Genecology of Douglas Fir in Western Oregon and Washington
J. BRADLEY ST CLAIR1,*, NANCY L. MANDEL1and KENNETH W. VANCE-BORLAND2
1USDA Forest Service, Pacific Northwest Research Station, 3200 SW Jefferson Way, Corvallis, OR 97331, USA and
2Oregon State University, Department of Forest Science, Corvallis, OR 97331, USA
Received: 26 May 2005 Returned for revision: 10 August 2005 Accepted: 6 September 2005Published electronically: 24 October 2005
? Background and Aims Genecological knowledge is important for understanding evolutionary processes and for
managing genetic resources. Previous studies of coastal Douglas fir (Pseudotsuga menziesii var. menziesii) have
been inconclusive with respect to geographical patterns of variation, due in part to limited sample intensity and
geographical and climatic representation. This study describes and maps patterns of genetic variation in adaptive
traits in coastal Douglas fir in western Oregon and Washington, USA.
? Methods Traits of growth, phenology and partitioning were measured in seedlings of 1338 parents from 1048
locations grown in common gardens.Relations between traits andenvironments of seed sources were explored using
regressions and canonical correlation analysis. Maps of genetic variation as related to the environment were
developed using a geographical information system (GIS).
? Key Results Populations differed considerably for adaptive traits, in particular for bud phenology and emergence.
Variation in bud-set, emergence and growth was strongly related to elevation and cool-season temperatures.
Variation in bud-burst and partitioning to stem diameter versus height was related to latitude and summer drought.
Seedlings from the east side of the Washington Cascades were considerably smaller, set bud later and burst bud
earlier than populations from the west side.
? Conclusions Winter temperatures and frost dates are of overriding importance to the adaptation of Douglas fir to
Pacific Northwest environments. Summer drought is of less importance. Maps generated using canonical correlation
analysis and GIS allow easy visualization of a complex array of traits as related to a complex array of environments.
The composite traits derived from canonical correlation analysis show two different patterns of variation associated
with different gradients of cool-season temperatures and summer drought. The difference in growth and phenology
between the westside and eastside Washington Cascades is hypothesized to be a consequence of the presence of
interior variety (P. menziessii var. glauca) on the eastside.
Key words: Pseudotsuga menziesii, genecology, geographical variation, adaptation, growth, phenology.
Describing and understanding the geographical structure of
genetic variation and its relation to environments are impor-
tant for understanding evolutionary processes and for man-
aging our heritage of genetic resources. Correlations
between genetic variation and environmental differences
amongseed sources suggest naturalselectionandadaptation
of genotypes to their environments, particularly when the
correlations make sense physiologically (Heslop-Harrison,
1964; Endler, 1986). Mapped genetic variation and an
understanding of natural genetic structure are used to
develop guidelines for seed movement in reforestation,
for managing breeding populations in advanced generation
breeding programmes, for evaluating conservation of gen-
etic resources, and for predicting and possibly mitigating
effects of climate change. Furthermore, knowledge of
geographical variation among populations of outcrossing,
undomesticated conifers may contribute substantially to
exploring the molecular basis of quantitatively inherited
adaptive traits through association genetics studies (Neale
and Savolainen, 2004).
Douglas fir (Pseudotsuga menziesii) is one of the most
ecologically and economically important trees in western
North America, and is planted as an exotic timber species in
Europe, New Zealand, Australia and Chile. It has one of the
widest natural ranges of any tree species, extending from
the Pacific Coast to the eastern slope of the Rocky
Mountains and from 19?N in Mexico to 55?N in western
Canada (Hermann and Lavender, 1990). Two varieties are
recognized: P. menziesii var. menziesii, called coastal
Douglas fir and found along the North American Pacific
Columbia), and P. menziesii var. glauca, called Rocky
Mountain or interior Douglas fir and found inland in the
mountains from British Columbia to central Mexico. Within
a region, Douglas fir can grow under a wide variety of
climatic conditions; in western Oregon and Washington,
it occurs from sea level to over 1700m.
