New Forests 12: 187-202, 1996.
© 1996 Kluwer Academic Publishers.
Printed in the Netherlands.
Effects of length of seed chilling period and sowing date
on family performance and genetic variances of
Douglas-fir seedlings in the nursery
FRANK C. SORENSEN1
1 Forestry Sciences Laboratory, Pacific Northwest Research Station, 3200 SW Jefferson Way,
Corvallis, OR 97331, U.S.A.
Received
16 January 1996; accepted in revised form 9 April 1996
Key words: time of emergence, date of bud set, date of bud burst, growing season length,
seedling height, nursery testing, genetic testing, Pseudotsuga menziesii
Application. Date of sowing and length of chilling period had a predictable effect on date
of emergence, seedling phenology, and seedling size; interacted with families; and had an
unpredictable effect on family and within-plot components of variance. To produce vigorous
seedlings early sowing and long seed chilling are recommended. Because families interacted
with treatments, it is important for genetic evaluations in the nursery that sowing date and
chilling period be as consistent as possible from year to year.
Abstract. Seeds of four full-sibling Douglas-fir families (F) were moist chilled (C) for 14,
33, and 77 days and sown (S) March 29, April 26, and May 24 at two densities (D = 111 and
200 seeds/m2), grown for 2 years in nursery beds and phenology and size traits recorded. The
study was analyzed in two parts: part I evaluated seed treatment effects and their interactions
with families; and part II investigated the effect of treatments on genetic variances, particularly
among-family (u�) and within-plot (u�) components and the intraclass correlation for families
C and S and
F for all traits. Early S combined with long C resulted in early emergence and gave
large seedlings with little loss and damage. Many interactions between C and F, and S and
F, were significant. Interactions involved rank changes for size but not for phenology traits,
and were larger for C x F than for S x F. Seedling density affected seedling size but not
(t1). In part I there were large and highly significant differences associated with
among
phenology, did not interact with seed treatments, and interacted significantly but weakly with
families. In part II,
C and S, but not D, had significant effects on
u�, u�, and
t1, but not in
a predictable manner. Because of significant interactions, it is recommended that standardized
seed treatments be used in family nursery tests. This should aid in keeping the results from
these tests as repeatable as possible. Long chilling and sowing as early as practicable are
recommended to minimize disease losses and winter damages and to provide good nursery
stock.
Introduction
Several studies have indicated, both in agricultural crops (Black and
Wilkinson 1963; Alessi and Power 1975; Gray 1976; Cook 1980) and
trees (Skeates 1986; Jenkinson and McCain 1993), including Douglas-fir
(Pseudotsuga menziesii (Mirb.) Franco) (Sorensen 1978), that date of seedling
emergence influences seedling size, as well as phenology (Dormling 1973;
188
Sorensen 1978) and survival or seedling yield (Cook 1980; Miller 1987;
Jenkinson and McCain 1993). Two factors, length of growing season and
synchrony between the developmental cycle of the plant and the annual
climatic cycle, may be important. In addition to direct effects on first-year
growth, there may be carry over effects into the subsequent year or years
(Heide 1977; Sorensen 1978; Schmidt-Vogt 1995). In Douglas-fir (Sorensen
1978) and Norway spruce (Picea abies (L.) Karst.) (Dormling 1973), prove­
nances responded differently to date of sowing, and in Norway spruce to year
of sowing (Schmidt-Vogt 1995). In annual crops, highly significant variety x
year and variety x year x location interactions have been observed (Robin­
son and Moll 1959). These and other observations point out the intimate
connection between the climatic and photoperiodic regime and the phasic
developmental cycle of the plant (Rasumov 1930; Watson and Baptiste 1938;
Hammerton 1975; Lawn 1979). If families react differently to these factors,
it could influence genetic and environmental variances and genetic rankings
in nursery evaluations (Johnson and Frey 1967).
Time of emergence can be affected by date of sowing and by length of time
that seeds are moist chilled or stratified prior to sowing. Chilling increases
both rate and uniformity of germination, and the effect is inversely related to
the incubation temperature; i.e., the lower the substrate temperature within
the germinable range, the more chilling increases the speed and uniformity of
germination (Weber and Sorensen 1990; Sorensen 1991). Therefore, depend­
ing on substrate temperature, genetic variances for time of emergence and for
nursery growth and phenology could be affected by sowing date and chilling
pretreatment.
The objective of the present test was to investigate the effect of length
of chilling period and date of sowing on genetic evaluation (early testing)
in nurserybed trials (e.g., Robinson and van Buijtenen 1979, Bastien and
Roman-Amat 1990). Particular emphasis was placed on family x sowing
date and family x chilling period interactions. The test was designed so
that the effect on among- and within-family variances could be analyzed.
Two sowing densities were used to determine if competition influenced the
magnitude of the effects, particularly effects carrying into the second year
(Fowler 1984; Magnussen 1989; Mazer and Schick 1991; Miller et al. 1994).
