Tree Physiology 26, 421–430
© 2006 Heron Publishing—Victoria, Canada
Temperature regulation of bud-burst phenology within and among
years in a young Douglas-fir (Pseudotsuga menziesii) plantation in
western Washington, USA
JOHN D. BAILEY1,2 and CONSTANCE A. HARRINGTON3
1
Oregon State University, Department of Forest Resources, 280 Peavy Hall, Corvallis, OR 97331, USA
2
Corresponding author (john.bailey@oregonstate.edu)
3
USDA Forest Service, Olympia Forestry Sciences Laboratory, 3625 93rd Avenue SW, Olympia, WA 98512-9193, USA
Received May 18, 2005; accepted August 13, 2005; published online January 16, 2006
Summary Past research has established that terminal buds of
Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) seedlings
from many seed sources have a chilling requirement of about
1200 h at 0–5 °C; once chilled, temperatures > 5 °C force bud
burst via accumulation of heat units. We tested this sequential
bud-burst model in the field to determine whether terminal
buds of trees in cooler microsites, which receive less heat forc­
ing, develop more slowly than those in warmer microsites. For
three years we monitored terminal bud development in young
saplings as well as soil and air temperatures on large, replicated
plots in a harvest unit; plots differed in microclimate based on
amount of harvest residue and shade from neighboring stands.
In two of three years, trees on cooler microsites broke bud 2 to
4 days earlier than those on warmer microsites, despite receiv­
ing less heat forcing from March to May each year. A simple
sequential model did not predict cooler sites having earlier bud
burst nor did it correctly predict the order of bud burst across
the three years. We modified the basic heat-forcing model to
initialize, or reset to zero, the accumulation of heat units when­
ever significant freezing temperature events (≥ 3 degree-hours
day – 1 < 0 °C) occurred; this modified model correctly pre­
dicted the sequence of bud burst across years. Soil temperature
alone or in combination with air temperature did not improve
our predictions of bud burst. Past models of bud burst have re­
lied heavily on data from controlled experiments with simple
temperature patterns; analysis of more variable temperature
patterns from our 3-year field trial, however, indicated that
simple models of bud burst are inaccurate. More complex mod­
els that incorporate chilling hours, heat forcing, photoperiod
and the occurrence of freeze events in the spring may be needed
to predict effects of future silvicultural treatments as well to in­
terpret the implications of climate-change scenarios. Develop­
ing and testing new models will require data from both field
and controlled-environment experiments.
Keywords: bud break, chilling, dormancy, freezing, heat forc­
ing, heat sum, photoperiod.
Introduction
The phenology of bud burst is fundamental to tree survival and
growth in temperate and boreal regions of the world (Sakai and
Larcher 1987). Early expansion of vegetative tissue is advanta­
geous for producing biomass and maintaining site dominance;
however, early tissue expansion also increases the risk of frost
damage from late-spring freezing temperatures (Heide 2003),
and spring frosts are two- to three-times more likely than fall
frosts to damage Douglas-fir in the Pacific Northwest (Timmis
et al. 1994). Plants have evolved mechanisms to use photoperiod and temperature cues to balance the benefits from early
bud burst with the probability of significant damage from
spring frost (Sakai and Larcher 1987, Hannerz 1999). Changes
in microclimate around plants or prevailing landscape climate
(IPCC 2001) will alter that balance within and among species
(Murray et al. 1989, Hänninen 1995, Guak et al. 1998).
Standard phenological models for temperate and boreal
trees include a minimum chilling requirement during the rest
phase of dormancy (late autumn and winter) that leads to full
growth competence of buds. This period is followed by peri­
ods of heat forcing, during the quiescence phase of dormancy
(in early spring), that ultimately produces bud burst (Fuchi­
gami et al. 1982, Hänninen 1995, Hannerz 1999, Hänninen
and Hari 2002, Tanja et al. 2003). Chuine et al. (1999) and
Hänninen and Hari (2002) concluded that there is little differ­
ence in accuracy among these various models, but the best
models (1) address chilling and forcing with functions more
sophisticated than simple degree-days and (2) consider forcing
temperatures sequential to the onset of quiescence, once chill­
ing requirements are fulfilled. Cannell and Smith (1983) and
422
BAILEY AND HARRINGTON
Murray et al. (1989) showed that a heat-forcing function for
boreal conifers may need to be modified by the duration of
chilling, but the need for this type of model has not been dem­
onstrated for Douglas-fir (Pseudotsuga menziesii (Mirb.)
Franco).
Wommack (1960, 1964) first demonstrated that Douglas-fir
has a chilling requirement, and greenhouse and growth-cham­
ber studies have established that chilling Douglas-fir seedlings
for 1200 h at 0 to 5 °C will result in rapid bud burst and proper
growth following planting for most seed sources (van den
Driessche 1975, 1977, Ritchie 1984, Lavender and Stafford
1985). In the forests of the Pacific Northwest, this chilling re­
quirement is naturally met in most years by early February and
can be enhanced for seedlings by artificial cold storage (Ritch­
ie 1984, Lavender and Stafford 1985). In the Pacific North­
west, there are typically less than 100 h below 0 °C, and such
sub-freezing temperatures are less efficient than temperatures
of 0–5 °C in meeting the chilling requirements for Douglas-fir
(Lavender 1981). Heat forcing of Douglas-fir buds comes
from the cumulative heat sum of temperatures > 5 °C (Camp­
bell and Sugano 1975, Hänninen 1995) with bud burst typi­
cally occurring in May. Photoperiod has been shown
important in growth chamber and greenhouse settings (Heide
1993a, 2003, Partanen et al. 1998), but relatively unimportant
in predicting the timing of bud burst in the field (Hänninen
1995, Chuine et al. 1999, Osbourne et al. 2000, Hänninen and
Hari 2002).
