Biodiversity and Conservation $, 779-794 (1996)

Genetic variation in low elevation Douglas-fir of

British Columbia and its relevance to gene
conservation*
YOUSRY A. EL-KASSABY
Pacific ForestProducts Limited, Saanich Forestry Centre, Saanichton, BC, V8M 1K1, and Faculty of Forestry,
University of British Columbia, Vancouver, BC, V6T 1Z4, Canada

KERMIT RITLAND
Department of Botany, Universityof Toronto, Toronto, Ontario, MSS 3B2, Canada

Received 7 August 1994; revised and accepted 4 May 1995

Patterns of variation were studied at 20 isozyme loci in 49 coastal, low-elevation Douglas-fir

populations in SW British Columbia and NW Washington State. Several components of variation
were estimated for each population including the number of alleles per locus N,, number of alleles
per polymorphic locus Na(951,inbreeding F,,heterozygosity H, and population divergence D. F was
near zero indicating nearly complete outcrossing within populations. H was quite high (16%) and in
accord with previous studies of Douglas-fir. D values were low (equivalent to Wrights Fsr of 0.08)
indicating levels of gene flow sufficient to largely homogenize populations. The parameters of
diversity No, N,~95~, H, and D showed little intercorrelation across populations. A homogenous
pattern of genetic relationship among populations was shown by the clustering of populations based
on their inferred relationship, and by the principal components of the matrix of inferred genetic
relationship. Because of the complex nature of gene diversity and the continuous nature of
population differentiation in Douglas-fir, it is difficult with isozyme markers to identify specific
populations of value for genetic conservation in this species.
Keywords: genetic variation; isozymes; Douglas-fir; genetic conservation

Introduction
The discovery of protein electrophoresis by Hunter and Markert (1957) and its
introduction to population genetics by Lewontin and Hubby (1966) has led to the
accumulation of vast amounts of information about the levels and structuring of allozyme
variation (Nero, 1978; Hamrick et al., 1979). In concert, a number of methods have
accumulated to characterize patterns of isozyme variation, including Wright's F statistics
(Wright, 1965), Nei's diversity statistics (Nei, 1973) and several measures of 'genetic
distance' (Nei, 1972; Hedrick, 1975; Reynolds et al., 1983). In forestry, allozymes and their
statistics have been useful for genetic studies of both experimental and natural populations
(Mitton, 1983; Ledig, 1968; Hamrick and Godt, 1989; EI-Kassaby, 1990).
Conifers are regarded as one of the most genetically variable group of species: in most

*This paper is dedicatedto J.C. Heaman on the occasionof his retirement.

0960-3115 © 1996Chapman & Hall
780 El-Kassaby and Ritland
species, isozyme heterozygosity is at least 15%, and normally at least 90% of the genetic
variation is found within a population (Hamrick and Godt, 1989). The life history of
conifers - long generations, large population sizes, wind-pollinated mating systems,
pollination mechanism, high fecundity, and long distance seed and pollen dispersal - are
likely responsible for this pattern of variation (Hamrick and Godt, 1989).
Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) is one of the most widely distributed
conifer species in the western region of the North American continent (Fowells, 1965). The
species has two recognized varieties (coastal: var menziesii and interior: var glauca)
(Fowells, 1965). The menziesi variety range extends from Vancouver Island and the coastal
mainland of British Columbia to northern California. In the coastal regions of British
Columbia, Douglas-fir is considered to be one of the most important commercial forest
tree species. Due to its high productivity, desirable wood quality, tree form, and growth
rate, Douglas-fir is being used extensively in reforestation, and programmes for selection
and breeding of Douglas-fir are underway (Heaman, 1985).
Gene conservation efforts have traditionally been focused on rare and endangered
species. The vulnerability of genetic variation in widespread species like Douglas-fir has
been less studied. As in most conifers, the majority of genetic variation resides within
populations of Douglas-fir, but there is small but significant component among
populations (GsT ranging between 1.8 and 7.1, see Yeh and O'MaUey, 1980; E1-Kassaby
and Sziklai, 1982; Li and Adams, 1989; Moran and Adams, 1989). Thus, despite the nearly
homogenous genetic structure of this species, there are probably populations with unique
attributes of value for genetic conservation of Douglas-fir.
In this study, the amount and pattern of variation in Douglas-fir is characterized, with
the twin goals of establishing a baseline for genetic variation in Douglas-fir and
determining an effective conservation strategy that protects the evolutionary potential of
this species. We sample 49 populations within the coastal low-elevation range of the
species in British Columbia. This area is subject to the highest rate of harvest and
reforestation, and hence is where gene conservation needs the most critical evaluation. We
estimate heterozygosity and polymorphism at isozyme loci, as well as 'population
divergence', a new measure of divergence which is essentially a component of Wright's V~~.
In addition, to provide a complete characterization of variation among populations, we
describe the distribution of alleles among populations, and perform both cluster analyses
and principle components analyses on the matrix of genetic relatedness. We conclude that
genetic variation is a complex quantity not totally represented by any single population or
selected group of populations, implying that complete maintenance of variation via the
preservation of a select group of populations is a difficult task. We also discuss the
problems with identifying rare alleles and the general limitations of isozymes for
characterizing genetic variation.

