Full text of DOI:10.1093/forestry/73.5.419

Variation am ong seed sources of
silver birch in Scotland
R. WORRELL1, E.P. CUNDALL2, D.C. MALCOLM1 AND R.A. ENNOS1
1
School of Forestry, Institute of Ecology and Resource Management, University of Edinburgh, Darwin Building,
Mayfield Road, Edinburgh EH9 3JU, Scotland
2
Forest Research, Forestry Commission Northern Research Station, Roslin, Midlothian EH25 9SY, Scotland
Summary
The results of five provenance trials of silver birch (Betula pendula Roth.) in Scotland are described.
These comprise: (1) two trials laid out in the 1970s and 1980s comparing improved Finnish stock
with unimproved Scottish material, and (2) three recently established provenance trails of Scottish
seed sources for which 1–2 year results are available. Use of planting stock of Scandinavian origin is
not advisable in Scotland, or probably elsewhere in the UK, as it has poor survival and grows slowly
(i.e. volume growth at years 16–22 is only 7–26 per cent of the Scottish controls). It is particularly
prone to climatic damage due to late spring frosts. Recently established trials of seed sources from
Scotland and Northern England showed significant variation among Scottish populations in growth
and dates of flushing and senescence. Variation in 1-year height growth between the fastest and
slowest growing provenances was about 30 per cent. Some geographic patterns of variation in early
height growth and flushing were apparent among different seed sources at year 1, but equally there
was considerable variation among sources which could not be simply attributed to location. A
preliminary study of isozyme variation suggests that: populations are unlikely to have passed
through severe population bottlenecks; that there is no consistent evidence for inbreeding within the
populations; extensive gene exchange occurs between populations by pollen and/or seed flow; and
there is no evidence for more than one postglacial origin for the nine populations tested. There is a
need for more information on provenance performance among indigenous populations of birch in
order to give useful guidance to nurseries, foresters and policy makers.
recent decades there has been widespread plant-
ing of continental origin broadleaved stock,
including birch, and only in the last few years
have indigenous sources become more widely
available. This raises the question of the relative
merits of different seed sources, both indigenous
and foreign, for the production of planting
stock.
Introduction
Management of broadleaved trees has become
increasingly important during recent years and
broadleaves now constitute about 60 per cent of
all forest planting in the UK. Silver birch (Betula
pendula Roth.) is a widely planted species, par-
ticularly in upland areas, and is recognized as a
species with timber production potential
(Lorrain-Smith and Worrell, 1991; Cameron et
al., 1995; Cameron, 1996; Worrell, 1999). In
Since about 1970, a small number of trials have
been established in Scotland to test the perform-
ance of different origins of planting stock of silver
© Institute of Chartered Foresters, 2000
Forestry, Vol. 73, No. 5, 2000

420
FO R EST RY
birch. There have been three types of trial, late frosts and winter dieback of the crowns than
namely:
indigenous seed sources. The few progeny trials
of B. pendula and B. pubescens Ehrh. have
demonstrated that there is significant variation in
height, diameter and form among the progeny of
selected individual trees at years 2–3 (Blackburn
and Brown, 1988).
1 Scandinavian plus trees. During the 1970s and
early 1980s experiments were laid out to assess
the suitability of improved Scandinavian plant-
ing stock using unselected Scottish material as
controls.
Studies on the genetic variation in silver birch
in other northern countries have shown consider-
able variation between origins within a country.
In Sweden this was closely related to latitude of
origin and seed transfer rules have been devel-
oped based on this information (e.g. Eriksson and
Johnsson, 1986; Rosvall et al., 1998; see also
Ennos et al., 1998). However, in studies in south-
east Norway, geographic patterns were less clear
and considerable transfer of planting stock
appeared to be possible, especially longitudinally,
before maladaptation occurred (Langhammer,
1982). Currently there is no information on any
geographic patterns of genetic variation in
growth characteristics in silver birch populations
2 British and other plus trees. During the late
1970s and 1980s a pilot tree breeding project
based at Aberdeen University led to the estab-
lishment of progeny trials investigating the
heritability of form and growth of selected
individual trees from a range of sources, mainly
from the north-east Scotland, but including
some selections from England and Europe
(Blackburn and Brown, 1988; Richardson,
1992).
3 Provenance trials of Scottish seed sources.
Beginning in 1995 a series of provenance trials
has been established to assess seed source vari-
ation among Scottish birch populations.
These trials do not represent a coherent in Britain. However there is some evidence of
research programme, but are the result of spo- clinal variation in morphological characteristics,
radic interest by a number of research insti- such as leaf size, of B. pubescens across Britain
tutions. Rook and Fletcher (1991) comment that (Pelham et al., 1988).
the research effort has been ‘fragmented, with
The two older birch provenance trials in Scot-
very diverse aims, has lacked depth of effort in land have now been assessed at ages 16 and 22
any one area and suffered from inadequate long years, respectively, and the first assessments of the
term funding’. The trials represent a sequence new series of provenance trials of British seed
where foresters wishing to plant silver birch sources have been carried out (at age 1 year). In
looked first to material already available from addition, a preliminary study on isozyme vari-
foreign breeding programmes and only later ation in Scottish silver birch populations, using
began to consider British indigenous populations. material from the new series of trials, has been
The recent series of provenance trials of Scottish carried out.
origin material, which might logically have been
a first step, belatedly recognizes the importance of
understanding variation among indigenous seed
sources.
The aims of this paper are therefore to:
• draw together information from all these trials
and provide an updated account of the per-
formance of different seed sources of silver
birch, both domestic and foreign;
• provide some very preliminary information on
patterns of adaptive genetic variation in British
birch populations;
• provide a summary of patterns of variation at
four isozyme loci in some Scottish birch popu-
lations. Such studies of selectively neutral vari-
ation provide insights into the postglacial
origins of these populations and the extent of
gene flow among populations.
Previous accounts of some of these trials have
shown that trees of Scandinavian origin, includ-
ing improved stock from Finland, grow more
slowly than Scottish birch, typically showing
3–10 year mean heights about 80 per cent of that
of Scottish origin stock. The survival of the Scan-
dinavian material is also about 80 per cent of that
of the Scottish controls, though form is often
better than that of the Scottish trees (Worrell,
1992; Samuel, 1996). Lines and Brown (1982)
noted that foreign origin trees planted in north-
east Scotland were more susceptible to early and

