genetic cor­
relations indicates that most of the interaction is being
2.2
2.4
2.6
'.1
count the heritabilities at the respective sites. The. next
2.8 clonal rMar\Y.Iml
step would be to identify environmental characteristics
that determine a good testing site.
Other factors besides height growth may be important to..
•covalence
decisions of stratification of planting regions. The Norway
spruce clonal program has actually been divided into two
...
.eo
zones above and below 300 meters. The two zones s1re based
-all
clonal moan
.70
on differences in crown architecture of trees from different
and the
elevations
relationship
to
snowbreak
damage.
Such information would not be available from the! present
.60
study for many years. Other factors important to the stra­
tification
'1'
.so
.40
'I
'I'
.30
••
.20
t
112
•
.10
'1
'I.
•
.J-
•""
2.0
1.8
nes that are stable as. defined by the stability pa1rameten.
ecowail'nce
moan
•
but may involve a reduction in gain when clones excluded
.
due to instability were superior in height. The gain from
a reduction in clone :.< site interaction must be compared
'.l
2.2
2.4
2.6
with the loss from excluding clones with high growth po­
2.8
clonal
mean. Y,.frrj
tential.
Identification of specific environmental conditions in­
of clones suited to specific environments. Such a fine­
::::' ...-
15.0
from a seed orchard approach to tree improvement. Within
a clonal program, it would involve considerable record
T
..•
J'
5.0
30
42e
15...
''Ill' "' :zo ·
"U''
7.0
'i',,
•
11S·· . ...
11.0
u.f"
keeping and the increase in costs would need to be weighed
'l
genotype-environment interactions, economics should be
considered.
\2
Acknowledgements
......... of
Ule
Sn>I.UA for establishing
and maintaining the trials, A. GEHRMA""' for help In as,sessment,
puter analysis. R. BuKooN and H.
criticisms of the manuscript.
•
21
•
•
WnLc�<OOIII' provided
valuable
Literature Cited
Au.ARI>, R. W.: RelationshiP between genetic diversity and con­
sistency of performance In different environments. Crllp Sci.
173:17
•
•
1.8
The authors gratefully acknowledge J.
H. HAASE for preparing the figures, and B. SEF.I.MANN for the com­
1.0
Fig.
against possible gains. Whatever approach is used to handle
'"w>
,a-
II
•
'll
tuning of clones and environments would not be poss ible
..
•
':}
llO
184
s <li• ecovalence, and Sn. Selection for stability within a
planting zone involves no increase in cost to the program,
volved in generating interactions can lead to identification
s.
9.0
from
fering. Individual buffering can be used by selecting clo­
... !l) '
'1' sa
....
0
'i
damage
of clones within a variety assures some population buf­
i'
,_'l
zones may include
nal varieties within planting zones. A sufficient number
'l
...
planting
Interactions may also be reduced by selecting stable clo­
....
•
'i'
,.,
1
'i'
of
disease and insects. wind and acid rain.
2.0
3.- Plot of s•
22
di
2.4
2.6
127-133, (1961).
-
1:
ALLARD, R. W. and A. D. BRADSHAw: Implications
of genotype-environmental Interactions in applied plant breeding.
2.8
clonal
mean y,_(m)
' ecovalence and s against clonal means.
1
Crop. Sci. 4: 503-- 508, (1964).
-
BECKER, H.
c.:
Correlations among
some statistical measures of phenotypic stability. Euplilytlca
835-840, (1981).
Forth
edition.
-
30:
BECKER, w. A.: Manual of Quantitative Genetics.
Academic
Enterprises.
Pullman,
WA,
(1984).
-
Table 8. - Expected genetic gains from clones planted at site y after resting at site x (In meters) Intensity of selec­
tion assumed to equal 1.596. Estimated clonal components .of variance and repeatabil!tles of clonal means for e ,et1
X with
site given In margin. Gain for selection at site
Planting
si te
Syke
i
':>.490
0.382
M ed ngen
Binnen
rn
Paderbo
KattenbUhl
n
-
-- --- ---
0.339
0.252
x
Latrtcn thaI
HOl zmi nden
Kat tenbuhl
0.286
0.269
overa 11
elec t o
i n
s
oi
..,,,
-·
0.188
0.439
o.ll2
o . o9
1
0.431
0 . 288
0.274
0 . 258
0.259
o.22o
o.4o5
o.35o
o.537
0 . 255
0.293
o.3o3
0.192
0.464
,:>.166
o.l95
0.149
,0.197
0.192
0.179
0.171
0.253
0.16 4
o . l 3 1l
o.28o
o.l8o
0.168
0 . 13
0.228
0 . 0 38
. 152
o . 58
0.155
o . l 44
0.
0.256
0 .1§5
l
. 262
1
0 . 77
;)(
sites assumes each environment represents
n
t:; stjn9 s_i_tc
0.354
___
Over 11
planting
aderoc:n
tslor.;en
,;J, 0 2
---,- - - - ­ .2;!!?