Compared with many tree species, Douglas fir popula-
tions are generally regarded as being closelyadaptedtotheir
environments with relatively steep clines associated with
steep environmental gradients (Rehfeldt, 1994). Results
from nursery common garden studies with seedlings indi-
cate that genetic variation in growth, germination and bud
phenology follows a clinal pattern with steep clines mainly
occurring along elevational gradients, but also related to
aspect, slope, latitude, longitude and distance to the
ocean (Hermann and Lavender, 1968; Griffin and Ching,
1977; Campbell and Sorensen, 1978; Griffin, 1978;
Campbell, 1979, 1986; Rehfeldt, 1979, 1982, 1983, 1989;
Sorensen, 1983; Campbell and Sugano, 1993). Results after
25 years from a coastal Douglas fir provenance study,
Washington and British
* For correspondence. E-mail email@example.com
Annals of Botany 96: 1199–1214, 2005
doi:10.1093/aob/mci278, available online at www.aob.oxfordjournals.org
Published by Oxford University Press on behalf of the Annals of Botany Company 2005
by guest on June 5, 2013
however, seem to contradict the findings from seedling
genecology studies (White and Ching, 1985). Despite
significant differences among 14 provenances from a wide
geographical region planted at five sites from British
Columbia to California, the provenance by planting location
interaction was non-significant and small. Their study, how-
ever, suffers the same drawbacks of many provenance stud-
ies, namely a limited number of populations from a limited
number of source environments planted over a limited num-
ber of test sites. Small sample sizes make it difficult to study
the relationship between genetic variation and environmen-
tal differences where environmental gradients are complex
and highly heterogeneous, and extrapolation of results to a
wider range of environments is often not possible. Further-
more, seed for each population may have come from a large
area relative to areas within which considerable genetic
differentiation may occur (e.g. Campbell, 1979), resulting
in population buffering and a reduced likelihood offindinga
Results from field tests of tree improvement programmes
have also provided insight into the structure of genetic
variation of coastal Douglas fir. Stonecypher et al. (1996)
summarized genetic tests established on lands owned
by Weyerhaeuser Company in western Oregon and
Washington that were designed to explore questions of
that family-by-planting location interaction was small rel-
ative to family and planting location effects, and where
significant, was predominately the result of a few families.
Johnson (1997) considered genetic correlations among test
sites in six breeding zones in Oregon and concluded that
breedingzones were not particularly large giventhat site-to-
site genetic correlations were relatively strong and were
unrelated to the differences among sites in elevation, lati-
tude or longitude. In contrast to these results, Campbell
(1992), using a different analytical approach, found signifi-
cant family-by-site interaction in several breeding zones in
Oregon. Similarly, Silen and Mandel (1993) showed clinal
variation in height growth based on results from progeny
tests in two breeding zones in Oregon. All of these studies,
however, were designed with specific objectives of manag-
ing variation within tree improvement programmes, and,
therefore, were limited in geographical range and the
range of genotypes and environmental conditions sampled.
They were not designed to provide a systematic sample of
environmental conditions of seed sources and planting loca-
tionsacrossthe rangesofnaturalvariationineach toexplore
adequately the relationship between genetic variation and
The objectives of this study are: (1) to describe and map
patterns of genetic variation in coastal Douglas fir in west-
ern Oregon and Washington; and (2) to determine the envi-
ronmental factors that are most strongly related to genetic
variation in adaptive traits. The methodology for mapping
genetic variation relies on deriving a response surface in
which the response of a trait for a genotype from a source
location is a function of the environment at the location.
Environment is measured directly as climate or measured
indirectly by geographical or topographical variables.
Response surfaces are best modelled by sampling the
independent variables evenly across the range of interest.
This is accomplished by a systematic sample on a geo-
graphical grid with attention paid to sampling contrasting
elevations within the grid. Sample intensity depends upon
the environmental variation within a region. More samples
per area are required, for example, in the environmentally
heterogeneous westernNorthAmericaascomparedwith the
more homogeneous eastern or southern United States.