VAR IAT IO N AM O N G SEED SO UR C ES O F SILVER BIR C H
421
Methods
Planting stock type and origin
Trial sites
Speymouth 33 and Teindland 113
At Speymouth 33 and Teindland 113 improved
Finnish material from breeding programmes was
used, which comprised progeny from polycrosses
between individual mother plus trees fertilized by
mixed pollen from a number of plus trees. The
Scottish controls were unselected material taken
from individual woods and no information is
available on how these were chosen or the exact
locations of the parent trees. Whilst it is likely
that they are from autochthonous populations,
this cannot be established with certainty. Details
of the Finnish crosses and the Scottish controls
are given in Appendix 1.
Details of the types of trial and their locations are
given in Table 1. The trials are mainly located in
north-eastern Scotland which is the main area in
the UK where silver birch is a major forest tree
species.
The trials of Scandinavian material at Spey-
mouth and Teindland (Morayshire) are only
about 7 km apart, but on strongly contrasting
sites. Speymouth 33 is on a hilltop fully exposed
to the sea to the north and east and on a poor
iron-pan soil with Calluna vulgaris heath as the
dominant vegetation type. Teindland 113 is at
110 m elevation on a relatively sheltered site with
a brown earth soil and the site was formerly used
for arable agriculture. Dunkeld 21 (Perthshire) is Dunkeld 21 and Moray 64
a relatively testing site for B. pendula, being fairly The new series of provenance trials at Dunkeld 21
high elevation (280 m) with a gleyed, stony soil and Moray 64 included 36 seed sources of B.
and in an elevation zone known to be prone to pendula from Scotland and northern England,
snow damage. Moray 64 (Morayshire) is adjacent together with one lot derived from a clonal seed
to Teindland 113 and is on a similar site. The Uni- orchard of plus trees (with origins around the
versity of Edinburgh trial is located in the uni- Great Glen area in Inverness-shire). In addition,
versity grounds and is therefore not strictly at Moray 64 one further lot of ‘improved birch’,
representative of forest sites in Scotland.
obtained commercially under the name ‘Elite
Table 1: Details of seed source trials of silver birch in Scotland
Location
Site type
———————————— ——————————————
Planting Region
year
Lat.
(°N)
Long. Elevation Rainfall Soil type
–1
Name of trial
Trial type
(°W)
(m)
(mm a )
Speymouth 33 Trial of Scandinavian
plus tree progeny with
1975 Moray
57 40
57 40
56 35
3 05
175
830 Iron-pan soil
unselected Scottish control
Teindland 113 Trial of Scandinavian
plus tree progeny with
1983 Moray
3 12
3 40
110
280
820 Brown earth
unselected Scottish control
Dunkeld 21
Provenance trial of 37 seed 1997 Tayside
lots from Scotland and
1200 Gleyed
upland
northern England
brown earth
(stony)
Moray 64
Provenance trial of 38 seed 1998 Moray
lots from Scotland and
northern England
57 40
3 12
3 10
110
70
820 Brown earth
University of
Edinburgh trial 8 Scottish and
1 Norwegian seed lot
Provenance trial of
1995 Midlothian 55 55
750 Loamy
brown earth

422
FO R EST RY
birch’ from Christie-Elite Nurseries, Forres, was
included, which is derived from selected material
from Scotland, England and Germany (three plots
in each replicate at Moray 64 consisted of this
planting stock). Two seed sources at Moray 64
(Glen Spean and Blaraidh, Glen Moriston)
showed low survival, which may have been due
to planting problems, and these have been
excluded from the analysis of that experiment.
In both experiments several sources of B.
pubescens were included, though these are not
considered in this paper. Several of the B. pendula
seed sources also included a small proportion of
B. pubescens plants as a result of difficulties on
the part of seed collectors in accurately telling the
two species apart in the field. Five seedlots
appeared to have considerable contamination
(20–40 per cent B. pubescens), but these lots have
been included in the 1-year analysis pending more
certain identification to species as they get older.
The locations of the seed sources are shown in
Figure 1 and a list of seed sources is provided in
Appendix 2. It is highly unlikely that all these
populations are autochthonous and at present
there is no means of distinguishing which are of
natural origin and which are of planted origin.
Those from the central and eastern Highland
region are most likely to be autochthonous,
whereas those from the western Highlands,
southern Scotland and northern England are most
likely to have been planted in the past (Worrell
and Malcolm, 1998).
Figure 1. The locations of the seed sources of B.
pendula included in the provenance trials at
Dunkeld 21 and Moray 64.
The University of Edinburgh provenance trial
This comprises nine seed sources from several
commercial nurseries (only two of these, Newtyle
Hill, Perthshire, and Loch Creran, Argyll, were the
same as those in the Dunkeld 21 and Moray 64
trials). Eight of the sources were Scottish and one
Norwegian, and the stock were of diverse sizes and
types. The trial was established mainly with a view
to developing assessment techniques, but never-
theless yielded some useful data on early patterns
of flushing and senescence. Details of the seed
sources are provided in Appendix 3 and the same
caveats about their autochthonous status apply.
1 for details). The variables assessed and the ages
of the assessments reported in this paper are given
in Table 2. Numerical scales used for assessing
flushing and senescence were as follows.
• Flushing stage. Visual assessments of the
progress of bud flushing were made using the
following scale:
1 = bud not swollen and still closed; 2 = bud
swollen but still closed; 3 = bud opening; 4 =
leaf starting to extend, but still bent back and
wrinkled; 5 = leaf open but not full size; 6 =
leaf fully open and full size.
• Senescence stage. Two variables were assessed
on the top five leaves of the leading shoot: (1)
number of leaves dropped (0–5); and (2) the
percentage of the area of the leaves which were
no longer green, i.e. yellow, brown or dropped
Experimental design and assessment
All the trials were complete randomized block
designs, with two to four replicates (see Appendix