We tested the simple sequential, bud-burst model with
Douglas-fir saplings and three natural winter–spring tempera­
ture profiles. Trees were in a field study with large, replicated
plots that exhibited small but consistent differences in air and
soil temperature. We hypothesized that: (1) buds of young
Douglas-fir saplings in warmer microsites would burst sooner
than those in cooler microsites, because they would experience
greater heat forcing in the spring after achieving the minimum
1200 h of chilling; and (2) differences in timing of bud burst
across years would be consistent with such sequential model
predictions. We also explored the usefulness of soil tempera­
ture in addition to, or instead of, air temperature in predicting
the sequence of bud burst within and among years.
Materials and methods
The Fall River Long-Term Site Productivity study area was es­
tablished in 1999 on a 20-ha, highly productive, low-elevation
(300 m), gently-sloping site in the Coast Range Mountains
60 km southwest of Olympia, WA, USA (Terry et al. 2001).
Mean annual precipitation is 226 cm falling mostly as rain
from September through May; temperatures are mild (9.2 °C
mean annual) with a mean January minimum of –0.1 °C (US­
DA NRCS 1999). Soils are a deep, well-drained silt loam over
silty clay loam (Typic Fulvudand in the Boistfort series) devel­
oped from highly weathered basalt with ash influences in the
upper horizons (Steinbrenner and Gehrke 1973).
Fall River was established to examine the responses of soil
processes, microclimate and Douglas-fir growth to varying
amounts of logging slash and residual coarse wood retention
and vegetation control following commercial timber harvest.
Harvest treatments in 1999 ranged from a bole-only harvest
(small-end log diameter of 10 cm) that left ~130 Mg ha – 1 of
slash and coarse woody debris uniformly scattered across
plots, to a total-tree harvest that also included the removal of
historic coarse woody residuals, leaving only woody material
smaller than 0.6 cm diameter in the plots (< 3 Mg ha – 1). Seed­
lings from a mixed seed lot of 23 local, half-sib families were
raised in a nursery bed for one year and then transplanted and
grown a second year. The 2-year-old seedlings were planted at
2.5-m spacing in spring 2000 on 48, 0.25-ha plots in a random­
ized complete block design (Terry et al. 2001). We sampled the
young saplings in 15 plots during three years (2001, 2002 and
2004); the specific plots sampled each year varied slightly
based on the needs of other research activities.
Air temperature was monitored 25 cm above the soil surface
with temperature probes and HOBO H8 data loggers (Onset
Computer Corp., Pocasset, MA) with 4–8 shielded probes in
each plot located midway between planted rows of seedlings;
the data loggers recorded at 30-minute intervals. Chilling
hours were identified as each hour when temperature was in
the 0–5 °C range and then summed beginning October 15 each
year. Freezing degree-hours were the summed product of neg­
ative degrees; the maximum contribution of any winter day
was 48 freezing degree-hours, equivalent to 12 h at –4 °C, re­
corded on March 7, 2002. Heat forcing was the summed prod­
uct of degree-hours above a range of threshold temperatures
from 4 to 10 °C (we present results based on 5 °C as the thresh­
old temperature in this paper). We weighted heat-forcing
hours with a function that varied from a minumum weight of
0.4 at 5.1 °C to a maximum weight of 1 at 30 °C (after Cleary
and Waring 1969). The maximum daily contribution to heat
forcing was 255 degree-hours, equivalent to just over 10 h at or
above 30 °C, recorded on April 26, 2004. We also recorded
soil temperature at a 10-cm depth with temperature probes and
HOBO H8 data loggers logging at 30-minute intervals and cal­
culated soil heating above a range of threshold temperatures
from 2 to 6 °C unweighted and summed after chilling require­
ments were fulfilled.
We monitored bud development on 30 to 55 seedlings per
plot in the first 2 years, 2001 and 2002, using three codes for
bud condition (Table 1). In 2001, seedlings averaged 0.6 cm in
basal diameter (measured 15 cm above groundline) and 0.4 m
in total height (corresponding values for 2002 were 1.6 cm in
diameter and 0.9 m in height). Bud development was not mon­
itored during spring 2003; however, we measured and moni­
tored bud development for a third year (2004) on 40 saplings in
each of 13 plots selected to cover the full range of tempera­
tures across the study area. In 2004, we used an expanded
five-code system (Table 1) more similar to the index used by
Murray et al. (1989) and Hannerz (1999). Photo guidelines
were used for training and in the field to achieve consistency
between two observers, who rotated their observations by plot
on a weekly basis and jointly rated 5% of the total sample each
week. Saplings averaged 5.4 cm in basal diameter (2.2 cm di­
ameter at breast height) and 2.3 m in total height in spring,
2004.
TREE PHYSIOLOGY VOLUME 26, 2006
WITHIN- AND AMONG-YEAR REGULATION OF BUD BURST
Table 1. Development codes describing terminal buds of Douglas-fir
saplings at the Fall River Long-Term Site Productivity study. Integer
coding (0, 1, 2) was used in 2001 and 2002; all five codes were used in
2004.
Code
Description
0
Retracted, dormant winter bud; dark reddish brown
color
Slightly swollen with some elongation, minor
separation of terminal bud scales
Elongated, swollen bud with scale separation
throughout the bud; lighter color overall
Extremely-swollen whitish buds, green tissue visible
between scales near terminus
Fully-ruptured bud with new green foliage visible and
protruding
0.5
1
1.5
2
We used plot means of temperature, chilling hours, heat
sums, bud condition code at given dates and mean date of ter­
minal bud burst as the response variables for each year and
performed simple t tests within each year, assuming unequal
sample size and variance (SAS 8.2, SAS Institute Inc., Cary,
NC). We tested for differences in each response variable be­
tween the four warmest and four coolest plots (based on air
and soil temperature profiles) for each year (the maximum
number consistently available) to maintain a balanced design.