Materials and methods
F amities (F)
Four full-sib and unrelated families were randomly chosen from about 30
control-pollinated crosses made in a natural second-growth stand in the
189
foothills of the central Oregon Cascade Range (44°35 ' N, 122°42' W, 300
m). Number of entries was restricted to four, because of the many treat­
ments applied. Families had not been tested previously for any germination
characteristics. Nursery location was 44°341 N, 123°271 W, 75 m (Corvallis,
Oregon, U. S.A.). All seeds were x-rayed, and only filled, apparently healthy
seeds were used in the tests.
Chilling periods (C), dates of sowing (S) and densities (D)
Chilling periods (14, 33, and 77 days) were spaced to be linear on the loga­
rithmic scale (Sorensen 1980, 1983). These periods span the range normally
used for Douglas-fir seeds. Fourteen days is marginally adequate for germi­
nation at 25 °C, but longer periods are necessary for complete germination at
lower temperatures (Allen 1960; Sorensen 1991). Seeds were soaked at room
temperature (ca. 22 °C) in aerated water for 24 hours, then the water drained,
and seeds placed in a cooler at 2-3 °C for the designated periods. Seeds were
not treated for mold during long stratification, but none formed on them.
Dates of sowing were March 29, April 26, and May 24 (28-day intervals).
March 29, the first S, is a little later than the mean emergence date of autumn­
sown seeds in our nursery, which was March 11 and March 22 in two previous
tests (Sorensen 1990; Sorensen and Campbell 1993), but the dates span the
normal times of outdoor sowing for the area. On designated sowing days,
seeds were surface dried, sown in small depressions in raised nursery beds,
and covered with a thin layer (ca. 0.5 cm) of granite grit. Our usual cultural
practices were followed: watering and weeding was done by hand as needed
with last watering and fertilization in mid-August; balanced soluble fertilizer
(20-20-20) was applied every 4 weeks at rate of 28 kg/h from time of seedcoat
shed (year 1) or budfiush (year 2) to mid-August.
Sowing densities were 111 (10 cm x 9 cm) and 200 (7 cm x 6.5 cm)
seeds/m2 • Based on visual impression, these D would not differentially affect
plant growth through shading until after bud flush in the second year.
Design and analyses
Experimental design included two blocks within each of two nursery beds.
As described below, beds and blocks were treated as four replications in one
analysis (see "Part I" below), whereas the beds were treated as the only two
replications in the second analyses (Part II).
Part!
Experimental design was split-split-split-plot. Density treatments, D1 and
D2 , were assigned to whole plots randomized within blocks. Sowing dates,
190
Si. S2 , and S3, were assigned to subplots randomized within whole plots.
Chilling times, C1 , C2 , and C3, were assigned to sub-subplots randomized
within subplots. Finally, families, F1 , F2 , F3, and F4, were assigned to sub-sub­
subplots randomized within sub-subplots. Each F was represented initially
by an 18-seed noncontiguous plot.
The test was conducted in raised nursery beds. There were two border
rows between the test seedlings and the edges of the beds and eight border
rows on each end of the test. Sub-subplots (different C sown on the same
date) were separated by one buffer row; subplots (different S) were separated
by two buffer rows, one row sown at each date; and D were separated by
six buffer rows, three at each density. Split-plot design was used primarily to
reduce the amount of between-plot buffering that would have been necessary
without the splits.
For the presentation of the results of this analysis, variability was appor­
tioned based on the percentage of total sums of squares (%SS) explained by the
different sources of variation (Hicks 1982; Hiihn et al. 1987). When compared
with equivalent variance components, residual %SS is usually underestimated
so that the proportion of variation explained by other terms should be consid­
ered maximal (Hicks 1982). Comparison based on %SS seemed suitable for
this material, because the traits represented different stages of development
of the same seedlings and all analyses used the same format.
Part Ila
The second part of the analysis had the purpose of investigating the effect of
treatments, particularly the C and S treatments, on genetic variances. Each
of the 18 sub-subplots (all combinations of 2 D, 3 C, and 3 S) was analyzed
as a "minitest" of the 4 F, each F represented initially by an 18-tree plot, and
replicated twice within a nursery bed. The format of the analysis of variance
of a single minitest is shown in Table 1.
Part /lb
In each of the 18 minitests, components of variance were determined for
family, within-plot, block, and experimental error (block x family), and
these components entered into a second analysis of variance. The goal of
the analysis primarily was to determine if C and S were systematically or
predictably influencing genetic variances (a} and a� in Table 1) and the
intraclass correlation [tf, or a}!(aF +a�)], and if D altered the influence of
C and S.
Seed treatment effects on the variance components for block and exper­
imental error were small and are not presented. The analysis of variance
format, which is split-split-plot, is given in Table 2. Plot coefficients of vari­
ation and skewnesses also were analyzed according to the format of Table 2.