VAR IAT IO N AM O N G SEED SO UR C ES O F SILVER BIR C H
423
(scale from 0–100 per cent, estimated to the Lamington; Cruach Ardura, Mull; Dall Mill,
nearest 10 per cent).
Black Wood of Rannoch; Glen Feshie, Kincraig;
Delnapot, near Elgin; Birkhill, Alford; Pannanich
Wood, Ballater – see Appendix 2 for details of
locations). Subsequent tests showed two of these
sources, Glen Feshie and Pannanich Wood, to
have a considerable proportion of B. pubescens
among the plants. Freshly expanded leaves were
sampled from approximately 48 individuals from
The assessments of flushing and senescence
were done during the spring and autumn (see
Table 3 for dates).
Isozyme studies
Isozyme variation was studied in nine of the each population (range 40–58). Leaves were
Scottish populations of B. pendula included in crushed over ice in extraction buffer, extracts
Dunkeld 21, chosen to represent the geographic were absorbed onto filter paper wicks, run on 11
range of the species in Scotland (i.e. Whitebridge, per cent starch gels for 5 h at 60 mA, and stained
Glengarry; Tomich, Glen Affric; Baitlaws, for the enzymes leucine amino peptidase (LAP),
Table 2: Variables assessed and ages of assessment
Name of trial
Variables assessed
Units
Age of assessment (year)
22
Speymouth 33 Mean height
Survival
m
%
Teindland 113 Mean height
Survival
m
%
16
Stem lesions
numbers > 4 cm long
Dunkeld 21
Moray 64
Mean height
Survival
Flushing stage
m
%
1
scale 1 (bud shut) to 5 (leaf open)
Mean height
Survival
m
%
1
Flushing stage
scale 1 (bud shut) to 5 (leaf open)
University of
Mean height
m
1 and 2
Edinburgh trial Survival
Flushing stage
Senescence stage
(% )
1 (bud shut) to 6 (leaf fully open)
% of leaves no longer green;
no. of leaves out of top 5 dropped
Table 3: Dates of assessments of flushing and senescence
University of Edinburgh trial
Flushing
1995
1996
1995
1996
8 and 28 April, 4 May
23 and 30 April, 8 May
3 and 20 October, 6 November
9 and 23 October, 12 November
Senescence
Dunkeld 21
Flushing
1998
1999
22 March (an assessment of a subset of 17 of the seed sources by Donelly (1998))
Moray 64
Flushing
20 March (an assessment of a subsample of six trees per plot for all seed sources by
Armstrong (1999))

424
FO R EST RY
phosphoglucomutase (PGM) and aspartate gene diversity (He), and mean Fis over loci was
aminotransferase (AAT). For details of buffer calculated for each population to determine the
systems see Brown (1999). Variation was scored level of inbreeding within populations. The extent
at four polymorphic isozyme loci: LAP1, LAP2, of isozyme differentiation among populations
PGM and AAT.
was measured by calculating the value of Gst.
Analysis
For Speymouth 33 and Teindland 113, differ-
ences between the Finnish and Scottish seed
sources in height growth, diameter growth and
Results
Performance of Scandinavian material
survival were analysed using ANOVA (Minitab). The average height growth of the Finnish prog-
Survival data were transformed to arcsine values. enies was 74 and 83 per cent of that of the Scot-
At Teindland 113 differences in the presence of tish controls at Speymouth and Teindland,
stem lesions were also analysed.
respectively, and the corresponding figures for
For Dunkeld 21 and Moray 64, differences diameter were 49 and 72 per cent (see Tables 4
among Scottish seed sources in height were and 5). The most striking contrast was in sur-
analysed using ANOVA (Forestry Commission vival; the Finnish material showing levels of only
statistical package) and relationships between 38 per cent (at Speymouth) and 61 per cent (at
measures of growth and flushing stage and lati- Teindland) of those of the Scottish controls. The
tude and longitude of origin were determined by volume production of the Finnish plots was only
regression analysis (Minitab).
about 7 per cent of that of the Scottish material
For the University of Edinburgh trial, differ- at Speymouth 33 and 26 per cent at Teindland
ences among seed sources in flushing and senes- 113. The Norwegian material (at Speymouth 33)
cence stage were analysed using ANOVA (Excel appears to perform equally poorly, with height
5 General Linear Model). Data for senescence and diameter being 64 and 49 per cent, respec-
which were expressed as percentages were trans- tively, of the Scottish controls and survival being
formed to arcsine values. Height values were not about half that of the Scottish material (plot
included in the data because the wide variation in volumes 8 per cent of the Scottish controls). This
initial plant sizes masked any apparent differ- is a clear case of maladaptation of Scandinavian
ences in seasonal growth.
material to site conditions in this part of Scotland,
even on the sheltered low elevation site at Teind-
land.
Isozyme study
At Teindland 113 the two Scottish controls also
Data analysis was carried out using POPGENE differed considerably in performance, with one
(Version 1.21). Diversity in each population was (Tummel) growing only marginally better than
measured as percentage polymorphic loci (P) and the average of the Finnish material in terms of
Table 4: Performance of Finnish, Norwegian and Scottish B. pendula at age 22 in experiment Speymouth 33
Mean height
Diameter at 1.3 m
Survival
———————————– ————————————– ————————————
m (% of Scottish control) cm (% of Scottish control) % (% of Scottish control)
Seed source
Mean of Finnish progenies 6.2
(74)
(64)
5.2
5.1
10.4
(49)
(49)
36
50
94
(38)
(53)
Tangen, Stange, Norway
Banchory, Scotland
5.4
8.4
The Finnish stock was progeny of plus trees in a breeding programme; the Norwegian and Scottish material
was from unselected sources.