Plot selection varied slightly among years because of move­
ment of sensors to address other objectives of the Fall River
study, occasional failure of sensors and changing microcli­
mates as the stand developed; however, five of the eight plots
were used every year and the same plot was used from year to
year more than 80% of the time. The small but consistent dif­
ferences in temperature (warm versus cool) were created by
variable amounts of harvest residue present across treatments,
variability in landform producing cold area drainages and
proximity to neighboring mature stands that cast shade. We
present specific information from a single, median plot
(Plot 10) for the study area to illustrate differences across
years.
Results
Cool versus warm plots
Lower temperatures were consistently associated with shading
from neighboring mature stands along the southern and west­
ern edges of the plantation, with a slight depression in the area
near the southwest corner and with slash accumulation left by
the bole-only removal treatment across the plantation. For ex­
ample, the lowest recorded air temperature in 2001 (–5.8 °C)
was in the bole-only treatment plot at a shaded western edge of
the plantation and the average hourly minimum temperature
was lower across all bole-only removal treatment plots relative
to total-tree harvest plots (P < 0.01). The four coolest plots
across the study site, on average, logged 13.9% of total winter
hours at or below 0 °C between October 15, 2000 and Febru­
423
ary 10, 2001 (the accumulation of 1200 h of chilling on aver­
age), whereas the four warmest plots logged freezing tempera­
tures for only 12.2% of that time period (P = 0.02). The
maximum one-day contribution to freezing (44 degree-hours)
was from a bole-only harvest plot in a shaded corner of the
study area; the maximum one-day contribution to heat forcing
(255 degree-hours) was from a total-tree harvest plot in an
unshaded portion of the study area. Overall, differences in
mean daily temperature between cool and warm plots were
typically less than 0.5 °C, difficult to discern (Figure 1a), and
significant at P < 0.05 for no more than about 10 days per
month. Only when freezing degree-hours, chilling hours, and
heat forcing were accumulated over the winter and spring did
the differences between cool and warm microsites become
readily apparent and statistically significant (Table 2, Fig­
ures 1b and 1c).
Contrary to what would have been predicted from the sim­
ple sequential heat-forcing model, however, mean terminal
bud burst of young Douglas-fir saplings on cool-microsite
plots was on average 2 and 3 days earlier, respectively,
(P < 0.03) relative to that on warm plots during 2001 and 2004
(Figure 2a). This earlier bud burst occurred despite similar or
lower springtime heat forcing (Table 2). Differences in bud
burst between the coolest and warmest plots were 3 and 4 days,
respectively, in 2001 and 2004. Furthermore, bud development
was more advanced in cool plots for 2–3 weeks before mean
terminal bud burst in both of those years (Figure 2a). This pat­
tern of bud burst among the four warmest and four coolest
plots was consistent in 2001 and 2004 (2001 shown in Fig­
ure 2b); there were no differences in either the rate of terminal
bud development or ultimate date of mean terminal bud burst
in 2002.
Annual variation
There were important differences among the three years of this
study in terms of temperature profiles and resulting chilling
hours and heat forcing (Figure 3). The winter of 2000 through
the spring of 2001 approximated the 50-year average for this
area (National Weather Service, Silver Spring, MD) and
1200 h of chilling were achieved on February 12, 2001 for the
median plot at Fall River (Figure 3a). Winter temperatures
were lower before 2002 bud burst and the chilling requirement
was fulfilled nine days earlier than the previous year (February
3, 2002). Conversely, winter months before 2004 bud burst
were warmer than average and the chilling requirement was
not met until nine days later than the average year (February
21, 2004). Theoretically, earlier fulfillment of 1200 chilling
hours translates to an earlier start of quiescence and accumula­
tion of heat forcing units; however, the relatively cool temper­
atures of any February contribute little to heat forcing and all
years showed similar heat forcing through much of March
(Figure 3b).
Our characterization of 2002 as a cool year (particularly a
cold winter) and 2004 as a warm year was supported during
the ensuing spring heat-forcing months of each year. March
2002 averaged 2.7 °C cooler than March 2001 and spring tem-
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
424
BAILEY AND HARRINGTON
Figure 1. The 2001 air temperature profiles (a), accumulation of chill­
ing hours (b) and accumulation of heat-forcing units (c) for two con­
trasting plots at Fall River, western Washington. Plot 23, classified as
a cool plot, receives morning shade during the winter and has accumu­
lated harvest residue on the soil surface; Plot 43, a warm plot, is
unshaded with bare soil exposed. Instantaneous differences are minor
(a) but cumulative differences become significant (b and c).
peratures produced an additional 965 chilling hours beyond
the minimum 1200 h (Figure 3a), 5% more than 2001. April
2004 averaged 3.8 °C warmer than April 2001 and spring tem­
peratures produced only 545 h of chilling beyond the 1200 h
requirement (15% less than 2001). However, total heat forcing
reflects this cool versus warm year characterization only for
2004 (Table 2). During 2001 (a year with winter and spring
temperatures close to the 50-year average) plots accumulated
on average 3200 weighted degree-hours based on our calcula­
tions. The spring months of 2002 had a slightly cooler temper­
ature profile, but the length of the heat-forcing period was
longer, producing 3800 (+18%) weighted degree-hours de­
spite being a cooler year. During the warm 2004 spring, the
same plots totaled 4800 (+51%) weighted degree-hours (Fig­
ure 3b shows data for Plot 10).
Although 2002 accumulated significantly more heating
units than 2001, mean date of terminal bud burst was 14 days
later than in the previous year. Clearly, a simple sequential
heat-forcing model could not account for these observed dif­
ferences. Our sequential heat-forcing model was not improved
across the years with changes to threshold temperatures (e.g.,
5 versus 4.4 °C for chilling) nor by various weighting func­
tions (or unweighted data) in our heat-forcing calculations.