Materials and methods

Plant materials'
Dormant vegetative buds were collected from 49 populations of the Province of British
Columbia Ministry of Forests' Douglas-fir provenance trials (Illingworth, 1978).
Provenance trials are large-scale field testing experiments which include several seed
sources from the natural range of a species. They are planted over several test sites in order
to investigate genotype × environment interaction and the suitability of seed source
Genetic variation in low elevation Douglas-fir 781
transfer (i.e. the matching of a group of genotypes to specific sites) for artificial
regeneration of forest tree species. This particular provenance trial included 88
populations from the Douglas-fir Pacific Northwest range and planted over 22 test sites.
The samples for this study came from Sooke (Mure Creek) and Mud Bay, on Vancouver
Island (Illingworth, 1978). A total of 2497 trees were sampled and the number of trees per
population ranged between 43 and 61 (average 51 trees). The location and names of the
sampled populations are presented in Fig. 1. Samples were then placed in plastic bags and
stored in shade or on ice to prevent protein deterioration. The branch samples were
shipped to the laboratory and stored at 2 ° C until protein extraction.
Newly-developed vegetative bud primordia were removed from the bud scales and
proteins were extracted using a slightly modified extraction buffer of Cheliak and Pital
(1984). Samples were electrophoresed on 11% horizontal starch gels using four gel-
electrode buffer systems. The buffer systems used were: Histidine citrate (pH 7.0) (Fildes
and Harris, 1966), morpholine citrate (pH 6.1) (Clayton and Tretiak, 1972), Tris citrate:
lithium borate (pH 8.5) (Ridgeway et al., 1970), and Tris citrate (pH 7.0) (Siciliano and
Shaw, 1976). A total of 20 loci were resolved for 12 enzyme systems: phosphoglucose
isomerase (PGI1, PGI2) E.C. 5.3.1.9, phosphoglucomutase (PGM1, PGM2) E.C. 5.4.2.2,
fructose diphosphatase (FDP2) E.C. 3.1.3.11, 6-phosphogluconie dehydrogenase (6PGD 1,
6PGD2) E.C. 1.1.1.44, shikimate dehydrogenase (SKDH1) E.C. 1.1.1.25, aspartate
amino-transferase (AAT1, AAT2) E.C. 2.6.1.1, malate dehydrogenase (MDH1, MDH2,
MDH3) E.C. 1.1.1.37, leucine aminopeptidase (LAP1, LAP2) E.C. 3.4.11.1, aconitase
(ACO1, ACO2) E.C. 4.2.1.3, isocitrate dehydrogenase (IDH) E.C. 1.1.1.42, glutamate
dehydrogenase (GDH) E.C. 1.4.1.2, and glucose-6-phosphate dehydrogenase (G6P) E.C.
1.1.1.49. The staining methods used followed those of Conkle et al. (1982) and O'Malley et
al. (1980). The Mendelian inheritance of the studied loci has been previously confirmed
(E1-Kassaby, 1981; EI-Kassaby et al., 1982a, b; Neale et al., 1984; Adams et al., 1990).

Data analyses
Gene and genotypic frequencies were compiled for each of the 49 populations. For each
population, the following were calculated: (1) the proportion of polymorphic loci (PLP)
for the 95% criterion, (2) the average number of alleles per locus N a, (3) the average
number of alleles per polymorphic locus Na~95~, (4) Wright's inbreeding coefficient F, (5) the
expected heterozygosity H and (6) population divergence D. A locus is polymorphic by the
95% criteria if the frequency of the most common allele is less than or equal to 95%.
Wright's F measures the excess homozygosity caused by inbreeding (primarily self-
fertilization), and ranges from 0 to 1 corresponding to complete outbreeding to complete
inbreeding. This was estimated using maximum likelihood which guarantees a minimum
variance estimate. The expected heterozygosity H is the probability that two randomly
chosen alleles, both from the same locus but different individuals, are different. For each
locus, it was estimated as 1 - piz, for Pi the frequency of allele i; this quantity was averaged
over loci to obtain H. The standard error of H was found using the methods of Nei and
Royehoudhury (1974), In addition, Nei's gene diversity statistics (Nei, 1973) were
computed.
The 'population divergence' D is defined as the excess homozygosity (gene identity) in a
population caused by genetic divergence (through drift and selection, but not mutation)
from a larger, 'reference' population. For population i, over all loci it is estimated by the
formula Di= cEjk (pZk--p2k)/pjk, where Pijk is the frequency of allele k at locus ] in
 782 El-Kassaby and Ritland

I I I I 53o~
127 ° 125° 123° 121°

LOCATION OF"
DOUGLAS-FIR
POPULATIONS

30 0 30 60 120kin
I I I I I
SCALE

--51o BRITISH COLUMBIA 51o~

(CANADA) I

-'0

C~

--49 °
KEI~ 49o--

~" ~'~,~ ~iii~iiiiiiii~ WASHINGTON

Po,,n, rre, Loco,o. ~iiiiiiiiiiiiiiiiiii~ (U.S.A.I
0 Population
--47 °
47 ° -

127° 125° 125° 121o

I I I I

Figure 1. Location of the 49 Pseudotsuga menziesii populations sampled for this study.