Lautentha 1
all
o.31!o
,
Holzminden
-
N a 1 ng e
-r<e
y
planti.ng at
equal proportion of area to be pla ted.
·- -- ··-­
··-
••
!?;!? L .!?;22L
o.Z61
0.247
o . 8o
0.8 2
___
••
1 48
!?;!?L
••
___
1
0.256
0.168
!?;} §
____
--
- m
;
0.044
o.213
____
o.o33
-- !! L .
o . 2 o7
0.221
0 . 22 5
0 . 182
0.329
o.56
o.7o
o.78
0.84
0.89
•
':?;':?£
o. o48
e
Table 9. - Correlation coefficients b tween clonal means for ten-year height growth among sites.
I
b
.._
Medingen
k
y e
Binnen
o. 7o
o.Bo
L:din'Jcn
Paderborn
o.£6
l'l\nnen
Kattenblihl
o . 58
0.64
o.7o
0 . 64
0.53
o . 63
IKattenblih 1
scale
Large
B.:
spruce (Ptcea
of Norway
propagation
netlca
G TI'etlc correlation as a concept for studying genotype-environ­
ment Interaction in forest tree breeding. S!lvae Genet!ca Z&: 168-­
R.
CoMsTocK,
-
175, (1977).
E. and R. H. MoLL:
Genotype-envi­
ronment Interactions. p. 164-196. In: w. D. H�.,.soN and
H.
F. Ro·
982,
,
A. UE JAMBLIN,.E, P. KRcTzscH, A. K!lstc, R. LINES, S. MAc:<ESEN
D.
to
Introduction
S.:
Longman. London and New York,
edition.
W. and G. N. WILKI:<so:<:
programme.
breeding
genetics.
quantitative
(1981).
-
Agric. Res.
14:
742-754, (1963).
components
15-23, (1971).
uctions.
Heredity
(1973)·
31: 339-354.
Genotype-environmental
-
FRIPP,
interactions
Y.
in
J.
and
C.
Coil.
For.,
Swed.
environment Interactions
;(
Unlv.
on
selection
commun:
Eplcea
P.:
87-95,
The impa1Ct of geno·
Australian
14:
pp.
Sci.,
Agr.
RA\'MO>m:
-
MoN ­
pepin!erc.
en
massale
radiata
Pinus
11-25, (1984).
-
-
- X
-
14,
18 p.,
PERKI;.;s, J. M. and J. L. JINKs: Environmental and geno­
Ro<'LUND,
-
n.:
The ef­
-
Ronl'ND,
Tree Impr. Arbor .• Harsholm, No. 5:
H.:
Vegetative
propagation
of forest
trees at the arboretum in Hersholm, Denmark. In: Prc,c. vegeta­
tive Propagation of Forest Trees-Physiology and Pra,=tice. Inst.
For. Imp. and Dep. For. Gen., Col!. For., Swed. Univ. Agr. Sci.•
pp. 103-128, (1977).
RocLt·No, H.: Stem form of cuttings related
-
to age and position of scions. (Picea abies L. KAIIST.). For. Tree
Impr. Arbor., Hersholm, No. 13, 24 p., (1979).
E.
-
Row•, P. R. and
ST.
CLAm,
R. H. ANDREw: Phenotypic stability for a systematic serices of corn
Schizophyllum
genotypes.
I. Analysis and character. Heredity 27: 393-407, (1971).
'
Fnu••·· Y. J.: Genotype-environment interactions in Schizophyl­
commune.
-
Gen.,
MATHESON, A. C. and C. A.
21-41, (1973).
tistlcnl methods for the analysis of genotype-environment lnter­
C.\'I'EN:
-
way spruce cuttings. For.
-
FREEMAN, J. H.: Sta­
-
For.
aNd crosses. Heredity 23: 339-356, (1968).
of variability. VIII. Relations between
27:
LEI;IsT!l, M.: Vegetative Propagation
fect of cyclophysis and topophysis on the rooting ability of Nor­
K.
genotypes grown In different environments and measures of these
environments. Heredity
-
type-environmental components of variability. III. Multiple lines
-
FnuBIAN, J. H. and J. M. PERKIN<: Environmental and genotype-en­
vironmental
Dep.
(1982).
The analysis of adaptation in a plant­
Aust. J.
197-203, (1977).
cultural treatment. Can. For. Serv. Inf. Rep. PI
Second
FINJ.AY,
0.61
MoRCENSTEns, E. K.: Interactions between genotype, sitE! and sllvi­
EBERHART, S. A. and w. A. Rt:ssELL: stability
parameters for comparing varieties. Crop. Sci. 6: 36-40, (1966).
FAI.cos•n,
0. 57
AFOCEL, Annales de Recherches Sylvlcoles 1982: 262-2811, (1982).
A.
on Norway spruce (Ptcea abies (L.) K,\RST.). Norwegian For. Res.