Such adjustments produced earlier/later dates to fulfilled
chilling requirements and higher/lower heat sum values; how­
ever, none of these adjustments altered the prediction of earlier
bud burst with higher temperatures, which was exactly the op­
posite of what we observed.
Soil temperature data, alone or in combination with air tem­
perature, was not useful in predicting date of bud burst. Soil
temperatures were correlated with air temperatures; thus, the
plots with lower air temperatures were the plots with lower soil
temperatures. Like air heat sums, accumulated soil heating at
bud burst was 22% greater in 2002 relative to 2001; thus, soil
temperature did not help explain why bud burst was later in
2002 than in 2001. Soil temperatures varied much less than air
temperatures and no freezing temperatures were recorded in
the soil.
Based on an examination of temperature data across years,
we hypothesized that cold weather or freeze events in the
spring (after 1200 chilling hours had been achieved) were
playing a role. We evaluated the possible effect of these events
by initializing (i.e., resetting to zero) the sum of heat forcing
units under several criteria. There were numerous dates each
year when air temperature was slightly below freezing (0.5 °C)
for up to 1 h. These minor freeze events did not seem useful in
our modeling exercises because they occurred so frequently.
We found, however, that if we used “three or more hours per
day at least 1 °C below freezing” to define a significant freeze
event, and, thus, the criterion for initializing the accumulation
of heat forcing, we could reverse the order of predicted bud
burst for 2001 and 2002. The date of the last significant freeze
event varied by more than a month across the three years (Ta­
ble 2). Initializing heat forcing resulted in bud burst occurring
at about the same number of heat forcing units in 2002 and
2001 (Figure 4) and in the correct sequence relevant to ob­
served bud burst. The accumulated number of heat-forcing
TREE PHYSIOLOGY VOLUME 26, 2006
WITHIN- AND AMONG-YEAR REGULATION OF BUD BURST
425
Table 2. Air temperature and bud burst (means with ranges) for four warm and four cool plots at the Fall River study site, western Washington.
Temperatures were measured 25 cm above the soil surface. Chilling hours were summed from October 15 to bud burst. Heat forcing is summed
from date of minimum chilling (1200 h) to bud burst as weighted degree-hours adapted from Cleary and Waring (1969). The last freeze event is de­
fined as the last date, after 1200 chilling hours have been achieved, with ≥ 3 degree-hours day – 1 below 0 °C.
Warm plots
Cool plots
P value
Winter–Spring 2001
Date 1200 chilling hours achieved
February 15 (Feb 14–16)
Total chilling hours
2036 (2000–2067)
Total heat forcing (degree-hours)
3212 (3143–3349)
Date of terminal bud burst
May 17 (May 16–19)
Date of last freeze event: April 14, 2001 (~32 days before bud burst)
February 9 (Feb 5–14)
2113 (2040–2209)
3227 (3088–3398)
May 15 (May 14–15)
0.0774
0.1549
0.8613
0.0042
Winter–Spring 2002
Date 1200 chilling hours achieved
February 4 (Feb 4–5)
Total chilling hours
2158 (2128–2203)
Total heat forcing (degree-hours)
3868 (3559–4127)
Date of terminal bud burst
May 28 (May 25–30)
Date of last freeze event: May 9, 2002 (~20 days before bud burst)
January 31 (Jan 26–Feb 3)
2275 (2233–2352)
3797 (3768–3826)
May 29 (May 28–31)
0.0759
0.0142
0.5942
0.3146
Winter–Spring 2004
Date 1200 chilling hours achieved
February 22 (Feb 20–24)
Total chilling hours
1733 (1709–1784)
Total heat forcing (degree-hours)
5026 (4828–5179)
Date of bud burst
May 9 (May 7–9)
Date of last freeze event: April 6, 2004 (~32 days before bud burst)
February 18 (Feb 15–21)
1774 (1557–1807)
4656 (4509–4814)
May 6 (May 5–8)
0.0836
0.1083
0.0099
0.0314
units was much higher at the date of bud burst in 2004 than in
the other two years; however, the modified model predicted
the sequence of bud burst correctly across the three years.
Differences in photoperiod when various thresholds were
reached may be important in determining the rate of develop­
ment; however, because of differences in temperature patterns
among years, the effect of photoperiod is confounded with dif­
ferences in number of chilling hours and heat-forcing units.
Photoperiod on the date of the last freeze event ranged from
13.11 h on April 6, 2004 to 14.76 h on May 9, 2002. Thus, the
photoperiod was 99 minutes longer on the date when we
started accumulating heat forcing in 2002 than in 2004. Photoperiod at the date of 50 or 100% bud burst differed much less
(less than 30 minutes) across the three years.
Discussion
The additional chilling hours associated with cooler microsites apparently caused earlier bud burst in this young Douglas-fir plantation for two of our three study years. We had a
well-replicated design with dozens of precise temperature sen­
sors and hundreds of young saplings of similar genetic compo­
sition, common in industry plantations. Genetic regulation of
bud burst has been well documented (Campbell and Sugano
1975, White et al. 1979, Li and Adams 1993, Myking and
Heide 1995), but our data set contained this genetic variability
within the error structure of the experiment and still yielded
significant differences among treatments. There were no pho­
toperiod differences among cool and warm microsites within
years and no consistent difference in foliar nitrogen (N) con­
centration between our cool and warm plots (Roberts et al.
2005). Our cooler sites may have had a slightly greater N avail­
ability because of leaching from accumulations of logging
slash (B. Strahm, data on file, University of Washington) and
N has been shown to promote earlier bud burst in Picea abies
(L.) Karst. (Fløistad and Kohmann 2004). However, it seems
unlikely that the small differences in N concentrations in soil
solution would accelerate bud development because foliar N
did not differ significantly between the warmer (open, to­
tal-tree harvest) and cooler (partially-shaded, bole-only har­
vest) plots. We suspect that the modest gain in growing season
(2–4 days) resulted from some benefit of additional chilling.