population i, PI~ is the frequency of that allele in the ensemble of populations, and
c = [E (nj - 1)] -1 for nj the number of alleles at locus j. To avoid bias due to sampling of a
finite number of populations, i was excluded from Pkt in computing D i. Also, alleles outside
of the frequency range of 0.01--0.99 (as a mean over all populations) were excluded because
such alleles incur sampling problems. However, another source of bias, the divergence of
the reference population from the ancestor of all contemporary populations, cannot be
removed. If this quantity is denoted Dr, its presence would bias downwards the true
Genetic variation in low elevation Douglas-fir 783
divergence of population i by the amount (1 - D~)D,. We expect that, given our large
ensemble of 49 populations, that D r would be small relative to D i.
The expected value of Di equals Wright's Fsr, so in essence, Di is the contribution of a
specific population i (or subdivision) to Fsr. This quantity has not been previously
measured for two reasons. First, the classic model of population subdivision assumes
equality of population sizes, so that populations have equal divergence, therefore making
average divergence, characterized by Fsr, a sufficient quantity to characterize populations.
Second, the appropriate statistic has not been developed. We expect that FsT varies
primarily because population size varies due to variation in the number of immigrants into
populations. Variable selection may also be responsible, but isozyme markers are
generally regarded to be largely neutral.
For each pair of populations i and l, the pairwise genetic relationship was likewise
estimated as Rit = C Ejk (PijkPtjk--P~k)/Pjk, where the above definitions hold (again, to
eliminate bias, populations i and I were excluded from the estimation of Pjk in Ru). This
formula is also a minimum-variance, linear estimator of relationship, and estimates the
probability of identity-by-descent of two homologous alleles sampled at random, one from
each population, relative to the ensemble of 49 populations.
Classically, workers have used Nei's genetic distances (Nei, 1972) to portray
relationships among populations of a species. However, Nei's distance measure is based on
a model of divergence via mutation and not drift. The initial stages of divergence such as
those separating populations of a species are primarily due to drift and selection; mutation
is a slow process that has a significant effect only in species comparisons. Another distance
measure, proposed by Reynolds et al. (1983), is designed to measure the same 'short-term'
divergence due to genetic drift as our measure of population relatedness does. The
properties of our estimator of population divergence and relatedness will be documented
in Ritland (manuscript in preparation).
To portray the structure of genetic relatedness among the 49 populations, two
multivariate analyses were performed. First, a dendrogram was constructed using the
unweighted pair group method (UPGMA, Sheath and Sokal, 1973) applied to the matrix
of inferred relationship. Second, principal components analysis (PCA) was performed on
the matrix of inferred relationship. This analysis (an 'ordination') is expected to be more
appropriate for populations of a species, since gene flow tends to convert the genealogical
tree into a network of relatedness. For this analysis, a 56 × 49 data matrix was used which
consisted of elements (P~k--Pk)/(Pk( 1 --Pk)) for i = 1-49 (populations) and k = 1-56
(alleles among loci, excluding rare and fixed alleles); the product with its transpose gave a
49 × 49 matrix of relationship between populations, from which the principal components
of relationship were extracted. This was achieved via the Karhunen-Loeve expansion,
using a FORTRAN program written by C.F. Murtagh (ST-ECF/ESA/ESO), obtained
from the Carnegie-Mellon University experimental StatLib gopher server (at
lib.stat.cmu.edu).

Results
Over all populations, the total genetic diversity H r (expected heterozygosity of pooled
populations) was 0.163 (SE = 0.002). Seventeen of the 20 loci studied (85%) revealed
detectable polymorphism in at least one population. The remaining three loci were
monomorphic in all populations. A total of 61 alleles were observed in this study (allele
784 El-Kassaby and Ritland
frequencies are available upon request). Eight of the polymorphic loci expressed three
alleles, while the remaining loci had either five (1 locus), four (6 loci), or two (2 loci) alleles.
Figure 2 gives the distribution of allele frequencies. This distribution gives the
probability that a randomly chosen allele lies within a certain frequency range (the chance
of choosing the allele is proportional to its frequency). It shows that the probability that a
randomly chosen allele is fixed (an allele at a monomorphic loci) is high: 47%. There is a
long tail of lesser frequent alleles, with a pronounced bump in the 0.1-0.2 frequency range,
suggesting an excess of 'rarer' alleles. Figure 3 gives the partitioning alleles among
populations. Three alleles are unique to single populations (e.g. are 'private' alleles), and
two alleles are unique to a pair of populations; 11 alleles were found in 10-40 populations,
while 25 alleles are found in all 49 populations.
The proportion of total genetic diversity among populations (Nei's Gsr) was 0.045
(SE = 0.010), or in other words, about 5% of the total genetic diversity (Hr) is due to gene
frequency differences among populations. The standard error for the GsT indicates that
this amount of subdivision is statistically significant.
Table 1 gives estimates of proportions of polymorphic loci (PLP), average number of
alleles per locus (No), average number of alleles per polymorphic locus (N,~95~),inbreeding
coefficients (F), expected heterozygosity (H) and population divergence (D). Over all
populations, an average of 52.6% (sE = 0.010) of the loci were polymorphic (95%
criterion), with individual population values ranging from 40 (four populations) to 70%
(Hob, USA).
Across these populations, the inbreeding coefficient F ranged from -0.557 (Texada Is.)
to 0.135 (Haney). The average F value was non-significantly positive (0.008, SE = 0.015),
suggesting that essentially no selfing occurs. However, positive F values were estimated in
39 of 52 populations, which has a probability under the null hypothesis of F = 0 of about
0.01 (Z2 = 6.5, 1 d.f.). Thus, some inbreeding is present; apparently a skewed distribution of
F values results in a near zero mean.

0.5

0.4

0.3!
.O
°
0 0.2
Q- ;
I
0.1 1
4

0.0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Frequency of allele
Figure 2. The distribution of allele frequencies. On the Y axis is the probability that a
randomly chosen allele (irrespective of locus) lies within the frequency specified on the
X axis.
Genetic variation in low elevation Douglas-fir 785
° ~