-
26:
(1977).
CIIAU:-<,
NA"s os and B. V1ss: The. IUFRO provenance experiment of 1964/68
Inst., 14 p., (1976).
o.48
of Forest Trees - Physiology and Practice. Inst. For. Imp. and
type
DIETRICHSON, J., C. CHRISTOPHE, J. F. CoLES,
-
(1963).
o.61
breeding programmes. Aust. For. Res.
mNsoN (Eds.); Statistical Genetics and Plant Breeding. Nat. Acad.
Sci. PUb!.
0 . 37
and problems connected with large-scale application. Sllvae Ge­
BURDON, R. D.:
-
0.59
0 . 52
Experiences with Picea abies cuttings propagation In Germany
abies (L.) KARST.) by cuttings. In: Symp. on Clonal Forestry. Swed.
Un!v. Agr. Research Notes 32, pp, 33-42, (1981).
o.4o
o . 61
[Ho 1 zmi n::len
BENTzEN,
o . 57
0.61
0.59
l?ed r orn
Lautentkal
Holzminden
Crop. Sci.
4:
563-567,
(1984).
-
J.
B.,
J.
serial propagation of
KLEixscHMIT, and J. SvoLaA: Juvenility and
iu.m commune. II. Assessing the environment. Heredity 28: 223-238,
Norway spruce clones (Ptcea abies KARST.). Sl!vae Genet. 34: 42-48,
studying genqtype-environment Interactions. Heredity 28: 209-222,
(19'72). - Hen,., M.: 13eitriige zur· Erfassung der phllnotypischen
Crop. sci.
(1972).
-
HARDWICK, R. C. and J. T. WooD: Regression methods for
(1985).
(1979).
- Hem;,
M.:
Beitrllge
Medlzin und Blologle
10:
ge'W'ogener
zugsbasis.
EDV
Government
Kl.EissCIIMIT. J .• W.
der
Medlz!n
MCLLrn, J.
Steck!lngsvermehrung
Prn:-;:isreife.
und
Biologic
11:
113-120, (1980).
ScmuoT und J. RAcz:
von
Sllvae Genetica 22:
Flchte
4-15.
(Picea
(1973).
abies
-
-
Entwicklung
KARST.)
zur
KLEINSCHMIT, J.:
A program for large-scale cutting propagation of Norway spruce.
N.Z.J. For. Sci.
: 359-366. (1974).
-
Kttl;.;scll>llT, J. and J. ScHMIDT:
E.: Selecting for stability of yield In maize.
549-551. (1967).
-
SHELBOL'RNE,
c.
J. A.: Genotype­
IUFRO Genetics-SABRAO Joint
Forest
Experiment
Station.
Tokyo,
symposium.
1972,
(1972).
-
SnEtoot·RsE, C. J. A. and I. J. THuuN: Early results !rom a clonal
zur Erfassung der phiinotyplschen
selection and testing programme with radlata pine. N.Z. J. For.
Sci. 4: 387-398, (1974).
Regress!onsmodelle sow!e .Anderungen der Be­
In
7:
tree Improvement.
112-117,
Stabllltat. II. ModlfikatiDn des Regressionsansatzes durch verwen­
dung
ScoTT, I.
environment Interaction: Its study and Its !mplicatloml in forest
Stnbil!Uit. I. Vorschlag elnlger auf Rang!nformation beruhenden
Stabllit!ltsparameter. EDV In
-
·
-
SHEtaovR"•· C. J. A. and R. K.
CAMPBELL:
The impact of genotype environment Interactions on tree impro­
vement strategy. IUFRO Joint Meeting on Advanced Generation
Breeding, Bordeaux, June 1976, (1976).
-
SKR0PPA, T.: A critical
evaluation of methods available to estimate the genotype X en­
vironment Interaction. Stud!a Forestalia Suecica
-
Sot'ILUCE,
166:
3--14, (1984).
A. C.: Field experiment on the kinds and sizes of
185
genotype-environment interaction. In: Papers 2nd Meeting, Work­
tion by cuttings of Picea abies In Sweden. In: Proc. V<egetatlve
1969, p.
Propagation of Forest Trees - Physiology and Practice. lnst. For.
MethodS Of Selec­
Imp. and Dep. For. Gen., Col!. For., Swed., Univ. Agric. Sci., pp.
ing r.roull on Quantitative Genetics. IUFRO, Raleigh, N.
49-61, (1970).
-
STONECYPHER,
R. and M.
AREBZ!
c.
tion, IUFRO Joint Meeting on Advanced Generation Breeding,
Bordeaux, June 1976, (1976). - WERNBft, M.: Vegetative propaga­
97-102, (1977).
- WR!CKE, G.: 'Ober elne Methode zur E',rfassung
der 6kologlschen Streubreite
tung
C7:
92-96, (1962).
in
Feldversuchen. z. Pfianzell%ilch­