Active management that fosters cooler microsites (e.g., Lang­
vall and Löfvenius 2002), at least in the absence of other major
growth-limiting factors, could result in earlier bud burst. Of
course, such practices could also increase the risk of spring
frost injury by lowering mean temperature in the weeks fol­
lowing bud burst.
Several authors have documented that additional chilling
promotes earlier bud burst in greenhouses and growth cham­
bers in several forest species (Cannell and Smith 1983, Murray
et al. 1989, Heide 1993b, Myking and Heide 1995, Partanen et
al. 1998). Van den Driessche (1975, 1977) and Ritchie (1984)
found that additional chilling (more than the minimum
1200 hours) of Douglas-fir seedlings led to earlier bud burst in
controlled-environment studies, particularly with cooler forc­
ing temperatures that more closely simulated those in our field
study. In one field experiment on conifer bud burst, Hänninen
(1995) found an unexplained discrepancy in his models exam­
ining the effects of major increases in temperature (3–6 °C) on
Pinus sylvestris L. saplings. His models predicted bud burst
~50 days earlier than it actually occurred with such warmer
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
426
BAILEY AND HARRINGTON
Figure 2. Bud development over successive weeks of observation at
Fall River, western Washington. See Table 1 for bud codes. Mean val­
ues for warm and cool plots are shown (a) for three study years. The
rate of bud burst in 2001 and 2004 was consistently higher for cooler
plots relative to warmer plots (shown by lines with one standard error
bars placed at monitoring times demonstrating significant differences;
n = 4). In 2002, rates of bud burst did not differ significantly between
cool and warm plots. Values for individual plots (b) are shown for
2001.
temperatures; it is possible that a lack of complete chilling
slowed bud development despite substantial heat forcing, or
that photoperiod played an undocumented role in the observed
effects. Though his model was inaccurate, major temperature
increases still correctly predicted earlier bud burst overall
(Hänninen 1995), a result supported by Osbourne et al. (2000)
as well as the results from the third, very warm year (2004) in
this study. Langvall and Löfvenius (2002) found delayed bud
burst under a range of Picea abies shelterwood densities that
effectively warmed microsites relative to clearcut areas over
Figure 3. Annual differences in: (a) chilling hours above 1200 and (b)
heat-forcing degree hours during the quiescent period before date of
bud burst for Plot 10, the median plot at Fall River, western Washing­
ton. Heat forcing is the sum of weighted degree-hours (see text for de­
tails). The line for each year begins the day 1200 chilling hours is
achieved and ends when bud burst is complete.
several years following harvest; however, these authors did not
present chilling-hour data because of the low chilling require­
ment of the species. In summary, it will require a more com­
plex model than we currently have to accommodate both the
negative influence of slight warming (due to delay in achiev­
ing, or to a decrease in, chilling hours) and the positive influ­
ence of major warming through increases in heat forcing
(Heide 1993b, Osbourne et al. 2000, Pop et al. 2000).
Though 1200 h at 0–5 °C is sufficient for normal bud devel­
opment in Douglas-fir following planting (van den Driessche
1977, Lavender 1981), it appears that the upper limit of benefi­
cial chilling for coastal Douglas-fir tree buds experiencing nat-
TREE PHYSIOLOGY VOLUME 26, 2006
WITHIN- AND AMONG-YEAR REGULATION OF BUD BURST
Figure 4. Revised heat forcing trajectories by year at Fall River, west­
ern Washington, based on initializing the accumulation of heat-forc­
ing units whenever a freezing event (> 3 degree-hours day – 1 below
0 °C) occurred. The line for each year ends when bud burst was com­
plete. When heat forcing is initialized for each freeze event, it rises
most rapidly after the last such event.
ural temperature profiles is actually closer to 2000 h. We
documented more rapid bud burst on cooler plots in two years
when chilling hours averaged 1750 and 2080 across plots (Ta­
ble 2), suggesting an effect of additional chilling up to 880 h
beyond the minimum required to break bud dormancy; how­
ever, we observed no marginal effect of increasing chilling to
2200 h. These observations are consistent with nursery and
greenhouse observations on Douglas-fir seedlings by Ritchie
(1984), who showed a 2-week decrease in days to bud burst
with 2100 versus 1200 chilling hours, and with the observa­
tions by van den Driessche (1975) that bud burst at 12.5–13 °C
was slow when Douglas-fir seedlings received 1250 chilling
hours and rapid when plants had received 2070 chilling hours.
The optimum chilling requirement is longer (2856 hours sug­
gested by Wells 1979) for Rocky Mountain Douglas-fir
(Pseudotsuga menziesii var. glauca); thus, it may vary with ge­
notype or geographic location as well.
Freezing events
Variation in bud burst among years in this 3-year study ap­
peared to be regulated by the amount of heat forcing accumu­
lated since the last significant freezing event (> 3 degree-hours
below freezing per day). Initializing heat forcing units at each
such freezing event led to more accurate prediction of bud
burst between the first two very dissimilar thermal years (Fig­
ure 4), and may be the only way to predict the 2-week delay in
bud burst during 2002. Such an incorporation of last-freeze
initialization might explain the anomalous bud burst pattern
documented by Hannerz (1999). Physiological responses to
minor freezing within bud tissues likely delays phenological
development, although the specific mechanism of response is
427
unknown. Short-term non-lethal freezing events can signifi­
cantly impact recovery of photosynthetic capacity in Pinus
sylvestris and Picea abies in boreal regions (Hänninen and
Hari 2002, Tanja et al. 2003). Rinne et al. (1997) explored the
role of abscisic acid in Betula spp. bud burst, Bergervoet et al.