191
Format of analyses of variance of Part Ila "minitests" - components of variance
estimated for blocks (u1), families (u}), experimental error (u1p) and within plot
(u�) among 4 families within each of the 18 combinations of sowing density, date of
sowing, and chilling period. Variance components from the minitests are then analyzed
according to the format in Table 2.
Table I.
are
Sources of variation
Degrees of freedom2
Expected mean squares 1
Total
Blocks
Families
BxF (experimental error)
Within plot error
bfn-1
(b-1)
(f-1)
(b-l)(f-1)
bf(n-1)
u�+n u1F +nf u1
u�+n u1F +nb u}
u�+n u1F
0"2
w
1 u2 are variance components for blocks (B), families (F) and interaction.
2 b = 2; f =4; n =seedlings per plot= 18 initially, later harmonic mean of survivors.
Table 2. Part Ilb format for analyses of variance components derived from the minitests
(e.g., analysis of u} 's and other components of variance obtained from each of the
tests analyzed according to the format of Table l).
Sources of variation
Degrees of freedom2
Expected mean squares 1
Total
Blocks (B)
Densities (D)
Error (a)
beds-I
(b-1)
(d-1)
(b-1) (d-1)
u� +eds u1
u� +bes Ob
0"2
Sowing date (S)
SxD
Error (b)
(s-1)
(s-1) (d-1)
d(b-1) (s-1)
ut+bed 01
ut+be Oh
ut
Chilling period (C)
CxD
CxS
CxDxS
Error (c)
(c-1)
(c-1) (d-1)
(c-1) (s-1)
(c-1) (d-1) (s-1)
ds(b-1) (c-1)
u�+bds Ob
u�+bs Obv
u�+bd Obs
u�+bObvs
0"2
c
a
1 u2 is the variance component for blocks (B); 02 are fixed effects for density (D), date
of sowing (S), and chilling period (C).
2 b = 2; d = 2; s = 3; c = 3.
Presumably because of effective transformations (see below), neither added
information to that provided by analyzing variance components alone, and
they are not included in the results.
Traits
The following seed and seedling traits were recorded:
192
EMERG - days from sowing to emergence (seed coat or hypocotyl visible
above ground surface). Observations made every second day from day
of sowing.
BS1 and BS2 - dates of bud set, first and second year. Observations made
biweekly first year, weekly second year. Buds recorded as set when bud
scales first visible between the needles of the terminal shoot.
BB2-date of bud burst. Observations made every second day. Buds recorded
as flushed when green needles first visible between the scales of the
terminal bud. EMERG, BS1 and 2, and BB2 recorded in Julian calendar
dates.
GSLl and GSL2 - first- and second-year growing season lengths for the
terminal shoot in days (BSl minus EMERG in year 1, and BS2 minus
BB2 in year 2).
H1 and H2 - heights in centimeters from ground level to base of terminal
bud at end of first and second year.
HI2 - height increment in centimeters during year 2 (H2 minus Hl).
D2 - diameter in millimeters below cotyledons at the end of the second
growing season.
All measurements were tested before analysis for the need for trans­
formation, so that variance would be independent of the mean. Where
transformation was necessary, log transformation was satisfactory.
Results
Establishment
Of the 5 184 seed spots sown, 5 036 (97.1 %) emerged. Of these 53 were
obvious mutants and deleted from further observation, leaving 4 984 (96.1 %)
emerged usuable seedlings. At the end of the first year, 4 866 (93.9%) were
alive and healthy and 4 788 (92.4%) were present at the end of year 2.
Percentage of emergence by sowing date (S l is earliest) was 98.7 (S l ), 96.0
(S2), and 96.7% (S3) and by chilling period (Cl is shortest) 93.2 (Cl ), 98.7
(C2), and 99.5% (C3). Family emergence ranged from 96.0 to 98.1%.
Mutant seedlings were predominantly curly needle phenotype (35 of 53),
which we typically find at low frequencies in Douglas-fir beds, particularly
if the seeds have been stored for several years as these had. It occurred in
193
all families, but frequencies differed among them (x2 = 10.99; d.f. 2; p
0.012). Other mutants were fused cotyledons (8 seedlings, 7 in one family)
and unclassified (10 mutants).
First-year mortality was 2.3% and most of it (1.9%) was due to late-season
damping off. S1 had less mortality than S2 and S3 (0.5% vs 2.6 and 2.6%;
x2 27.48; d.f. 2; p < 0.001). Longer C also had less mortality (1.3 and
1.4%) than Cl (3.0%) (x2 16.23; d.f. = 2; p < 0.001). The differences were
small but they were consistent with other observations that early emergence
decreases incidence of damage from soil-borne pathogens (Bloomberg 1973;
Jenkinson and McCain 1993).