0
0 25
0
CXl 20
t-
O 15
E
t~
~" 10
09

d 0 10 20 30 40
Z
No. of populations that have the allele
Figure 3. The partitioning of alleles among populations. On the Y axis is the number of
alleles found in the number of populations specified on the X axis. (notice: two alleles
were found in no natural populations; they were found exclusivelyin the seed orchard
populations studied by E1-Kassaby and Ritland, 1996).
Across the 49 populations, the average number of alleles per locus was 2.143 (SE = 0.015)
and varied between 1.75 (Kennedy Lake) and 2.35 (Mead Cr.). The average number of
alleles per polymorphic locus was 2.691 (SE = 0.181) and varied between 2.31 (Klinaklini
Low) and 3.11 (Tinhat Mtn.). This estimate (Na(95)) is considered to be a more informative
parameter for the number of alleles per locus since it is not confounded by the proportion
of monomorphic loci as in N a. The expected heterozygosity H ranged from 0.122
(Darrington, USA) to 0.198 (Port Hardy) and averaged 0.163 (SE = 0.004). The population
divergence values D ranged from 0.032 (Parksville) to 0.369 (Kennedy Lake), with a mean
of 0.077 (SE = 0.007). The interrelationships among these three major measures of
diversity, N,, H and D, are plotted in Fig. 4. Expected heterozygosity H did show a positive
correlation with the number of alleles N,, but D was not correlated with H, nor was D
correlated with N,. As one would expect, PLP95 showed strong intercorrelation with N,
(data not shown).
The hierarchical structure of genetic relatedness among the 49 populations is presented
in Fig. 5. The dendrogram shows lack of genetic grouping in relation to geography (see
Fig. 1, for the geographic location of the 49 populations). The clustering of populations is
almost continuous, with no major groupings evident (the clustering of Nei's genetic
distances also showed the same pattern, with no statistically significant clusters found,
based upon the method given in Ritland, 1989). Estimates of genetic relatedness between
populations plotted against their physical distance (distances were classified into seven
categories of equal distance increments) produced a slight correlation of genetic
relationship with physical distance, with populations of only the closest distance category
showing a significant relationship (these are populations within - 5 0 kin). There is some
indication that, over all categories, relatedness decreases with an increase in distance,
which is the direction expected.
The principal components analysis further demonstrated the lack of structure for
786 El-Kassaby and Ritland
Table 1. Parameters of isozyme diversity for the 49 populations: expected heterozygosity H,
proportion of polymorphic loci (PLP, 95% criterion), number of alleles per locus Na, number of
alleles per polymorphic loci (95% criterion) Na~osrinbreeding coefficients F, population divergence
D

Population PLP95 N, N~t95~ F H D

Alberni 0.55 2.25 2.73 0.088 0.153 0.037
Bowen Island 0.55 2.25 2.73 0.035 0.162 0.061
Cassidy 0.45 2.05 2.67 0.054 0.146 0.043
Caycuse 0.50 2,25 2.80 0.022 0,186 0.096
Chehalis 0.50 2.15 2.70 0.086 0.162 0.058
Chilliwack High 0.60 2.25 2.75 0.027 0.159 0.054
Chilliwack Low 0.50 2.20 2.80 0.024 0.158 0.080
Courtenay 0.40 2.20 3.00 - 0.084 0.144 0.030
Darrington USA 0.40 2.15 2.88 0.035 0.122 0.060
Denman Island 0.55 2.25 2.73 0.051 0.188 0.094
Duncan 0.60 2.15 2.50 0.037 0.169 0.068
Empress Mtn. 0.50 2.15 3.00 -0.061 0.181 0.050
E. Thurlow 0.65 2.25 2.69 - 0.026 0.164 0.040
Forbidden Plat. 0.50 2.30 2.90 0.083 0.177 0.082
Franklin R. 0.50 2.10 2.60 0.055 0.151 0.064
Gold River 0.50 2.10 2.50 -0.085 0.152 0.094
Granite Falls USA 0.40 2.10 2.88 - 0.030 0.139 0.078
Haney 0.55 2.10 2.64 0.135 0.160 0.049
Harrison Lake 0.55 2.05 2.55 0.057 0.166 0,039
Hernando Is. 0.65 2.05 2.46 0.118 0.181 0.081
Hoh, USA 0.70 2.30 2.79 0.016 0.181 0.094
Hope 0.50 2.10 2.80 -0.132 0.126 0.051
Jevis 0.50 2.15 2.90 -0.045 0.164 0.191
Jeune Landing 0.55 2.15 2.55 -0,170 0.185 0.106
Jordan River 0.55 2.20 2.91 0.002 0.t69 0.118
Kaouk River 0.55 2.15 2.64 0.062 0,189 0.071
KIinaklini High 0.55 2.00 2.45 0.061 0.190 0.086
Klinaklini Low 0.65 2.00 2.31 0.046 0.157 0.050
Kelsey Bay 0.55 2.10 2.64 -0.017 0.160 0.075
Kennedy Lake 0.45 1.75 2.44 0.097 0.138 0.369
Malcolm Island 0.40 1.90 2.38 0.053 0.143 0,116
Mead Cr. 0.50 2.35 3.00 0.048 0.181 0,067
Nimpkish 0.65 2.15 2.54 0,007 0.169 0,058
Nanaimo L. 0.45 2.15 2.78 0.050 0.166 0.070
Parksville 0.55 2.20 2.55 -- 0.081 0.146 0.032
Pitt River 0.50 2.05 2.60 0,024 0,171 0.055
Port Hardy 0.55 2.20 2.73 0.001 0.198 0.141
Powell River 0.50 2.15 2.70 0.037 0.161 0,070
Quadra Island 0.65 2.05 2.46 0.034 0.167 0,047
San Juan R. 0.50 2.15 2.90 0.065 0.157 0.071
Sechelt 0.50 2.05 2.40 0.009 0.158 0.035
Shelton, USA 0.45 2.10 2.67 -0.010 0.148 0.054
Sooke 0.55 2.20 2.82 0.085 0.161 0.076
Squamish 0.55 2.20 2.64 0.012 0.169 0.045
Stella Lake 0.45 2.15 2.67 0,096 0.160 0.070
Tahsis 0.50 2.15 2.60 -0.108 0,154 0.092
Texada Island 0.55 2.05 2.55 - 0.557 0.171 0.050
Tinhat Mountain 0.45 2.25 3.11 0.057 0.163 0.046
Toba River 0.55 2.25 2.82 0,039 0.172 0,084
Mean 0.526 2.143 2.691 0.008 0,163 0.077
SE among populations 0.010 0.015 0.181 0.015 0.002 0.007
Genetic variation in low elevation Douglas-fir 787