This approach to mapping genetic variation was developed
by Campbell (1979, 1986). His approach utilizes open-
pollinated seed from a single parent at most locations,
and duplicate samples at some locations in order to test
lack of fit of the models and to estimate family within-
types. His studies of coastal Douglas fir, however, were of
limited geographical range (Campbell, 1986; Campbell and
Sugano, 1993), and the models used geographical and topo-
graphical variables as surrogates for environments at source
locations. Climate models have since been developed that
provide reliable estimates of climate at each location
(Daly et al., 1994). Better climate data improve the models
of genetic variation as a function of the environments
at source locations, and provide greater insight into the
environmental factors that shape genetic structure through
natural selection. Furthermore, a larger geographical scale
is needed to explore the implications of climate change for
adaptation of Douglas fir to future climates. This study
considers genetic variation of growth and adaptive traits
as a function of environments over the whole range of
coastal Douglas fir in western Oregon and Washington
using a sampling strategy that is both extensive and inten-
sive. A geographical information system (GIS) is used to
display the response surface as a map of genetic variation.
MATERIALS AND METHODS
Sampling from natural populations
Wind-pollinated seed was collected from 1338 parent trees
of Douglas Fir [Pseudotsuga menziesii (Mirb.) Franco var.
menziesii] in naturally regenerated stands at 1048 locations
in western Oregon and Washington (Fig. 1). Most of the
seed was obtained from previous collections of the USDA
Forest Service, USDI Bureau of Land Management, Oregon
Department of Forestry and Northwest Tree Improvement
Cooperative. In 1993 and 1995, seed was collected from an
additional 231 parents to cover areas not sampled by the
other collections. The range of coastal Douglas fir in west-
ern Oregon and Washington was well sampled, although
sampling intensity was lower along the Washington coast
and in urban and agricultural areas around Puget Sound and
Willamette Valley. At 291 locations (28%), cones were
collected from two parent trees from the same elevation
and aspect, but separated by at least 100m. These paired
samples were used to estimate average variance among
families within locations and to test lack of fit to our gene-
The environments of seed source locations were charac-
terized using geographical, topographical and climatic data.
The geographical and topographical data were obtained
St Clair et al. — Genecology of Douglas Fir
by guest on June 5, 2013
from GIS coverages using a 90-m digital elevation model
(DEM). Variables included latitude, longitude, linear dis-
tancetothe sea,elevation, slope,aspect andsunexposure on
21 March (an integrative function of latitude, aspect, slope
and local topography). Climatic data were obtained from
Elevation Regressions on Independent Slopes Model), a
statistical–geographical model in which climate parameters
are predicted for 4 · 4km grid cells using localized regres-
sion equations of climate as a function of elevation with
greater weight given to climate data from nearby weather
stations of similar elevation and topographical position
(Daly et al., 1994; see www.ocs.orst.edu/prism/prism_new.