=
=
=
=
=
Seedling phenology and growth
S and C strongly and significantly influenced seedling phenology and size. The
effects on several traits and a comparison with family differences are shown
in Figure 1. Early S and long C affected all traits in the same direction, except
for EMERG (early S increased, long C decreased time to emerge) and H2/D2
(early S increased the size of the ratio, C had no effect). Relative importance of
main effects and interactions, given as a percentage of total sums of squares, is
shown in Table 3. The results were consistent with expectations and in general
agreement with other reports where comparable traits have been measured.
One exception is the ratio, H2/D2. Early S, compared with late S, increased
the ratio in this study and decreased it in a previous one (Sorensen 1978).
Some consequences and unexpected aspects of the results are pointed out
below.
Seeds sown later did not emerge as rapidly as expected based on temper­
ature. After long chilling (C3), the averages were 22, 18, and 18 days for
S1, S2, and S3 seeds to emerge, respectively. A recording 8-day mechanical
thermometer was in the nursery area. Air temperatures were summed in
degree-hours (0C) above 4.4 °C from S to mean date of emergence for the C3
treatment only. Approximate degree-hours from S to EMERG were 3 500,
4 100 and 4 800 for S1, S2, and S3, respectively. The reason for the increase
in degree-hours to emergence with advance of the season was not clear.
Because increase of autumn frost resistance is associated with bud
development (Campbell and Sorensen 1973), delayed BSl can influence
susceptibility to early frosts. A frost of -5 °C on October 29 damaged
terminal needles on the main shoot of 2.05% of the seedlings. The frequency
of damage was not large enough to test with analysis of variance, but treat­
ment main effects (D, S, C, and F) were tested with chi-square analyses.
All main effects, except D, were highly significant, with the percentage of
frost-damaged seedlings increasing with later sowing and decreasing with
long chilling. Family differences depended on S. With early sowing (S1),
Table 3. Sums of squares and statistical significances for sources of variation in Part I analyses
in seedling size and phenology traits (Table I) given
in percentages of total sums of squares.
Sources of variation
EMERG 1
BSI
Sowing date (S)
Chilling period (P)
11.3***2
40.5***
7.7***
BS2
_3
BB2
GSLI
GSL2
HI
H2
HI2
D2
H2/D2
30.4***
3.5***
68.3***
6.1 ***
10.2***
55.3***
7.8***
26.7***
5.4***
5.2***
5.3***
31.6***
30.4***
30.9***
Density (D)
-
Family (F)
4.9***
38.4***
67.6***
27.3***
5.3*
13.l***
4.2***
9.2*
14.8***
65.0***
12.8***
34.9***
44.4***
SxF
CxF
0.8***
2.2***
0.3***
0.4***
2.3***
l.7***
0.9**
l.7***
0.8***
0.6**
0.4***
0.7***
0.9***
1.0***
0.5***
1.0***
0.3*
l.7***
-
0.4 *
-
l.6**
0.9***
-
0.6**
0.4***
DxF
SxC
SxD
37.0***
-
-
-
-
-
1.3***
-
-
1.0***
0.8***
-
-
-
2.0***
-
-
-
0.9**
0.3**
0.8 **
l.9***
0.3*
0.7**
4.6***
0.4 *
0.7 ***
0.9**
l.7 ***
29.0**
4.9***
0.7**
2.0***
0.8***
1.2**
CxD
SxCxF
-
-
-
SxDxF
0.8**
CxDxF
SxCxD
SxCxDxF
Percent SS explained4
Residual
58.5
88.1
72.5
65.9
91.3
79.5
1.7
2.7
10.7
8.4
2.1
6.2
78.4
2.3
76.1
3.3
74.7
4.3
81.8
68.6
7.3
4.9
1 Traits are, EMERG, days from sowing to seedling emergence; BS, date
of bud set; BB, date of bud burst; GSL, growing season length of terminal
shoot in days; H, total height; HI, height increment; D, diameter below cotyledons; HID, height diameter
ratio. Numbers I and 2 after the trait
abbreviations refer to first and second year in the nursery bed.
2 Significance, *,p < 0.05; **,p < 0.01; ***,p < 0.001.
3
Value not given if mean square is not significant.
4 Percentage of
total sum of squares explained by significant terms.
.......
\0
..;...
195
Cl
c
"i
�
40
E 30
g
Figure 1. Comparative effects of sowing date (S1
March 29, S2 April 26, S3 May 24)
and seed chilling period (C l , C2, and C3
14, 33, and 77 days, respectively) and family
(F l through F4 four biparental crosses) on several seedling phenological and size traits:
EMERG, days from sowing to seedling emergence; BS and BB, date of bud set and bud burst;
GSL, growing season length for terminal shoot extension; H and D, height and diameter; and
FROST, percentage of seedlings frost damaged in first autumn. A number 1 or 2 after trait
symbols refers to 1 or 2 years after sowing. If the effect was not significant, the overall mean
value was given to all levels of the treatment (e.g., trait BS 2).