~_ 0.22 0.24
U
0

"~ 0.18 • :,:. 0.16

J
@
"o
I:
o ., ::
= 0.14
0 •• •
~. 0.08
Q.
0
":1;|
io
e •

X a.
0.10 , , 0.~ i i i

0.00 0.08 0.16 0.24 1.6 1.8 2.0 2.2 2.4

Population divergence D Numberofalleles

, 0.22

i 0.18 • Ii' s .
• | •
• e
0.14

0.10 I I I

1.6 1.8 2.0 2.2 2.4

Number of alleles
Figure 4. Interrelationships of number of alleles, gene diversity, and population divergence among
the 49 natural populations of Douglas-fir.

genetic relationship. The first two principal components only accounted for about 30% of
the total variance, and in general, the proportion of variance explained by successive PC
axis was small, with only a gradual increase. In fact, ten principal component axes are
needed to account for 90% of the variance, indicating a very complex pattern of
relationship. In Fig. 6, the location of each population is projected onto the first two
principal component axes. No correlation with geography is evident (see Fig. 1).

Discussion

A high amount of genetic diversity exists at the species level for Douglas-fir. The percent of
polymorphic loci in the studied 49 populations was 85%, which is higher than that of
gymnosperms (71.1 __-2.6), temperate-zone species (63.5 _+2.3), species wind-pollinated
outcrossing mating system (53.0 ___2.2), early succession species (56.9 + 11.6), and those
with widespread geographic range (74.3 ___7.7) (Hamrick et al., 1992). Its average number
of alleles per locus was 2.14; this value is also higher than that of gymnosperms
(1.83 _+0.58), temperate-zone species (1.81 ± 0.06), wind-pollinated outcrossing mating
788 EI-Ka~abyand Ritland

~ - ~ N a n i a m o L.
Sechelt
Hoh USA

ii __U______
' ~ Franklin R.
_ f~-T-a-h-s-i s Denman island
"-~Forbidden Plat.

Duncan
Klinaklini High
~Mead Cr.
. ----Jordan River

~~~ .
. ~
~
-~Caycuse
old River
Cassidy
Tinhat Mtn.
~ Sooke
-~- San Juan R.
H a r r i s o n Lake
Squamish
Jervis
Granite Falls USA
Parksville
Hope
Darrington USA
~ Klinaklini Low
L ~ E. Thurlow
Stella Lake
~ P o w e l l River
Port Hardy
Alberni
~ Kennedy Lake
Courtenay
~~-.~Chilliwack Low
U
Shelton USA
-~~-Empress Mtn.

~ ~ ' " ~ C h e h a l i s
Toba River
~ Texada Is.
. ~ . ~ Malcolm Is.
Bowen Island
~ Pitt River
- - ~ Kelsey Bay
. ~ Quadra Is.
~~~ - ~ Hernando
Kaouk River
Is.

~ Jeune Landing
~gare 5. Dendrogramofin~rred geneticrelationshipamongthe 49 natural populations.
constructedviatheUWPGM.
Genetic variation in low elevation Douglas-fir 789
0.4
Jeune Landing

Denman Island
Malcolm Is.

Gold River
0.2 - Jordan River Kennedy Lake
Mead Cr. Pitt R ~ y Bay
Hemando Is.
Texada Is.
t-
Forbidden ..~uadra Is.
O KlinakliniL t ~"ouk H~ver
Q.
E Cassidy Albeml~hilliwad~.~l Island
o
Nimpkish E. Thudow
Sechelt
Q. 0.0 - Klinaklini Hi SquamishChehalis
Stslla Lake

c" Haney
L,.
Caycuse Franklin R.Naniam° ILiohUSTAOt~aRiver
Q.
Chilliwack Hi
Harrison Lake Hope
0
0 Darrington USA

Sooke Courtenay Port Hardy

if)
ParksvilleTinhat Mtn.
o -0.2 -
t-
Shelton USA
O
o
Empress Mtn.
Gm. Falls USA Powell River
13_