html). PRISM has been used extensively to map precipita-
tion and temperature in the United States, Canada and other
countries, and is particularly well suited to mountainous
from PRISM (Parameter-
terrain. Climate data are based on the averages for the
years 1961–1990. Climate values at specific parent tree
locations are determined as distance-weighted averages
of the four nearest grid cells using the LATTICESPOT
function with the bilinear interpolation option in ARC/
INFO. Climate variables included monthly, seasonal and
annual averages for minimum and maximum temperature,
precipitation, daily temperature fluctuation and aridity
(a ratio of precipitation to temperature); dates of 50%
probability of last spring frost and first autumn frost;
frost-free period; and seasonal ranges in temperature and
Common garden procedures
Breeding values of parent trees were estimated by grow-
ing seedlings in a common garden. Seeds were stratified at
3?C for approximately 60 d before sowing in April in raised
nursery beds in Corvallis, Oregon. To evaluate a large num-
ber of parent trees, tests were established in three successive
years (1994–1996) using different sets of families. Because
it was not possible to assign families to sowing years such
that sowing years contained an equivalent sampling across
the study area, 66 families from 50 well-distributed loca-
tions were includedin all three sowing years. These families
served as a genetic checklot to allow for adjustment of year
effects (White and Hodges, 1989; Rehfeldt, 1989). Each
year families were randomly assigned to five-tree row
plots (12cm between rows and 7cm between seedlings)
in each of four raised beds with each bed treated as a
block. In order to evaluate differences in rate of emergence,
four seeds were sown per position for a total of 20 seeds per
plot; seedlings were later systematically thinned to one per
Seedlings were grown for 2 years during which they were
measured for traits of emergence, bud phenology, growth
and partitioning (Table 1). Mean rate and standard deviation
of the rates of emergence were determined following pro-
cedures given in Campbell and Sorensen (1979) based on
the cumulative number of seedlings out of 20 that emerged
in a plot. Height and bud-set were measured at the end of the
first growing season. Bud-set was measured weekly as the
number of days since 1 January that terminal bud scales
were firstvisiblefollowinganysecondflushes.Atthe begin-
ning of the second growing season, bud-burst was measured
twice a week as the number of days since 1 January that
green needles were first visible emerging from the terminal
bud. At the end of the second growing season, bud-set was
again measured. Whole seedlings were harvested by care-
fully excavating the soil from around the roots. Seedlings
were measured for stem diameter (1cm above the root
collar), height from root collar to terminal bud, height to
the bud-scar resulting from second flushing and length of
the longest root. Dry weights of shoots and roots were
determined after drying the seedlings at 80?C for 24h.
To avoid problems with families from different regions
being tested in different years, year-to-year environmental
differences were removed from the data by standardizing
FI G. 1. Study area and locations of parents (grey dots).
St Clair et al. — Genecology of Douglas Fir
by guest on June 5, 2013
species that is commonly used to demarcate the varieties. A
study of the same transect using presumably neutral RAPD
markers showed a distinct boundary between varieties
coinciding with the break in the distribution (Aagaard
et al., 1995), which agrees with results from allozymes
(Li and Adams, 1989) and terpene composition (Zavarin
and Snajberk, 1973). In contrast to Douglas fir of the
Washington Cascades, those of the eastside Oregon
Cascades shares a genetic affinity with the coastal variety,
and probably migrated from glacial-age populations on the
One possibility is that the decreased growth and earlier
bud-set in the Washington transition may be an adaptation
to higher drought on the eastside, but this is not consistent
with the finding that seedlings from the eastside are also
smaller and set bud earlier for a given value of precipitation
or July aridity. Seedlings from the eastside Washington
Cascades are simply less vigorous, consistent with the
idea that they are of the interior variety. Estimates of pollen-
and seed-mediated gene flow from neutral DNA markers
could potentially shed light on the presence and nature of
barriers to gene flow across this narrow transition zone, as
well as provide evidence for or against the hypothesis of
Adaptation of Douglas fir populations to their environ-
ments appears to be largely a consequence of trade-offs
between selection for traits to avoid exposure to cold and
traits that confer high vigour in mild environments. Winter
temperatures are of greatest importance to population
differentiation. Selection for drought avoidance by early
bud-burst also appears to have resulted in population
differentiation. An important unanswered question arising
from this work is: what specific genetic and epigenetic
phenomena are responsible for geographical variation
observed in adaptive traits? To address this fundamental
question, parents from this study are currently being
genotyped at candidate genes presumably involved in
cold hardiness and drought tolerance. Associations between
DNA-level polymorphism and traits measured in this study
should provide a first step to elucidating the genes or path-
ways responsible for adaptive variation in Douglas fir
(Neale and Savolainen, 2004).
We thank the USDA Forest Service National Forests, USDI
Bureau of Land Management, Oregon Department of For-
estry and Northwest Tree Improvement Cooperative for
providing seed for this study. We are grateful to Jeff Riddle
for his care in raising seedlings and data collection. Richard
Cronn, Glenn Howe, Tim Max, Frank Sorensen and Cathy
Whitlock provided helpful comments on earlier versions of
DNA productsamong RAPDprofilesare frequentand stronglydiffer-
strategies for conifer cold hardiness. In: Bigras FJ, Columbo SJ, eds.