(1999) identified the role of microtubule formation in the regu­
lation of growth hormones in Pinus sylvestris, Amasino
(2004) proposed competing enzymes with different Q10 values
and several authors have speculated that there is a role for
cytokinins and gibberellins (Reid and Burrows 1968, Laven­
der et al. 1973). Further research on last-freeze initialization
under field temperature profiles may provide insight into one
or more of these mechanisms.
Our last-freeze initialization model consistently places an­
nual bud burst in its observed sequence and roughly aligns the
total heat forcing required for bud burst in 2001 and 2002, but
it does not explain the shorter, 20-day interval to bud burst ob­
served in 2002 (Figure 4). This 40% greater “efficiency” in
heat forcing later in the spring after the last freeze may reflect:
(1) the advantages of 5% additional chilling, (2) the compres­
sion of similar and abundant heat forcing units into a shorter
time interval (Figure 4), or (3) some predisposition to more
rapid bud development later in the season (e.g., a longer
photoperiod or changes in light quality as spring progresses or
both). Adjusting the weighting function for heat forcing
changed the values of various calculations but did not change
the order of predicted bud burst across years, nor the percent
differences in amount of heat forcing at time of bud burst.
Our addition of last-freeze initialization to the sequential
bud-burst model did not predict or explain the much greater
heat forcing associated with bud burst in 2004 (the warmest
year). Similar to the model-predicted results in Hänninen
(1995), our heat-forcing function predicted a much earlier bud
burst in 2004 than we actually observed (Figure 4). In other
words, mean terminal bud burst should have occurred 10 days
earlier, near April 28 (at ~2000 weighted degree-hours), if bud
burst were solely controlled by the amount of heat forcing fol­
lowing the last significant freeze. Again, adjusting the weight­
ing functions for heat forcing uniformly changed values in all
years but did not affect the pattern. Slower bud development in
2004 may be explained by reduced chilling hours or much
shorter photoperiod at the time of the last freeze event or bud
burst. Simple length of time since the last significant freezing
event may also play some role in regulating bud burst. The
time between last freezing event and bud burst ranged from 20
to 32 days (Figure 4), perhaps indicating the existence of a ba­
sic physiological recovery period similar to that suggested by
Hänninen and Hari (2002) or Tanja et al. (2003).
Soil temperature
Soil temperatures did not improve our models of bud burst.
Soil temperatures were much less variable than air tempera­
tures and did not fall below the freezing point; thus, they were
less effective than air temperatures in predicting bud break.
Cold soil has been shown to delay bud burst of Douglas-fir
(Lavender 1973, Lopushinsky and Max 1990) but there is no
evidence that soil temperature is the primary driver of bud
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
428
BAILEY AND HARRINGTON
burst. Because the viscosity of water is directly proportional to
water temperature (Kramer 1934), cold soil will reduce water
transport (Carlson and Miller 1991) and, thus, retard cell elon­
gation; however, there is no evidence that the actual sensing
mechanism driving bud burst is outside the buds themselves.
Furthermore, soil temperatures did not explain the order of
bud burst between microsites or among years in our study or in
the study by Tanja et al. (2003).
Climate warming
Any climate warming in temperate and boreal latitudes (IPCC
2001) will lead to winter temperature profiles that provide, on
average, fewer chilling hours per year for Douglas-fir in the
Pacific Northwest; therefore, trees, especially those at the pe­
riphery of the species’ range, may experience a less than opti­
mum number of chilling hours for bud development. If temp­
erature variability does not increase, however, the same
warmer profiles would produce earlier dates for the last signif­
icant non-lethal freezing temperature and greater cumulative
heat forcing after that last freeze. As in the very warm third
year of this study (2004), the latter two effects would more
than compensate for delayed or reduced cumulative chilling
hours, leading overall to earlier bud burst (Hänninen 1995,
Pop et al. 2000). However, current heat-forcing components of
standard, sequential bud-burst models consistently over-pre­
dict bud burst and may miss the rank order of bud burst among
various climate scenarios as they do not incorporate the effects
of freezing temperatures. Additional research with climatecontrolled chambers programmed to mimic natural tempera­
ture profiles should enable rapid development of more accu­
rate models of bud burst. The importance of additional chill­
ing, timing of freezing events and photoperiod could be
evaluated with trials designed to test specific hypotheses.
Whether a warmer climate in temperate and boreal latitudes
produces a longer and more productive growing season de­
pends on many assumptions about future climate (Spittlehouse
2003) and is well beyond the scope of this paper. However, two
phenological issues are worth considering. (1) The 50-year av­
erage winter-spring temperature profile for our study area,
which is in the central portion of the species’ range, results in
about 2100 chilling hours, whereas 2004, which averaged 2 °C
warmer during the winter and spring, produced only 1745
chilling hours. Another 3 °C increase in temperature will re­
duce chilling hours to the demonstrated minimum for
Douglas-fir (i.e., 1200 chilling hours), below which odd
growth patterns will likely develop (Lavender 1981). If insuf­
ficient chilling occurs, long photoperiods will presumably
trigger bud burst (as demonstrated for Pinus taeda seedlings
by Garber 1983). However, it remains to be determined what
photoperiod will trigger bud burst with low chilling accumula­
tion, and whether the timing of that photoperiod will effec­
tively shorten the growing season from its current length. (2)
The more erratic weather patterns likely to occur with climate
warming (IPCC 2001) could increase the likelihood of lethal
freezing events for buds and newly emergent tissues (Hännin­
en 1996), although short photoperiods could retard bud burst
beyond the time of such events.