-0.4 -

Jen/is

-0.6 I I I I I

-0.8 -0.4 0.0 0.4 0.8

Projection of first principle component
Figure 6. Projections of the 49 populations in the first two dimensions of principal components.
790 EI-Kassaby and Ritland
system (1.84 _ 0.05), and early succession species (1.75 _+ 1.9), but lower than species with
widespread geographic range (2.56 -+ 0.31) (Hamrick et al., 1992).
The expected heterozygosity (H) of the studied populations (0.163 +_0.002) is slightly
higher than that of gymnosperms (0.151 ___0.008), early succession species (0.137 _ 0.031),
similar to temperate-zone species (0.166_ 0.008), but lower than those with wind-
pollinated outcrossing mating system (0.173 _+0.07), and widespread geographic range
(0.257 ___0.039) (Hamrick et al., 1992). Comparing the same heterozygosity parameters
(per cent of polymorphic loci, average number of alleles per locus, and expected genetic
diversity) of the studied 49 populations with that reported from other Douglas-fir studies
produced similar results indicating that our samples are a good representation of the
species (Yeh and O'Malley, 1980; EI-Kassaby and Sziklai, 1982; Markel and Adams, 1987;
Li and Adams, 1989; Moran and Adams, 1989).
In the 49 populations studied, the proportion of total heterozygosity due to population
differentiation, GsT, amounted to a significant 0.045. This value is lower than that reported
for gymnosperms (0.073 ___0.010), temperate-zone species (0.092 _+0.010), species with
wind-pollinated outcrossing mating system (0.077 _+0.009), early succession species
(0.095 _+0.020), and those with widespread geographic range (0.033 + 0.033) (Hamrick et
al., 1992). This low value of Gsr reflects a lack of spatial genetic differentiation reflecting
the large geographic scale represented in the present study compared to those reviewed by
Hamrick et al. (1992) and also is indicative of extensive gene flow via pollen and/or seeds.
These factors reduce the effect of geographic isolation of breeding populations, thus
decreasing the chance for genetic divergence. Since Gsr is equivalent to Wright's FsT, one
can use the classic relationship FsT = 1/(1 + 4Nm) to estimate the average number of
immigrants per population per generation as N m - - (1--GsT)/(4GsT ) producing an
estimate of 5. This is a very high gene-flow estimate (see Ellstrand, 1992), and indicates that
pollen and seed dispersal is substantial in Douglas-fir perhaps involving frequent
long-distant transport.
The inference of high gene flow is supported by the results obtained from the pattern of
genetic relationship between populations, wherein relatedness showed little structure
(Fig. 5), and little correlation with physical distance (Fig. 6). Workers have not previously
used PCA to analyse the genetic relationship matrix, but rather have used PCA to analyse
allelic frequencies (Guries, 1984) or multilocus genotypes (Conkle and Westfall, 1984:
Knowles, 1985; O'Reilly etal., 1985; Yeh etal., 1985,1986; Markle etal., 1988) to investigate
the relationship between grouping of populations to geography. Despite different
definitions of data, the results are similar. For example, Markle et al. (1988) also used
principal components analysis for the analysis of allozyme data for 22 Douglas-fir
populations from southwest Oregon, found the first principal component to explain only
3% of the variation, and 48 of the 78 principal components to account for 90% of the
variation.
The observed low Fls values (0.008 _ 0.015) indicate that little self-fertilization occurs in
this species. The mating system of Douglas-fir has been extensively studied and high
estimates of outcrossing rates have been reported for experimental and natural
populations (EI-Kassaby et al., 1981; Shaw and Allard, 1982: Neale and Adams, 1985:
Ritland and E1-Kassaby, 1985; Yeh and Morgan, 1987).
In our study, we estimated several components of genetic variation for each of the 49
populations were estimated (Table 1). Three of these, N a, (average number of alleles per
locus), H (expected heterozygosity) and D (divergence of a population from a reference
Genetic variation in low elevation Douglas-fir 791
population), seem to each measure a different facet of genetic diversity in Douglas-fir. The
number of alleles per locus No is of importance for longer-term response to selection, as
rare as well as common alleles are equally important in the long-term and are weighted
equally by this measure. The expected heterozygosity H is of importance for shorter term
response to selection, since the response to selection is proportional to pi(1 -p~) which is
exactly what heterozygosity is proportional to. The genetic divergence D measures the
'uniqueness' of a population in terms of its genetic distance from a reference population
consisting of the ensemble of populations under study. Genetic divergence occurs because
of either genetic drift or selection (natural selection can change the frequency effectively
neutral markers through their linkage to selected loci; this process is termed genetic
'hitchhiking').
The fact that these three measures (No, H and D) showed little intercorrelation among
populations (Fig. 4) indicates that no one single population, or even a group of
populations, can be identified as genetically 'most valuable'. This conclusion is further
supported by the distribution of alleles among populations (Fig. 3). Fifteen of the 61
electrophoretic alleles occurred in 11 or fewer populations. This distribution indicates that
a subsample of populations would likely miss many of the rare alleles. This component of
variation is imporant for the longer-term adaptability of Douglas-fir to such agents as
climatic changes, since the occasional rare allele will eventually be the allele favoured by
selection.
Isozymes are well known to represent only a portion of the genome. Most isozymes are
enzymes involved with primary metabolism. Many other types of genes, including
regulatory, most structural genes, ribosomal and tRNA loci, as well as genes in the
mitochondria and chloroplast, are not part of this group. Furthermore, even if they do
adequately represent variation for these other classes of genes (which is debatable), the
information content in a sample of only 20 neutral or nearly neutral loci is quite limited. It
is difficult to detect more than gross differences in genetic variation, and the different
facets of genetic variation further complicate this. At the minimum, we need more
informative types of molecular markers, such as RAPDs or microsatellites, to make better
inferences about patterns of genetic variation in Douglas-fir.
Douglas-fir has been studied extensively for several adaptive and quantitative attributes
and trends that are associated with geography were reported (Irgens-Moller, 1967;
Hermann and Lavender, 1968; Griffin and Ching, 1977; Campbell et al., 1978; Illingworth,
1978; Rehfeldt, 1983; Loopstra and Adams, 1989). Most of these studies were conducted
using a large number of seed sources representing populations covering large areas and
tested in either common gardens or provenance trials. While these studies are useful from
the selection and breeding point of view, their utility to gene conservation should be
evaluated in the light of genotype x environmental interaction or the environment in which
these trials are established.

Acknowledgements
The authors thank S. Barnes, C. Cook, L. EI-Kassaby, D. MacLeod, and B. Newbery for
technical assistance, D. Ashbee, C. Heaman, J. Woods, and C. Ying from the British
Columbia Ministry of Forests' Research Branch for providing access to research materials.
This work was supported in part by Pacific Forest Products Limited, a grant from the
792 EI-Kassaby and Ritland
Science Council of British Columbia to Y.A.E., and a Natural Sciences and Engineering
Research Council of Canada grant to K.R.