Conifer cold hardiness. Dordrecht: Kluwer Academic Publishers,
Cascades. Ecology 60: 1036–1050.
transfer in southwest Oregon. Silvae Genetica 35: 85–96.
Campbell RK. 1992. Genotype*environment interaction: a case study for
Douglas fir in western Oregon. Research Paper PNW-RP-455.
Portland, OR: US Department of Agriculture, Forest Service, Pacific
Northwest Research Station, 21 pp.
Campbell RK, Sorensen FC. 1978. Effect of test environment on the
expression of clines and on delimitation of seed zones in Douglas
fir. Theoretical and Applied Genetics 51: 233–246.
Campbell RK, Sorensen FC. 1979. A new basis for characterizing ger-
mination. Journal of Seed Technology 4: 24–34.
Campbell RK, Sugano AI. 1979. Genecology of bud-burst phenology in
Douglas fir: response to flushing temperature and chilling. Botanical
Gazette 140: 223–231.
Campbell RK, Sugano AI. 1993. Genetic variation and seed zones of
Douglas fir in the Siskiyou National Forest. Research Paper PNW-
Pacific Northwest Research Station, 23 pp.
Wiley & Sons.
Cressie NAC. 1991. Statistics for spatial data. New York: John Wiley &
Critchfield WB.1984. Impactof the Pleistoceneon the geneticstructureof
North American conifers. In: Lanner RM, ed. Proceedings of 8th
North American Forest Biology Workshop. Logan, UT: Utah State
mapping climatological precipitation over mountainous terrain.
Journal of Applied Meteorology 33: 140–158.
Endler JA. 1986. Natural selection in the wild. Princeton, NJ: Princeton
Haddock PG, Walters J, Kozak A. 1967. Growth of coastal and interior
provenances of Douglas fir (Pseudotsuga menziesii (Mirb.) Franco) at
Vancouver and Haney in British Columbia. Faculty of Forestry
Research Paper No. 79. Vancouver: University of British Columbia.
Hermann RK. 1985. The genus Pseudotsuga: ancestral history and past
distribution. Special Publication 2b. Corvallis, OR: Forest Research
Laboratory, Oregon State University.
Hermann RK, Lavender DP. 1968. Early growth of Douglas fir from
various altitudes and aspects in southern Oregon. Silva Genetica 17:
Department of Agriculture, Forest Service, 527–540.
Heslop-Harrison J. 1964. Forty years of genecology. In: Cragg JB, ed.
Advances in ecological research. Vol. Two. New York: Academic
2003. From genotype to phenotype: unraveling the complexities of
cold adaptation in forest trees. Canadian Journal of Botany 81:
Griffin AR. 1978. Geographic variation in Douglas fir from the coastal
ranges of California. II. Predictive value of a regression model for
seedling growth variation. Silvae Genetica 27: 96–101.
GriffinAR,ChingKK. 1977.Geographicvariation in Douglasfirfrom the
coastal ranges of California. I. Seed, seedling growth and hardiness
characteristics. Silvae Genetica 26: 149–157.
Johnson GR. 1997. Site-to-site genetic correlations and their implications
on breeding zone size and optimum number of progeny test sites for
coastal Douglas fir. Silvae Genetica 46: 280–285.
Johnson GR. 2002. Genetic variation in tolerance of Douglas fir to Swiss
needle cast as assessed by symptom expression. Silvae Genetica 51:
Li P, Adams WT. 1989. Range-wide patterns of allozyme variation in
Douglas fir (Pseudotsuga menziesii). Canadian Journal of Forest
Research 19: 149–161.
St Clair et al. — Genecology of Douglas Fir
by guest on June 5, 2013
tant hardwoods. Miscellaneous Publication No. 1146. Washington,
DC: US. Department of Agriculture, Forest Service.
Mardia KV, Kent JT, Bibby JM. 1979. Multivariate analysis. London:
Morgenstern EK. 1996. Geographic variation in forest trees. Vancouver:
Neale DB, Savolainen O. 2004. Association genetics of complex traits in
conifers. Trends in Plant Sciences 9: 325–330.