Tree stress, damage and mortality caused by changes in
phenology could become as important as pests and fire (Innes
and Peterson 2003, Spittlehouse 2003), or interact with pests
and fire in shaping temperate and boreal ecosystems to match
future climates. For example, Clancy et al. (2004) suggested
spring phenology was the most important factor associated
with stand-level susceptibility of Douglas-fir to western
spruce budworm (Choristoneura occidentalis Freeman) be­
cause early bud burst in Douglas-fir would coincide with instar
emergence.
Our study was located on a relatively homogeneous slope in
a simple stand type (plantation) with relatively subtle differ­
ences in microclimate within years. Much larger variability in
temperature patterns can be anticipated if more complex stand
structures are considered and different natural temperature
patterns occur; thus, future models of bud burst will need to be
more complex. Determining the mechanisms by which trees
sense and respond to winter and spring temperatures by exper­
imentation under a wide range of natural conditions as used in
this study will be fundamental in developing and testing future
models.
Acknowledgements
We thank Weyerhaeuser Company for providing access to the study
area and data on tree size. Partial funding was provided by the USDA
Forest Service-American Forest and Paper Association Agenda 2020
research program and Northern Arizona University.
References
Amasino, R. 2004. Vernalization, competence, and the epigenetic
memory of winter. Plant Cell 16:2553–2559.
Bergervoet, J.H., H. Jing, J.W.E. van den Hout, R. Delmondez de Cas­
tro, B. Kunneman, R.J. Bino and S.P.C. Groot. 1999. Expression of
beta-tubulin during dormancy induction and release in apical and
axillary buds of five woody species. Physiol. Plant. 106:238–245.
Campbell, R.K. and A.I. Sugano. 1975. Phenology and bud burst in
Douglas-fir related to provenance, photoperiod, chilling and flush­
ing temperature. Bot. Gaz. 136:290–298.
Cannell, M.G.R. and R.I. Smith. 1983. Thermal time, chill days and
prediction of budburst in Picea sitchensis. J. Appl. Ecol. 20:
951–963.
Carlson, W.C. and D.E. Miller. 1991. Target seedling root system size,
hydraulic conductivity, and water use during seedling establish­
ment. In Target Seedling Symposium: Proceedings, Combined
Meeting of the Western Forest Nursery Associations. USDA FS,
Gen. Tech. Report RM-200, pp 53–65.
Chuine, I., P. Cour and D.D. Rousseau. 1999. Selecting models to pre­
dict the timing of flowering of temperate trees: implications for tree
phenology modelling. Plant Cell Environ. 22:1–13.
Clancy, K.M., Z. Chen and T.E. Kolb. 2004. Foliar nutrients and in­
duced susceptibility: genetic mechanisms of Douglas-fir resistance
to western spruce budworm defoliation. Can. J. For. Res. 34:
939–949.
Cleary, B.D. and R.H. Waring. 1969. Temperature: collection of data
and its analysis for the interpretation of plant growth and distribu­
tion. Can. J. Bot. 47:167–173.
TREE PHYSIOLOGY VOLUME 26, 2006
WITHIN- AND AMONG-YEAR REGULATION OF BUD BURST
Fløistad, I.S. and K. Kohmann. 2004. Influence of nutrient supply on
spring frost hardiness and time of bud break in Norway spruce
(Picea abies (L.) Karst.) seedlings. New For. 27:1–11.
Fuchigami, L.H., C.J. Weiser, K. Kobayashi, R. Timmis and L.V.
Gusta. 1982. A degree growth stage (°GS) model and cold acclima­
tion in temperate woody plants. In Plant Cold Hardiness and Freez­
ing Stress. Eds. P.H. Li and A. Sakai. Academic Press, New York,
pp 93–116.
Garber, M.P. 1983. Effects of chilling and photoperiod on dormancy
release of container-grown loblolly pine seedlings. Can. J. For.
Res. 13:1265–1270.
Guak, S., D.M. Olsyzk, L.H. Fuchigami and D.T. Tingey. 1998. Ef­
fects of elevated CO2 and temperature on cold hardiness and spring
bud burst and growth in Douglas-fir (Pseudotsuga menziesii). Tree
Physiol. 18:671–679.
Hannerz, M. 1999. Evaluation of temperature models for predicting
bud burst in Norway spruce. Can. J. For. Res. 29:9–19.
Hänninen, H. 1995. Effects of climatic change on trees from cool and
temperate regions: an ecophysiological approach to modelling of
bud burst phenology. Can. J. Bot. 73:183–199.
Hänninen, H. 1996. Effects of climatic warming on northern trees:
Testing the frost damage hypothesis with meteorological data from
provenance transfer experiments. Scand. J. For. Res. 11:17–25.
Hänninen, H. and P. Hari. 2002. Recovery of photosynthesis of boreal
conifers during spring: a comparison of two models. For. Ecol.
Manage. 169:53–64.
Heide, O.M. 1993a. Dormancy release in beech buds (Fagus syl­
vatica) requires both chilling and long days. Physiol. Plant. 89:
187–191.
Heide, O.M. 1993b. Daylength and thermal time responses of budburst during dormancy release in some northern deciduous trees.
Physiol. Plant. 88:531–540.
Heide, O.M. 2003. High autumn temperature delays spring bud burst
in boreal trees, counterbalancing the effect of climatic warming.
Tree Physiol. 23:931–936.
Innes, J.L. and D.L. Peterson. 2003. Proceedings introduction: Man­
aging forests in a greenhouse world—context and challenges. In
Climate Change, Carbon and Forestry in Northwestern North
America. US Forest Service PNW-GTR-614, pp 1–9.
Intergovernmental Panel on Climate Change (IPCC). 2001. Climate
change 2001: the scientific basis. Contribution of Working Group 1
to the third assessment report of the IPCC. Cambridge University
Press, Cambridge, 786 p.
Kramer, P.J. 1934. Effects of soil temperature on the absorption of wa­
ter by plants. Science 79:131-132.