=
=
=
=
=
percentage of frost-damaged seedlings was 0.1 % and there was no difference
among F (x2 = 3.00; d.f. = 3; p = 0.392); after S3, the range among F was
zero to 10. 2% damaged seedlings, and family variance was highly significant
<x2 60.48; p < 0.001).
In addition to the direct effect of date of emergence on BS1 timing, there
was a carryover influence on date of BB2. Seed treatments that gave seedlings
=
196
with early BS1 gave seedlings with late BB2. The regression relation was
close (BB2 132.7-0.149 BS l ; t 13.65; d.f. 7; p < 0.001); i.e., a 10-day
advance in BSl 1.5-day delay in BB2.
Early emergence and accompanying longer GSLl seemed to have a
disproportionately large affect on H l . For example, mean GSLl for S3
seedlings was 122.1 days; GSLl for S l seedlings was 153.3 days or 25.6%
longer. For S3 seedlings, Hl was 8.8 cm; H l for S l seedlings was 16.6 cm or
88.6% greater. Two factors could have contributed to the disproportionality.
First, part of GSLl was used in seedling establishment (radicle elongation,
cotyledon elongation and seed shed, and early needle elongation) before any
epicotyl extension begins. The S3-Sl difference in time allotted to epicotyl
extension probably was proportionately greater than was the difference in
GSL as I recorded it. Also, elongation tends to be exponential, with the result
that loss of GSL at the end of the extension period had a larger-than-equivalent
effect on H1. Second, relative to S1 seedlings, growing season of S3 seedlings
continued late into the autumn when conditions for shoot extension were less
favorable.
Seedling form was analyzed as the second-year height/diameter ratio.
Density and sowing date had strong effects (Table 3). On average, stockier
seedlings were in less dense plots and in plots whose seeds had emerged
later. In an earlier study (Sorensen 1978), early emerging, early budsetting
plants were stockier than late emerging plants. Because diameter increment
continues after bud set, the discrepancy presumably was due to differences
in dates of bud set and in conditions for growth after bud set between the two
studies. In any case, seedling quality and health were not positively related to
stockiness in this study, because the stockiest plants were the small seedlings
from late sowing.
Seedlings grown at wide spacing were significantly taller and had signifi­
cantly greater second-year height increment than seedlings grown at narrow
spacing. Since GSL2 did not differ with density, the greater increment at wide
spacing must have been due to greater rate of elongation.
=
=
=
=
Interactions between seed treatments and families
Many interactions between F and C and F and S were highly significant (Table
3). Of the interactions, C x F tended to be larger than S x F interactions,
particularly for size traits, even where S main effects were larger than C main
effects. Interactions for size traits, compared with phenology traits, tended
more often to involve rank changes or to have more potential for rank changes.
S x C x F interactions were highly significant for size traits, but insignificant
for GSLl and GSL2 (Table 3). As observed in connection with BS1 and frost
damage, however, even small interactions involving phenological traits can be
197
rather important if a damaging event is timed right. Density, or competition,
did not affect interactions between C and F, and S and F (Table 3, 3- and
4-way interactions involving D were not significant).
Family and within-family variances
Part II of the study investigated the effect of seed treatments on variance
components, specifically among-family (a°}) and within-plot (o-�) and on
the intraclass correlation (t1) for families. F-values from the analyses of the
components (see Table 2 for format) are presented in Table 4 for several
traits frequently measured in nursery genetic evaluations. The F-values show
that C and S do indeed affect genetic variances, but the significance of both
main effects and interactions differs greatly from trait to trait. With regard to
changes in variances, no pattern emerges to indicate that certain seed treat­
ments should be used for family nursery tests. The effect of seed treatments
on components of variance and intraclass correlations for H2 are shown in
Figure 2, which illustrates the lack of pattern.
Discussion
Both early S and long C promoted early emergence. In addition, long C
gave more rapid, complete, and uniform emergence. With earlier emergence,
first-year seedlings had a longer extension season, became taller, set buds
earlier, had less loss from nursery diseases, and escaped damage from a
late-October frost. In the second year, seedlings that had emerged early had
a shorter extension season, but greater overall shoot extension and larger
diameters. Clearly, largest and healthiest seedlings were associated with early
emergence, as also has been reported for survival (Miller 1987) and seedling
vigor in other species (references given in "Introduction").
The inverse relation in the second year between amount of extension
and extension period is probably due to early emerging first-year seedlings
setting buds early and developing more preformed growth (Tranquillini et
al. 1980). At the same time, the longer year-2 extension period of seedlings
that emerged late the first year suggested a compensation mechanism that
dampens early environmentally caused size differences. This is apparently
analogous to the seed-weight environmental effect on Douglas-fir seedling
size, which also was observed to decrease over time (Sorensen and Campbell
1993).