Neter J, Wasserman W. 1974. Applied linear statistical models.
Homewood, IL: Richard D. Irwin, Inc.
State of Oregon Department of Forestry.
Randall WK, Berrang P. 2002. Washington tree seed transfer zones.
Olympia, WA: Washington State Department of Natural Resources.
Rehfeldt GE. 1977. Growth and cold hardiness of intervarietal hybrids of
Douglas fir. Theoretical and Applied Genetics 50: 3–15.
Rehfeldt GE. 1979. Ecological adaptations in Douglas fir (Pseudotsuga
menziesii var. glauca) populations. I. North Idaho and north-east
Washington. Heredity 43: 383–397.
Rehfeldt GE. 1982. Ecological adaptations in Douglas fir populations. II.
Western Montana. Research Paper INT-295. Ogden, UT: US Depart-
ment of Agriculture, Forest Service, Intermountain Forest and Range
Experiment Station, 8 pp.
Rehfeldt GE. 1983. Ecological adaptations in Douglas fir (Pseudotsuga
menziesii var. glauca) populations. III. Central Idaho. Canadian
Journal of Forest Research 13: 626–632.
Rehfeldt GE. 1989. Ecological adaptations in Douglas fir (Pseudotsuga
menziesii var. glauca): a synthesis. Forest Ecology and Management
Rehfeldt GE. 1994. Evolutionary genetics, the biological species, and the
ecology of the Interior cedar–hemlock forests. In: Proceedings of
Pullman, WA: Washington State University, 91–100.
von Rudloff E. 1973. Chemosytematic studies in the genus Pseudotsuga.
III. Population differences in British Columbia as determined by
volatile leaf oil analysis. Canadian Journal of Forest Research 3:
SAS Institute Inc. 1999. SAS/STAT?user’s guide, Version 8. Cary, NC:
SAS Institute Inc.
on the population genetic structure of Douglas fir (Pseudotsuga
menziesii) in the Great Basin. Canadian Journal of Forest Research
Silen RR, Mandel NL. 1993. Clinal genetic growth variation within two
Douglas fir breeding zones. Journal of Forestry 81: 216–220.
Sorensen FC. 1979. Provenance variation in Pseudotsuga menziesii seed-
lings from the var. menziesii—var. glauca transition zone in Oregon.
Silvae Genetica 28: 37–119.
suga menziesii) from the western Siskiyou Mountains of Oregon.
Ecology 34: 696–702.
pole pine in central Oregon. Research Paper PNW-RP-453. Portland,
OR: US Department of Agriculture, ForestService,Pacific Northwest
Research Station, 30 pp.
Results from genetic tests of selected parents of Douglas fir (Pseudot-
suga menziesii (Mirb.) Franco) in an applied tree improvement
program. Forest Science 42: Monograph 32: 1–35.
sal and late Quaternary history in the Pacific Northwest. Japanese
Journal of Ecology 32: 159–187.
Wartenberg D. 1985. Canonical trend surface analysis: a method for
describing geographic patterns. Systematic Zoology 34: 259–279.
Wells PV. 1983. Paleobiogeography of montane islands in the Great
Basin since the last glaciopluvial. Ecological Monographs 53:
WhiteTL, ChingKK. 1985.Provenancestudy ofDouglasfir in the Pacific
ica 34: 84–90.
in forest tree improvement. Dordrecht: Kluwer Academic Publishers.
White TL, Ching KK, Walters J. 1979. Effects of provenance, years, and
planting location on bud burst of Douglas fir. Forest Science 25:
America during the past 125 kyr. Nature 388: 57–61.
Worona MA, Whitlock C. 1995. Late Quaternary vegetation and climate
history near Little Lake, central Coast Range, Oregon. Geological
Society of America Bulletin 107: 867–876.
Zavarin E, Snajberk K. 1973. Geographical variability of monoterpenes
St Clair et al. — Genecology of Douglas Fir
by guest on June 5, 2013