VAR IAT IO N AM O N G SEED SO UR C ES O F SILVER BIR C H
425
Table 5: Performance of Finnish and Scottish B. pendula at age 16 in experiment Teindland 113
Mean height
Diameter at 1.3 m
Survival
———————— ————————– ————————
m
(% of
Scottish
controls)
cm
(% of
Scottish
controls)
%
(% of
Scottish
control)
Mean no. of stem
lesions in lower
2 m of stem
Seed source
Mean of Finnish progenies
Tummel, Scotland
7.8
7.9
(83)
6.8
8.0
(72)
59
94
(61)
1.49
0.2
Glen Prosen, Angus, Scotland 10.7
10.9
100
0.1
The Finnish stock was progeny of plus trees in a breeding programme and Scottish material was from
unselected sources. The values for the Finnish sources of percentage growth, survival and presence of stem
lesions use the average of the two Scottish controls for comparison
height, though its diameter and survival were Castle Howard (N. England) and Sand Hutton
better than the Scandinavian material (see Table (N. England). Those showing poor rankings
5). There was also some variation among the (≥ 25) were: Leanachan (Inverness-shire), Birkhill
Finnish progenies with one (seed source 8) per- (Aberdeenshire), Dall Mill (Perthshire), Achnatra
forming relatively well. This had survival, height (Argyll) and Cruach Ardura (Argyll).
and diameter of 87, 97 and 85 per cent of the
mean of the Scottish material, respectively (Arm- groups, and taken across the two experiments,
strong, 1999). southern groups (i.e. N. England, Borders and
The seed sources were allocated to geographic
At Teindland 113 a high proportion of the Angus) appeared to show faster growth than
Finnish material had stem lesions (average 1.5 more northerly and easterly ones (i.e. Great Glen
lesions over 4 cm long per tree in the lower 2 m and Aberdeenshire) (see Table 6). The ranking of
of stem), with necrotic bark around them. The these groups was generally similar for both trials.
Scottish material was virtually unaffected This trend was confirmed by regression analysis
(average 0.15 lesions per tree). The crowns of the at Moray 64, where a relationship was apparent
Finnish material also showed considerable thin- between height at year 1 and location of origin,
ning, with a high proportion of dead twigs, par- with southern and western origins being taller
ticularly on the slower growing origins. These than northern and eastern ones:
symptoms are interpreted as the result of late
Height (cm) = 146.0 – 1.81 latitude (°N) + 1.21
spring frost damage (D. Redfern, personal com-
longitude (°W)
munication).
(F2,31 = 8.80, P ≥ 0.001 (Armstrong, 1999)).
However, it is clear from Table 6 that there is
Variation among British seed sources
considerable variation among sources within
Growth and survival
regions which cannot be attributed simply to
There was significant variation in height growth location. Whilst there is some correspondence
at year 1 among seed sources from northern between the ranking of the sources between the
Britain at both Dunkeld 21 and Moray 64 (at two experiments, a considerable proportion have
Dunkeld F2,36 = 2.29, P < 0.01; at Moray F2,37
=
quite different rankings, suggesting that geno-
2.95, P < 0.001; see Table 6). The fastest growing type ꢀ environment interactions were present.
seed sources at Dunkeld 21 and Moray 64 grew More detailed analysis of geographic variation at
at 113 and 116 per cent of the experiment mean this early age was not considered worthwhile and
and the slowest growing at 89 and 80 per cent, is planned when the trials reach 3 years of age.
respectively. The sources achieving high rankings
The material from crosses between plus trees in
(< 10) for growth rate in both experiments were: the Great Glen area showed only slightly higher
Killin (Perthshire), Montreathmont (Angus), growth rates than the experiment mean at
Dunalastair (Perthshire), Locharbriggs (Borders), Dunkeld 21 (103 per cent of experiment mean)