Langvall, O. and M.O. Löfvenius. 2002. Effect of shelterwood den­
sity on nocturnal near-ground temperature, frost injury risk and
budburst date of Norway spruce. For. Ecol. Manage. 168:149–161.
Lavender, D.P. 1981. Environment and shoot growth of woody plants.
Research Paper No. 45, Forest Research Laboratory. Oregon State
Univ., Corvallis, 47 p.
Lavender, D.P. and S.G. Stafford. 1985. Douglas-fir seedlings: some
factors affecting chilling requirement, bud activity, and new foliage
production. Can. J. For. Res. 15:309–312.
Lavender, D.P., G.B. Sweet, J.B. Zaerr and R.K. Herman. 1973.
Spring shoot growth in Douglas-fir may be initiated by gibberellins
exported from the roots. Science 182:838–839.
Li, P. and W.T. Adams. 1993. Genetic control of bud phenology in
pole-size trees and seedlings of coastal Douglas-fir. Can. J. For.
Res. 23:1043–1051.
Lopushinsky, W. and T.A. Max. 1990. Effect of soil temperature on
root and shoot growth and on budburst timing in conifer seedling
transplants. New For. 4:107–124.
429
Murray, M.B., M.G.R. Cannell and R.I. Smith. 1989. Date of budburst of fifteen tree species in Britain following climatic warming.
J. Appl. Ecol. 26:693–700.
Myking, T. and O.M. Heide. 1995. Dormancy release and chilling re­
quirement of buds of latitudinal ecotypes of Betula pendula and
B. pubescens. Tree Physiol. 15:697–704.
Osbourne, C.P., I. Chuine, D. Viner and F.I. Woodward. 2000. Olive
phenology as a sensitive indicator of future climatic warming in the
Mediterranean. Plant Cell Environ. 23:701–710.
Partanen, J., V. Koski and H. Hänninen. 1998. Effects of photoperiod
and temperature on the timing of bud burst in Norway spruce
(Picea abies). Tree Physiol. 18:811–816.
Pop, E.W., S.F. Oberbauer and G. Starr. 2000. Predicting vegetative
bud break in two arctic deciduous shrub species, Salix pulchra and
Betula nana. Oecologia 124:176–184.
Reid, D.M. and W.J. Burrows. 1968. Cytokinen and gibberellin-like
activity in the spring sap of trees. Experientia 24:189–190.
Richie, G.A. 1984. Effect of freezer storage on bud dormancy release
in Douglas-fir seedlings. Can. J. For. Res. 14:186–190.
Rinne, P., H. Hänninen, P. Kaikuranta, J.E. Jalonen and T. Repo.
1997. Freezing exposure releases bud dormancy in Betula pub­
escens and B. pendula. Plant Cell Environ. 20:1199–1204.
Roberts, S.D., C.A. Harrington and T.A. Terry. 2005. Harvest residue
and competing vegetation affect soil moisture, soil temperature,
N availability, and Douglas-fir seedling growth. For. Ecol. Manage.
205:335–350.
Sakai, A. and W. Larcher. 1987. Frost survival of plants: responses
and adaptation to freezing stress. Springer-Verlag, Berlin, 321 p.
Spittlehouse, D.L. 2003. Water availability, climate change and the
growth of Douglas-fir in the Georgia Basin. Can. Water Res. Bull.
28:673–688.
Steinbrenner, E.C. and F.E. Gehrke. 1973. Soil survey of the McDon­
ald Tree Farm. Weyerhaeuser Company, Tacoma, WA, 56 p.
Tanja, S., F. Berninger, T. Vesala et al. 2003. Air temperature triggers
the recovery of evergreen boreal forest photosynthesis in spring.
Global Change Biol. 9:1410–1426.
Terry, T.A., R.B. Harrison and C.A. Harrington. 2001. Fall River
long-term site productivity study: objectives and design. Weyer­
haeuser Company, Centralia, WA. Forestry Research Technical
Note—Long-term Site Productivity Paper 01–1, 10 p.
Timmis, R., J. Flewelling and C. Talbert. 1994. Frost injury prediction
model for Douglas-fir seedlings in the Pacific Northwest. Tree
Physiol. 14:855–869.
van den Driessche, R. 1975. Flushing response of Douglas-fir buds to
chilling and to different air temperatures after chilling. B.C. For.
Serv. Res. Note 71, 22 p.
van den Driessche, R. 1977. Survival of coastal and interior Douglasfir seedlings after storage at different temperatures, and effective­
ness of cold storage in satisfying chilling requirements. Can. J. For.
Res. 7:125–131.
USDA Natural Resources Conservation Service, National Water and
Climate Center; and Oregon State University Spatial Climate Anal­
ysis Service. 1999. Parameter-elevation Regressions on Independ­
ent Slopes Model (PRISM). Available online from http://www.
wcc.nrcs.usda.gov/water/climate/prism/prism.html.
Wells, S.P. 1979. Chilling requirements for optimal growth of Rocky
Mountain Douglas-fir seedlings. USDA FS, Research Note INT–
254, 9 p.
White, T.L., K.K. Ching and J. Walters. 1979. Effects of provenance,
years, and planting location on bud burst of Douglas-fir. For. Sci.
25:161–167.
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
430
BAILEY AND HARRINGTON
Wommack, D.E. 1960. Effect of winter chilling and photoperiod on
growth resumption in Douglas-fir (Pseudotsuga menziesii (Mirb.)
Franco). M.Sc. Thesis. Oregon State Univ., Corvallis, 82 p.
Wommack, D.E. 1964. Temperature effects on the growth of Douglas-fir seedlings. Ph.D. Thesis. Oregon State Univ., Corvallis,
176 p.
TREE PHYSIOLOGY VOLUME 26, 2006