References
Adams, T.W., Neale, D.B., Doerksen, A.H. and Smith, D.B. (1990) Inheritance and linkage of
isozyme variants from seed and vegetative bud tissues in coastal Douglas-fir [Pseudotsuga
menziesii var. menziesii (Mirb.) Franco]. Silvae Genet. 39, 153-67.
Campbell, R.K. and Sorensen, F.C. (1978) Effect of test environment on expression of clines and on
delimitation of seed zones in Douglas-fir, Theor. AppL Genet. 51, 233-46.
Cheliak, W.M. and Pitel, J.A. (1984) Techniques for starch gel electrophoresis of enzymes from
forest tree species. Patawawa National Forestry Institute. Info. Rep. PI-X-42, 1-49.
Clayton, J.W. and Tretiak, D.N. (1972) Amine-citrate buffers for pH control in starch gel
electrophoresis. J. Fish. Res. Bord Can. 29, 1169-72.
Conkle, M.T. and Westfall, R.D. (1984) Evaluating breeding zones for ponderosa pine in California.
In Progeny Testing: Proc. of Servicewide Genetics Workshop, (D. Miller, ed) Charleston, S.C.
(1983) pp. 89-98. Washington, DC: USDA For. Serv.
Conkle, M.T., Hodgskiss, P.D., Nunnally, L.B. and Hunter, S.C. (1982) Starch gel electrophoresis of
conifer seeds: a laboratory manual. USDA, For. Serv. Gen. Tech. Rep. PSW.64, 1-18.
EI-Kassaby, Y.A. (1981) Genetic interpretation of malate dehydrogenase isozymes in some conifer
species. J. Hered. 72, 451-2.
EI-Kassaby, Y.A. (1990) Genetic variation within and among conifer populations: review and
evaluation of methods. In Biochemical markers in population genetics of forest trees
(H.H. Hattemer, S. Fineschi, F. Cannata and M.E. Malvolti, eds) pp. 59-74. The Hague: SPB
Academic Publishing bv.
El-Kassaby, Y.A. and Ritland, K. (1996) Impact of selection and breeding on the genetic diversity in
Douglas-fir. Biodiv. Conserv., 6, 795-813.
EI-Kassaby, Y.A. and Sziklai, O. (1982) Genetic variation of allozyme and quantitative traits in a
selected Douglas-fir [Pseudotsuga menziesii var. menziesii (Mirb.) Franco] population. For.
Ecol. Manage. 4, 115-26.
E1-Kassaby, Y.A., Yeh, F.C. and Sziklai, O. (1981) Estimation of the outcrossing rate of Douglas-fir
[Pseudotsuga menziesii (Mirb.) Franco] using allozyme polymorphism. Silvae Genet. 30, 182-4.
E1-Kassaby, Y.A., Sziklai, O. and Yeh, F.C. (1982a) Linkage relationships among 19 polymorphic
allozyme loci in coastal Douglas-fir [Pseudotsuga menziesii var. menziesii (Mirb.) Franco]. Can,
J. Genet. Cytol. 24, 101-8.
EI-Kassaby, Y.A., Yeh, F.C. and Sziklai, O. (1982b) Inheritance of allozyme variants in coastal
Douglas-fir [Pseudotsuga menziesii var. menziesii (Mirb.) Franco]. Can J. Genet. Cytol. 24,
325-35.
Ellstrand, N.C. (1992) Gene flow among seed plant populations. New For. 6, 241-56.
Fildes, R.A. and Harris, H. (1966) Genetically determined variation of adenylate kinase in man.
Nature 209, 261-3.
Fowells, H.A. (1965) Silvics of Forest Trees of the United States. U.S.D.A., For. Serv. Agriculture
Handbook No. 271 Washington, DC: USDA For. Serv. pp. 1-762.
Griffin, A.R. and Ching, K.K. (1977) Geographic variation in Douglas-fir from coastal ranges of
California. I. Seed, seedling growth and hardiness characteristics. Silvae Genet. 26, 149-56.
Guries, R.P. (1984) Genetic variation and population differentiation in forest trees. In Proc. 18th
North Am. P))r. Biol. Workshop (R.M. Lanner, ed) pp. 119-31. Utah: Logan.
Hamrick, J.L. and Godt, M.J.W. (1989) Allozyme diversity in plant species. In D(fferentiation
patterns in higher plants (K. Urbanska, ed.) pp. 53-67. New York: Academic Press.
Hamrick, J.L., Linhart, Y.B. and Mitton, J.B. (1979) Relationships between life history
characteristics and electrophoretically detectable genetic variation in plants. Ann. Rev. Ecol.
Svst. 10, 173-200.
Genetic variation in low elevation Douglas-fir 793
Hamrick, J.L., Godt, M.J.W. and Sherman-Broyles, S.L. (1992) Factors influencing levels of genetic
diversity in woody plant species. New For. 6, 95-124,
Heaman, J.C. (1985) A breeding program in the coastal Douglas-fir (Pseudotsuga menziesii (Mirb.)
Franco). In Proc. 18th Can. Tree Improv. Assoc. Part L (C.W. Yeatman and T.J.B. Boyle, eds)
pp. 186-8.
Hedrick, P.W. (1975) Genetic similarity and distance: Comments and comparison. Evolution 29,
362--6.
Hermann, R.K and Lavender, D.P. (1968) Early growth of Douglas-fir from various altitudes and
aspects in southern Oregon. Silvae Genet. 17, 143-51.
Hunter, R.L. and Markert, C.L. (1957) Histochemical demonstration of enzymes separated by zone
electrophoresis in starch gels. Science 125, 1294-5.
Illingworth, K. (1978) Douglas-fir provenance trials in coastal British Columbia- results to six years
after planting. In: Proc. IUFRO joint meeting of working parties $2-02-05, 06, 12, and 14 Vol. 1.
(K. Ching and O. Sziklai, eds) pp. 411-26. Vancouver: Vol. 1:411--426. Victoria, B.C., Canada:
B.C. Ministry of Forests, Information Branch.