426
FO R EST RY
Table 6: Heights of seed sources and ranking in eight geographic groups at trials Dunkeld 21 and Moray 64
Dunkeld
Moray 64
——————————————
—————————————–
Height
(cm)
Rank
(group rank)
Height
(cm)
Rank
(group rank)
Seed source
Great Glen
Leanachan
Glen Spean
White Bridge
Blaraidh, Glen Moriston
Tomich
Strathgarve
Kyloag
Mean
34.7
30.3
36.1
37.1
38.3
35.3
35.3
35.3
28
–
22
–
12
24=
24=
(8)
42.1
–
46.5
–
45.2
47.2
48.1
45.8
31
–
24
–
26=
20
15
(7)
Strathspey
Straanruie
Delnapot
Spindlemuir
Mean
36.4
38.6
39.4
38.1
21
11
5
50.8
48.6
42.8
47.4
9=
14
29
(3)
(5)
Aberdeenshire
Muir of Dinnet
Birkhill, Alford
Invermossat
Mean
37.9
34.5
34.5
35.6
16
39.4
42.7
47.4
43.2
34
30
19
(8)
29=
29=
(7)
Angus
Montreathmont
Silvie, Alyth
Ingilsmaldie
Mean
40.4
37.3
38.0
38.6
3
17
15
(1)
52.1
49.0
46.7
49.3
6
13
22
(3)
Perthshire
Dunalastair
38.8
35.0
32.9
38.2
41.9
37.4
8=
26
34
13=
1
(4)
51.2
40.7
46.8
48.0
53.0
47.9
8
33
21
16=
5
Dall Mill, Rannoch
Newtyle Hill, Dunkeld
Craiganour, Rannoch
Killin
Mean
(4)
West Coast
Langbank
Achnatra
Creagan, Loch Crearan
Cruach Ardura, Mull
Mean
35.9
33.2
41.2
34.8
36.3
23
33
2
27
(6)
49.9
45.4
45.2
43.8
46.1
11
25
26=
28
(6)
Borders
Glentress
39.0
38.2
33.9
38.8
37.2
34.0
39.8
36.7
37.2
7
13=
32
10
18
31
4
41.2
54.9
49.7
48.0
46.6
47.8
51.5
56.6
49.5
32
4
Floors, Kelso
Mellerstain, Kelso
Petersmuir
Baitlaws
Wauchope
Locharbriggs
Elibank
12
16=
23
18
7
20
(5)
2
(2)
Mean

VAR IAT IO N AM O N G SEED SO UR C ES O F SILVER BIR C H
427
Table 6: Continued
Dunkeld
Moray 64
————————————————– —————————————–
Height
(cm)
Rank
(group rank)
Height
(cm)
Rank
(group rank)
Seed source
N. England
Finsthwaite
Castle Howard
Sand Hutton
Mean
36.9
38.8
39.3
38.3
19
8=
6
56.9
50.8
56.2
54.6
1
9=
3
(2)
(1)
Mean of all seed sources
FC selected crosses from Great Glen
‘Elite birch’ (mixed origin)
37.1
38.1
–
49.0
45.8
55.4*
* Mean of three entries.
and showed lower growth rates at Moray 64 (93 significant variation among the seed sources in
per cent of experiment mean). This relatively flushing and senescence was recorded on most,
poor growth is presumably a reflection of the but not all, assessment dates during both years 1
slower growth of sources from the Great Glen and 2. The 2-year data were regarded as more
area (the Great Glen group was ranked 7 and 8 reliable than the 1-year data because in 1995 (1)
for height growth in the trials). The ‘Elite birch’ the plants had only been planted some 4 months
grew faster than the experiment mean at Moray prior to flushing and before that had experienced
64 (113 per cent of experiment mean) and only different conditions in the different source nurs-
three of the sources at this trial exceeded it in eries, which may have affected flushing; and (2)
height growth.
summer 1995 was exceptionally hot and dry and
There was no significant variation in survival leaf-fall in 1995 was clearly affected by drought.
at Dunkeld 21 or Moray 64 and survival rates The analyses of year 2 data demonstrated the
were generally very high (around 95 per cent) following.
despite severe spring frost episodes at Dunkeld in
1997 and 1998.
• Significant differences for flushing were found
among Scottish seed sources on 23 April and 8
May (F2,7 = 3.4, P = 0.024; F2,7 = 3.3, P =
0.027) but not on 30 April (F2,7 = 1.54, P =
0.23). Clear differences in senescence were
found on both 23 October and 12 November
(F2,7 = 67.3, P ≥ 0.000; F2,7 = 24.3, P ≥ 0.000,
respectively). Differences were greater for time
of senescence than for flushing; the progress of
senescence between the earliest and latest seed
sources averaged about 9 days, whereas for
flushing this was 4–5 days.
The mean rates of annual height growth for the
Scottish material over the lifespan of the two
older trials at Speymouth 33 and Teindland 113
were 38 cm and 58 cm respectively; with the best
origin (Banchory) growing at 67 cm a–1 at Teind-
land. These values probably represent the
expected range of growth rates for unselected
stock in this part of northern Scotland.
Phenology
There was significant variation among seed
sources in flushing stage at the beginning of the • Early flushing Scottish seed sources tended also
second growing season at Dunkeld 21 and Moray
64 (Donelly, 1998; Armstrong, 1999). At
Dunkeld, flushing stage was related to location of
origin, with southern origins flushing earlier than
northern ones (see Figure 2).
to senesce earliest in autumn and vice versa.
The average growing season length was 181
days, with only a 7 per cent variation between
the seed sources with the longest (187 days)
and shortest (175 days) growing seasons
(defined as the period between trees achieving
At the University of Edinburgh garden trial,