Long C is particularly important if the seeds are sown early when the
nursery soil is cold (McLemore 1969; Tanaka et al. 1986; Sorensen 1990;
Weber and Sorensen 1990; Barnett 1993). Compared to short C, long C
198
Table 4. Listing of F-values and their significances for sources of variation (date of sowing,
chilling period, and sowing density) that affected among-family and within-plot components of variance and the intraclass correlation coefficient for families in several seedling
size and phenology traits. A significant F-value means that the indicated source of variation influenced the among-family or within-plot components of variance or influenced the
intraclass correlation coefficient.
Sources of variation
BS11
BB2
BS2
HI
H2
D2
A. Among-family component of variance (a})
Date of sowing (S)
3.15
9.90**
5.71*2
0.54
7.96**
2.89
7.09**
2.30
0.73
4.45*
2.59
2.74
2.15
2.40
1.25
2.99*
2.35
1.10
0.29
3.67*
0.68
S x C
0.06
6.08**
S x D
4.34*
1.79
0.85
0.13
0.25
0.04
C x D
0.30
0.14
0.26
0.18
0.81
0.98
S x C x D
1.14
0.39
0.58
0.15
0.73
0.73
Chilling period (C)
Sowing density (D)
B. Within-plot component of variance
(a;)
Date of sowing (S)
4.56*
Chilling period (C)
0.45
0.70
3.86*
Sowing density (D)
0.99
12.51**
3.13
0.54
7.42*
S x C
1.65
0.45
0.62
16.93***
0.88
15.28***
10.00***
28.25***
0.33
3.08
3.11*
2.69
6.05**
0.55
6.68**
S x D
3.76*
1.51
0.28
0.18
0.96
3.61*
C x D
1.05
3.94
0.39
0.18
0.12
0.10
S x C x D
1.04
0.76
0.87
0.28
0.51
0.71
2.05
6.68**
5.57*
4.48*
3.85
5.68**
C. Intraclass correlation coefficient [t1 = a}/(u} +
5.18*
u;)]
Chilling period (C)
6.73**
5.47*
0.04
3.02
Sowing density (D)
0.03
0.00
0.68
S x C
2.64
0.77
5.66*
4.17*
2.76
0.15
4.59**
Date of sowing (S)
1
2.80
0.80
23.32***
11.38***
S x D
1.28
0.82
1.73
0.51
0.65
1.91
C x D
2.28
0.84
1.37
3.94*
1.85
4.11*
S x C x D
1.75
0.59
1.00
0.59
0.53
1.35
Traits are, BS 1 and BS2, dates of bud set, years 1 and 2; BB =date of bud burst; H1, H2
and D2, total heights or diameters, year 1 or year 2.
2 Significance of F-values, *, p < 0.05; **, p < 0.01; ***, p < 0.001.
both increases rate of emergence and reduces the variation among seeds in
emergence time. Long C is still beneficial but less important under conditions
when the substrate is warm.
Seed treatment x family interactions for height were similar to those
observed for provenances (Sorensen 1978). Different genetic entries appar­
ently differ in sensitivity to date of emergence. In the current family test, the
same family (Fl) showed the least difference in response to both S and C.
199
0.04
Oc1·
II C2
.C3
0.03
2
A
F
CJ
0.02
0.01
0.06 �-------�
B
0.05
0.04
2
w
CJ
0.03
0.02
0.01
o.s ..---,
0.5
t,
0.4
0.3
$1
$2
S3
(ai) and within-plot (a�) components of variance and intraclass
correlation coefficients [tf = ai/(ai + a�)] for families for second-year height (loge cm)
after 3 dates of sowing (SI = March 29, S2 = April 26, S3 = May 24) and 3 seed chilling
periods (Cl, C2, and C3 14, 33, and 77 days, respectively).
Figure 2. Among-family
=
200
Because of the effect of seed treatments on both rate and uniformity of
emergence, it was anticipated that varying S and C might alter the relation
between a} and a� in the seedling traits (Prout and Barker 1989; Mazer and
Schick 1991). Although both S and C did indeed influence a� , a� , and t1
(the intraclass correlation coefficient for families), they did so mostly in an
unpredictable way (e.g., Figure 2). The only apparent generality was that long
C caused the effect of S on a} and a � to be small (Figure 2, A and B).
Conclusions
The study was designed to investigate the effect of C, S, and D on family
evaluations in the nursery. Different C and S caused family rank changes for
size but not for phenological traits. Interactions were larger for C x F than for
S x F and smallest for D x F. C and S affected among-family and within-plot
variances, but not in a predictable manner, except that long C reduced the
effect of S on variances. Consistent C and S are needed for consistent genetic
evaluations and long C and early S are recommended because they promoted
early emergence, and gave the largest seedlings with least disease incidence
and least autumn frost damage. Independent observation in Weyerhaeuser
Company Douglas-fir nurseries support these recommendations (Yasuomi
Tanaka, 1995, personal communication).