Irgens-Moller, H. (1967) Patterns of height-growth initiation and cessation in Douglas-fir. Silvae
Genet. 16, 56--8.
Knowles, P. (1985) Comparison of isozyme variation among natural stands and plantations: jack pine
and black spruce. Can. J. For. Res. 15, 902--8.
Ledig, F.T. (1986) Heterozygosity, heterosis, and fitness in outbreeding plants. In Conservation
biology (M.E. Soulr, ed.) pp. 77-104. Sunderland, MA: Sinauer.
Lewontin, R.C. and Hubby, J.L. (1966) A molecular approach to the study of genetic heterozygosity
in natural populations. II. Amount of variation and degree of heterozygosity in natural
populations of Drosophila pseudoobscura. Genetics 54, 595-609.
Li, P. and Adams, W.T. (1989) Range-wide patterns of allozyme variation in Douglas-fir
(Pseudotsuga menziesii). Can. J. For. Res. 19, 149-61.
Loopstra, C.A. and Adams, W.T. (1989) Patterns of variation in first-year seedling traits within and
among Douglas-fir breeding zones in southwest Oregon. Silvae Genet. 38, 235-43.
Markel, S.A. and Adams, W.T. (1987) Patterns of allozyme variation within and among Douglas-fir
breeding zones in southwest Oregon. Can. J. For. Res. 17, 402-7.
Markel, S.A., Adams, W.T. and Campblee, R.K. (1988) Multivariate analysis of allozyme variation
patterns in coastal Douglas-fir from southwest Oregon. Can. J. For. Res. 18, 181-7.
Mitton, J.B. (1983) Conifers. In Isozymes in plant genetics and breeding, Part B (S.D. Tanskley and
T.J. Orton, eds) pp. 443-72. Amsterdam: Elsevier Science Publishers.
Moran, G.F. and Adams, W.T. (1989) Microgeographical patterns of allozyme differentiation in
Douglas-fir from southwest Oregon. For. Sci. 35, 3-15.
Neale, D.B. and Adams, W.T. (1984) The mating system in natural and shelterwood stands of
Douglas-fir. Theor. Appl. Genet. 71, 201-7.
Neale, D.B., Weber, J.C. and Adams, T.W. (1985) Inheritance of needle tissue isozymes in
Douglas-fir. Can. J. Genet. Cytol. 26, 549-68.
Nei, M. (1972) Genetic distance between populations. Am. Nat. 106, 283-92.
Nei, M. (1973) Analysis of gene diversity in subdivided populations. Proc. Natl. Acad. Sci. USA 70,
3321-3.
Nei, M. and Roychoudhury, A.K. (1974) Sampling variances of heterozygosity and genetic distance.
Genetics. 76, 379-90.
Nevo, E. (1978) Genetic variation in natural populations: patterns and theory. Theor. Pop. Biol. 13,
121-77.
O'Malley, D.M., Wheeler, N.C. and Guries, R.P. (1980)A manual for starch gel electrophoresis. Staff
Paper Series. Madison: University of Wisconsin.
O'Reilly, G.J., Parker, W.H. and Cheliak, W.M. (1985) Isozyme differentiation of upland and
lowland Picea mariana stands in northern Ohio. Silvae Genet. 17, 233-50.
794 EI-Kassaby and Ritland
Rehfeldt, G.E. (1983) Genetic variability within Douglas-fir populations: implications for tree
improvement. Silvae Genet. 32, 9-14.
Reynolds, J., Weir, B.S. and Cockerham, C.C. (1983) Estimation of the coancestry coefficient: basis
for a short-term genetic distance. Genetics 105, 767-79.
Ridgeway, G.J., Sherburne, S.W. and Lewis, R.D. (1970) Polymorphism in the esterases of Atlantic
herring. Trans. Am. Fish. Soc. 99, 147-51.
Ritland, K. (1989) Genetic differentiation, diversity, and inbreeding in the mountain monkeyflower
(Mimulus caespitosus) of the Washington Cascades. Can. J. Bot. 67, 2017-24.
Ritland, K. and EI-Kassaby, Y.A. (1985) The nature of inbreeding in a seed orchard of Douglas-fir as
shown by an efficient multilocus model. Theor. Appl. Genet. 71, 375-84.
Shaw, D.V. and Allard, R.W. (1982) Estimation of outcrossing rates in Douglas-fir using isozyme
markers. Theor. Appl. Genet. 62, 113-20.
Siciliano, M.J. and Shaw, C.R. (1976) Separation and visualization of enzymes on gels. In
Chromatographic and electrophoretic techniques, Vol. H (I. Smith, ed.) pp. 185-209. London:
Heinemann Medical Books.
Sneath, P.H.A. and Sokal, R.R. (1973) Numerical taxonomy. San Francisco: W.H. Freeman.
Wright, S. (1965) The interpretation of population structure by F-statistics with special regard to
systems of mating. Evolution 19, 395-420.
Yeh, F.C. and O'Malley, D.M. (1980) Enzyme variation in natural populations of Douglas-fir,
Pseudotsuga menziesii (Mirb,) Franco, from British Columbia. I. genetic variation patterns in
coastal populations. Silvae Genet. 29, 83-92.
Yeh, F.C. and Morgan, K. (1987) Mating system and multilocus associations in a natural population
of Pseudotsuga menziesii (Mirb.) Franco. Theor. Appl. Genet. 73, 799--808.
Yeh, F.C., Cheliak, W.M., Illingworth, B.P.K., Trust, D.C. and Pryhitka, B.A. (1985) Population
differentiation in lodgepole pine, Pinus contorta spp. latifolia: a discriminant analysis of
allozyme variation. Can. J. Genet. CytoL 27, 210-18.
Yeh, F.C., Khalil, M.A.K., E1-Kassaby, Y.A. and Trust, D.C. (1986) Allozyme variation in Picea
mariana from Newfoundland: genetic diversity, population structure, and analysis of
differentiation. Can. J. For. Res. 16, 713-20.