428
FO R EST RY
Figure 2. The relationship between flushing score and latitude of origin for 17 provenances at Dunkeld 21
scored in March 1998 (2 = late flushing, bud swollen but still closed; 4 = advanced flushing, leaf extended
but not fully open).
a flushing score of 5 in spring and having lost outcrossing breeding systems (Brown, 1999).
≥ 50 per cent of their green foliage in autumn). Measures of diversity should be treated with
Flushing dates in 1996 varied for the different caution since only four isozyme loci were used in
origins from 7–12 April and senescence dates their calculation. A significant excess of hetero-
ranged from 4 to 12 November.
zygotes in comparison with random mating
• The Norwegian control was clearly different expectations was found in three populations, and
from all Scottish seed sources in growth a significant deficit of heterozygotes in two popu-
rhythm, flushing some 15 days earlier and lations. However the mean value of Fis for all
senescing some 16 days earlier than the average populations was very close to zero, suggesting no
for the Scottish seed sources.
consistent, significant level of inbreeding within
• There were no apparent geographic patterns in these B. pendula populations. The mean estimate
flushing. The seed sources showing earliest of Gst over loci was 0.062, indicating that only 6
senescence appear to be those from the north per cent of isozyme variation in Scottish B.
west coast (Cluanie and Loch Creran) and pendula was found among populations (see
from higher elevations (Blackmount and Glen- Tables 7 and 8).
livet) compared with eastern and southern seed
sources; for example the senescence scores (i.e.
percentage of leaf area not green) for 12
November 1996 for these geographic groups
were 64.5, 57.4 and 53.2, respectively.
However, the trial has too few seed sources to
be able to test adequately for geographic pat-
terns.
Discussion
These early results for the performance of differ-
ent seed sources raise three questions which are
critical for forest practice in Britain, namely:
• Are foreign sources acceptable as seed sources
in the UK?
• No significant differences were recorded in the
numbers of leaves lost on the leading shoot • Is it possible to identify particularly promising
during senescence and this is therefore not a
good measure of the progress of senescence.
seed sources and how far may they be trans-
ferred before there is evidence of maladapt-
ation?
Isozyme studies
• Do studies of genetic markers reveal aspects of
the postglacial history of the species which
need to be taken into account in choice of seed
source or other aspects of genetic conser-
vation?
The four isozyme loci scored were polymorphic
in most of the populations and the overall level of
gene diversity within populations (He = 0.224) is
similar to that found in other tree species with

VAR IAT IO N AM O N G SEED SO UR C ES O F SILVER BIR C H
429
Table 7: Summary of isozyme variation (% P, % polymorphic loci; He, gene diversity) and inbreeding (Fis) in
nine Scottish populations of B. pendula
Population
n
% P
He
Fis
Glengarry
Affric
Lamington
Mull
Black Wood of Rannoch
Glen Feshie
Delnapot
Alford
47
58
40
54
45
46
49
56
40
48.3
100
75
75
100
50
100
100
100
75
0.241
0.194
0.234
0.165
0.219
0.235
0.299
0.227
0.201
0.224
–0.154
–0.245
–0.006
+0.058
+0.344
–0.102
–0.108
–0.222
–0.005
–0.049
Ballater
Mean
86.1
n = sample size.
Table 8: Proportion of gene diversity detected among
populations (Gst) for Scottish B. pendula
has been reported for Betula in other north tem-
perate regions following exceptionally warm
periods in late winter (Braathe, 1957). Climatic
damage on British B. pendula was also reported by
Blackburn and Brown (1988), who attributed this
to spring frosts and excessive cold and desiccation
in a period of cold weather in late winter. There is
some evidence that the poor performance of Scan-
dinavian material in Scotland is mirrored when
Scottish material is planted in Scandinavia; a Scot-
tish seed source showed particularly low survival
in a trial at Ås in south-east Norway (Langham-
mer, 1982). Thus the contrast in environments
between Scotland and Scandinavia appears to be
simply too great for B. pendula to cope with.
The particularly poor performance of the
Finnish material on the exposed infertile site at
Speymouth, as opposed to Teindland, suggests
that maladaptation is worse on poor upland sites
than on lowland ones. The variability in per-
formance among the Finnish progenies at Teind-
land is interesting, as it suggests some variation in
adaptability among the Finnish mother trees.
However, it does not detract from the conclusion
that use of Finnish origin planting stock is
unacceptable. The results of these trials should
dissuade any further attempts to cut short the
improvement process for birch by importing
improved foreign planting stock.
Locus
Gst
LAP1
LAP2
PGI
AAT
Mean
0.044
0.059
0.042
0.101
0.062
Data are given for four individual isozyme loci, and
the mean value over loci
Scandinavian sources
This study provides good evidence that the use of
Scandinavian planting stock is not advisable in
Scotland or probably elsewhere in the UK. In the
two provenance trials described in this paper, the
plots of improved Scandinavian birch only
attained average volume production of between 7
and 26 per cent of that of unselected Scottish
sources and were susceptible to climatic damage.
Their performance had worsened since the assess-
ment at age 10 reported by Worrell (1992) and
Samuel (1996), suggesting that they had experi-
enced continuing problems of maladaptation. The
reason for the poor performance of the
Scandinavian stock appears to be spring frost
damage as a result of dehardening occurring too
early during mild periods in late winter (D.
Redfern, personal communication). Many of the
Scandinavian trees had developed stem lesions
suggesting death of the cambium, and thinning of
the crowns was evident. Similar climatic damage
Variation among Scottish sources
There is clearly considerable variation among
Scottish populations in growth and phenology.