Pacific Northwest seed plants and nurseries stratify Douglas-fir seeds up to
13 weeks. A caution with longer chilling is that seeds with low stratification
requirements can germinate during chilling (Tanaka 1976) if seeds have to
be held for any reason after an already long chilling. Outdoor nursery sowing
times are dictated by rainfall and frost. Early emerging (late February to early
March) seedlings from autumn sowing have been damaged by spring frost in
our beds. Sowing probably could be advanced 2-3 weeks prior to the March
29 date used in this study, but this will vary with nursery. The C3/S1 treatment
seemed optimal or nearly so for our conditions, but the results indicated that
longer chilling and earlier sowing both could be beneficial provided they
stayed with safe limits for the particular material and nursery (Jenkinson et
al. 1993).
Acknowledgments
Richard S. Miles maintained the test and made many of the measurements;
Nancy Mandel ran and summarized many analyses; Roger G. Petersen
reviewed the statistical procedures. Yasuomi Tanaka and three anonymous
referees provided helpful comments on an earlier draft. Their contributions
are gratefully acknowledged.
201
References
Alessi, J. and Power, J.F. 1975. Response of an early-maturing com hybrid to planting date
and population in the Northern Plains. Agron. J. 67: 762-765.
Allen, G.S. 1960. Factors affecting the viability and germination behavior of coniferous seed.
IV. Stratification period and incubation temperature, Pseudotsuga menziesii (Mirb.) Franco.
For. Chron. 36: 18-29.
Barnett, J.P. 1993. Presowing treatments affect shortleaf pine seed germination and seedling
development. Tree Planters' Notes 44: 58--62.
Bastien, J.Ch. and Roman-Amat, B. 1990. Predicting Douglas-fir (Pseudotsuga menziesii
(Mirb.) Franco) volume at age 15 with early trails. Silvae Genet. 39: 29-35.
Black, J.N. and Wilkinson, G.N. 1963. The role of time of emergence in determining the
growth of individual plants in swards of subterranean clover (Trifolium subterraneum L.).
Austral. J. Agric. Res. 14: 628--638.
Bloomberg, W.J. 1973. Fusarium root rot of Douglas-fir seedlings. Phytopathology 63: 337341.
Campbell, R.K. and Sorensen, F.C. 1973. Cold-acclimation in seedling Douglas-fir related to
phenology and provenance. Ecology 54: 1148-1151.
Cook, R.E. 1980. Germination and size-dependent mortality in Viola blanda. Oecologia 47:
115-117.
Dormling, I. 1973. Photoperiodic control of growth and growth-cessation in Norway spruce
seedlings. IUFRO Working Party 2.01.4, Symposium on Dormancy in Trees. K6rnik,
Poland, Sept. 5-9, 1973. 16 pp.
Fowler, N.L. 1984. The role of germination date, spatial arrangement, and neighbourhood
effects in competitive interactions in Unum. J. Ecol. 72: 307-318.
Gray, D. 1976. The effect of time to emergence on head weight and variation in head weight
at maturity in lettuce (Lactuca sativa). Ann . Appl. Biol. 82: 569-575.
Hammerton, J.L. 1975. Experiments with Cyperus rotundus L. III. Seasonal variations in
growth. Weed Res. 15: 339-348.
Heide, 0. M. 1977. Regulering av vekst of kvile hos ulike 0kotyper av gran. (Control of growth
and dormancy in Norway spruce ecotypes.), pp. 1-10. In: Dormling, I. and Eriksson, G.
(Eds) Experimental Genekologi. Department of Forest Genetics, Royal College of Forestry,
Stockholm. Research Note 27. (Swedish with English summary and captions.)
Hicks, C.R. 1982. Fundamental Concepts in the Design of Experiments. 3rd edition. Holt,
Rinehart and Winston. New York.
HUhn, M., Kleinschmit, J., Svolba, J. 1987. Some experimental results concerning age depen­
dency of different components of variance in testing Norway spruce (Picea abies [L.]
Karst.) clones. Silvae Genet. 36: 68-71.
Jenkinson, J.L. and McCain, A.H. 1993. Winter sewings produce 1--0 sugar pine planting stock
in the Sierra Nevada. USDA Forest Serv., Pacific SW Res. Sta., Berkeley, Calif., Res. Pap.
PSW-RP-219. 10 p.
Jenkinson, J.L., Nelson, J.A., Huddleston, M.E. 1993. Improving planting stock quality-the
Rumbold experience. USDA Forest Serv., Pacific SW Res. Sta., Berkeley, Calif., Gen.
Tech. Rep. PSW-GTR-143. 219 p.
Johnson, G.R. and Frey, K.J. 1967. Heritabilities of quantitative attributes of oats (Avena sp.)
at varying levels of environmental stress. Crop Sci. 7: 43-46.
Lawn, R.J. 1979. Agronomic studies on Vigna spp. in south-eastern Queensland. I. Phenological
response of cultivars to sowing date. Austral. J. Agric. Res. 30: 855-870.