430
FO R EST RY
This suggests that it is worthwhile to attempt to general, then this scale of difference in flushing
identify specific sources most suitable for plant- times leads to severe maladaptation. It remains to
ing, and that selection and breeding within the be seen whether the 4–5 day variation among
best populations could result in significant gains Scottish sources in the same trial is enough to lead
in volume production and quality. The scale of to significant variations in susceptibility to frost
the differences in 1-year height growth between damage. In this respect there was somewhat con-
the fastest and slowest growing seed sources tradictory evidence from the trials: at Dunkeld
(from about 85 to 115 per cent of the experi- particularly severe spring frost events in 1998 did
mental mean) is similar to that reported from not appear to lead to significant damage; whereas
regional scale trials of B. pendula in Norway at the University of Edinburgh trial there was
(Langhammer, 1982).
considerable shoot dieback in some seed sources.
There is some evidence of geographic patterns Ongoing work testing resistance to frost and des-
among the seed sources, especially at Moray 64, iccation in a selection of Scottish seed sources will
where southern and western sources initially cast more light on this (Bruce and McKay, 1999).
appear to be relatively fast growing. This con-
There are increasing signs that some commer-
firms the well-established pattern that north- cially available birch planting stock is susceptible
wards transfer of planting stock tends to lead to to climatic damage, especially the so-called ‘Elite
faster growth. However, there is considerable birch’ which was made available commercially
variation among sources within regions and the when the Aberdeen University breeding work
simplistic view that local sources inevitably ceased (Richardson, 1992). In the more extreme
perform better than sources from other regions of cases this has led to death of trees in the sapling
Scotland is not substantiated by these preliminary stage. This may be a result of the original plus
growth data. Part of this variation may be trees containing a proportion of southern English
attributable to the fact that not all the popu- or European material and/or of a tendency of plus
lations sampled in the trial are autochthonous, trees to occur at low elevations and on sheltered
and the inclusion of any planted sources will sites. Such material would be most likely to
cloud the emergence of geographic patterns. Sec- experience problems when planted on poorer
ondly, and more importantly, the provenance upland sites. In fact the reputation of silver birch
trials are only 1 year old, and this is very early to as a hardy species is to some extent a misunder-
expect to observe differences, as the plants have standing, stemming from a lack of appreciation of
only been subjected to one season’s weather and the differences between silver birch, which is not
are therefore unlikely to have encountered the particularly hardy, and downy birch, which is.
sorts of extremes which lead to adaptive differ- This reinforces the need to be relatively con-
entiation. More detailed analysis of growth pat- servative when recommending setting upper
terns when the experiments are older is expected limits to transfer distances, at least until more
to cast useful light on the question of geographic information becomes available.
variation among seed sources.
The issue is complicated by the fact that early
growth rate alone is not a good indicator of
Isozyme study
overall performance, and slower growing seed The results of the isozyme study were largely as
sources which may be less prone to frost-related anticipated for a wide-ranging, outcrossing tree
damage, could well turn out to be more desirable species with extensive powers of both pollen and
for timber production. The variation in flushing seed dispersal. A high degree of polymorphism
times recorded at the University of Edinburgh was detected for the small number of isozyme loci
trial and at Dunkeld 21 is potentially important, scored, indicating that none of these populations
as it suggests the possibility of increased risk of are likely to have passed through severe popu-
frost damage on the part of the earlier flushing lation bottlenecks. Genotype frequencies within
seed sources. The Norwegian seed source in the populations generally conform to expectations
University of Edinburgh trial flushed some 15 under random mating. There was no consistent
days earlier than Scottish material and if this is evidence for inbreeding within the populations.
representative of Scandinavian material in Differentiation was small, on the high end of the