Magnussen, S. 1989. Effects and adjustments of competition bias in progeny trials with single­
tree plots. For. Sci. 35: 532-547.
Mazer, S.J. and Schick, C.T. 1991. Constancy of population parameters for life-history and
floral traits in Raphanus sativus L. II. Effects of planting density on phenotype and heri­
tability estimates. Evolution 43: 1888-1907.
202
McLemore, B.F. 1969. Long stratification hastens germination of loblolly pine seed at low
temperatures. J. For. 67: 419-420.
Miller, T.E. 1987. Effects of emergence time on survival and growth in an early old-field plant
community. Oecologia 72: 272-278.
Miller, T.E., Winn, A.A. and Schemske, D.W. 1994. The effects of density and spatial distri­
bution on selection for emergence time in Prune/la vulgaris (Lamiaceae). Am. J. Bot. 81:
1-6.
Prout, T. and Barker, J.S.F. 1989. Ecological aspects of the heritability of body size in
Drosophila buzzatii. Genetics 123: 803-813.
Rasumov, V.I. 1930. U ber die photoperiodische Nachwirkung in Zusammenhang mit der
Wirkung verschiedener Aussaatterminen auf die Pflanzen. Planta 10: 345-377.
Robinson, H.F. and Moll, R.H. 1959. Implications of environmental effects on genotypes
in relation to breeding, pp. 24-31. Proc. 14th Annual Hybrid Corn Industry Research
Conference.
Robinson, J.F. and van Buijtenen, J.P. 1979. Correlation of seed weight and nurserybed traits
with 5-, 10-, and 15-year volumes in a loblolly pine progeny test. For. Sci. 25: 591-596.
Schmidt-Vogt, H. 1995. EinfluB der Umweltbedingungen auf die Jugendentwicklung von
Fichten-Provenienzversuchen (Picea abies (L.) Karst.). Alig. Forst- u. Jagdztg 166: 121127.
Skeates, D.A. 1986. Time of germination: a factor in rate of growth of accelerated transplants
of black spruce (Picea mariana (Mill) B.S.P.). Tree Planters' Notes 37(1): 20-24.
Sorensen, F.C. 1978. Date of sowing and nursery growth of provenances of Pseudotsuga
menziesii given two fertilizer regimes. J. Appl. Ecol. 15: 273-279.
Sorensen, F.C. 1980. Effect of date of cone collection and stratification period on germination
and growth of Douglas-fir seeds and seedlings. USDA Forest Serv., Pacific NW Forest and
Range Exp. Sta., Portland, Oregon. Res. Note PNW-346. 11 p.
Sorensen, F.C. 1983. Relationship between logarithms of chilling period and germination or
bud flush is linear for many tree species. For. Sci. 29: 237-240.
Sorensen, F.C. 1990. Stratification requirements for germination of western larch (Larix occi­
dentalis Nutt.) seed. USDA Forest Serv., Pacific NW Research Sta., Portland, Oregon.
Res. Note PNW-RN-493. 11 p.
Sorensen, F.C. 1991. Stratification period and germination of Douglas-fir seed from Oregon
seed orchards: two case studies. USDA Forest Serv., Pacific NW Research Sta., Portland,
Oregon. Res. Note PNW-RN-499. 22 p.
Sorensen, F.C. and Campbell, R.K. 1993. Seed weight - seedling size correlation in coastal
Douglas-fir: genetic and environmental components. Can. J. For. Res. 23: 275-285.
Tanaka, Y. 1976. Stratification and other pre-treatments of Douglas-fir for nursery germination,
pp. 163-173.In: Hatano, K., Asakawa, S., Katsuta, M., et al. (Eds) Proceedings ofIUFRO
International Symposium on Physiology of Seed Germination. Fuji, Japan.
Tanaka, Y., Kleyn, N.J., Harper, L.M. 1986. Seed stratification of Engelmann spruce and
lodgepole pine: the effect of stratification duration and timing of surface-drying. For.
Chron. 62: 147-151.
Tranquillini, W., Lechner, F., Oberarzbacher, P., Unterholzner, L. and Holzer, K. 1980. Uber das
HOhenwachstum von Fichtenklonen in verschiedener Seeh1>he. Mitt. der forstl. Bundes­
Versuchsanst. Wien. 129: 7-25.
Watson, D.J. and Baptiste, E.C.D. 1938. A comparative physiological study of sugarbeet and
mangold with respect to growth and sugar accumulation. I. Growth analysis of the crop in
the field. Ann. Bot. 2: 437-480.
Weber, J.C. and Sorensen, F.C. 1990. Effects of stratification and temperature on seed germi­
nation speed and uniformity in central Oregon ponderosa pine (Pinus ponderosa Doug!.
ex Laws.). USDA Forest Serv., Pacific NW Research Sta., Portland, Oregon. Res. Pap.
PNW-RP-429. 13 p.