VAR IAT IO N AM O N G SEED SO UR C ES O F SILVER BIR C H
431
range expected for outcrossing tree species, indi- Cameron, A.D., Dunham, R.A. and Petty, J.A. 1995
The effects of heavy thinning on stem quality and
timber properties of silver birch. Forestry 68,
275–286.
Donelly, A. 1998 Provenance variation in silver birch
(Betula pendula Roth.). Unpublished honours thesis,
University of Edinburgh.
Ennos, R.A., Worrell, R and Malcolm, D.C. 1998 The
genetic management of native species in Scotland.
Forestry 71, 1–23.
Eriksson, I. and Jonsson, A. 1986 A review of the genet-
ics of Betula. Scandin. J. For. Res. 1, 421–434.
Langhammer, A. 1982 [Growth studies on silver birch
(Betula verrucosa Ehrh.) in Norway]. Scientific
Report of the Agricultural University of Norway 14,
43 pp [in Norwegian].
Lines, R. and Brown, I.R. 1982 Broadleaves for the
uplands. In Broadleaves in Britain: Future Manage-
ment and Research. Proceedings of a Symposium,
University of Technology, Loughborough. D.C.
Malcolm, J. Evans and P.N. Edwards (eds). Institute
of Chartered Foresters, Edinburgh, pp. 141–149.
Lorrain-Smith, R. and Worrell, R. [eds] 1991 The Com-
mercial Potential of Birch in Scotland. Forestry
Industry Committee of Great Britain, London.
Pelham, J., Gardiner, A.S., Smith, R.A. and Last, F.T.
1988 Variation in Betula pubescens Ehrh. (Betu-
laceae) in Scotland: its nature and association with
environmental factors. Bot. J. Linn. Soc. 96,
217–234.
cating extensive gene exchange between popu-
lations by pollen and/or seed flow. There was no
indication of substantial differentiation for par-
ticular groups of populations, and therefore no
evidence for more than one postglacial origin for
the nine populations (Brown, 1999).
The results reported in this paper still only
provide a sketchy account of adaptive genetic
variation in birch, despite this species having
received some research attention over the last
25 years. There is clearly a need for more infor-
mation on provenance performance in birch in
order to give useful guidance to nurseries,
foresters and policy makers. Some of this will
become available from the series of provenance
experiments now being established, but this still
only includes seed sources from northern Britain,
and similar trials located in England, with English
and Welsh sources, would be useful.
Acknowledgements
The authors wish to thank the following people and
organizations who contributed to this paper: staff of
the Forestry Commission Forest Research including
Sam Samuel and members of the Technical Support
Unit; the following students at the University of Edin-
burgh: Linda Armstrong, Andrew Donelly and Nicola
Brown; and the owners and agents who gave access to
woods for seed collection.
Richardson, C.D. 1992 A review of the methods and
strategies for the genetic improvement of birch in
Scotland. Unpublished M.Sc. thesis, University of
Aberdeen.
Rook, D.A. and Fletcher, A.M. 1991 Research on the
genetics of birch in Britain. In The Commercial
Potential of Birch in Scotland. R. Lorrain-Smith and
R. Worrell (eds). Forestry Industry Committee of
Great Britain, London, pp. 43–54.
Rosvall, O, Andersson, B. and Ericsson, T. 1998
Species-specific guidelines for choosing forest
regeneration material for northern Sweden. The
Forestry Research Institute of Sweden Redgorelse 1,
66 pp.
References
Armstrong, L. 1999 Provenance variation in silver birch
(Betula pendula Roth.). Unpublished honours thesis,
University of Edinburgh.
Blackburn, P. and Brown, I.R. 1988 Some effects of
exposure and frost on selected birch progenies.
Forestry 61, 219–234.
Braathe, P. 1957 Is there a connection between the birch
dieback and the March thaw of 1936? For. Chron. Samuel, C.J.A. 1996 Growth characteristics of locally
33, 358–363.
grown native provenances compared with translo-
cated and foreign stock. In Native and Non-native in
British Forestry. Proceedings of a Meeting, Uni-
versity of Warwick 1995. P.R. Ratcliffe (ed.). Insti-
tute of Chartered Foresters, Edinburgh, pp.
100–111.
Brown, N. 1999 Isozyme analysis of silver birch
(B. pendula Roth.) in Scotland. Unpublished M.Res.
thesis, University of Edinburgh.
Bruce, J. and McKay, H. 1999 A study of the frost har-
diness status in silver birch seedlings of nine Scottish
origins. Unpublished Report to Scottish Forestry Worrell, R. 1992 A comparison between European and
Trust, Edinburgh.
British provenance of some British native trees:
growth survival and stem form. Forestry 65,
253–280.
Cameron, A.D. 1996 Managing birch woodlands for
the production of quality timber. Forestry 69,
357–374.
Worrell, R. 1999 Birch Woodland Management

432
FO R EST RY
Handbook. Highland Birchwoods, Timber Growers Received 6 January 2000
Association and Aberdeen University, 56 pp.
Worrell, R. and Malcolm, D.C. 1998 Anomalies in the
distribution of silver birch (Betula pendula Roth.) in
Scotland. Bot. J. Scotland 50, 1–10.

VAR IAT IO N AM O N G SEED SO UR C ES O F SILVER BIR C H
433

434
FO R EST RY
Appendix 2
The locations of the seed of B. pendula included in the provenance trials at Dunkeld 21 and Moray 64
Latitude
(° min)
Longitude
(° min)
Grid reference
Great Glen
Leanachan
NN 222771
NN 290808
NH 284007
NH 376163
NH 315283
NH 408605
NH 662910
56 50
56 53
57 03
57 12
57 18
57 35
57 52
4 56
4 48
4 50
4 41
4 47
4 40
4 15
Glen Spean
White Bridge
Blaraidh, Glen Moriston
Tomich
Strathgarve
Kyloag
Strathspey
Straanruie
Delnapot
NH 996154
NJ 169370
NJ 153648
57 13
57 24
57 39
3 40
3 22
3 25
Spindlemuir
Aberdeenshire
Muir of Dinnet
Birkhill, Alford
Invermossat
NJ 433999
NJ 595163
NJ 495184
57 04
57 14
57 15
2 55
2 40
2 50
Angus
Montreathmont
Silvie, Alyth
Ingilsmaldie
NO 569548
NO 273484
NO 665713
56 40
56 37
56 49
2 42
3 10
2 32
Perthshire
Dunalastair
NN 733614
NN 589568
NO 040423
NN 590580
NN 560310
56 43
56 40
56 33
56 41
56 26
4 03
4 18
3 34
4 18
4 19
Dall Mill, Rannoch
Newtyle Hill, Dunkeld
Craiganour, Rannoch
Killin
West Coast
Langbank
Achnatra
Creagan, Loch Crearan
Cruach Ardura, Mull
NS 490720
NN 110090
NM 970440
NM 680300
55 55
56 14
56 33
56 23
4 34
5 02
5 18
5 45
Borders
Glentress
NT 286406
NT 700280
NT 650380
NT 490660
NS 979308
NT 587075
NY 991810
NT 355365
55 38
55 32
55 37
55 53
55 33
55 21
55 06
55 36
3 08
2 28
2 33
2 49
3 37
2 39
3 34
3 01
Floors, Kelso
Mellerstain, Kelso
Petersmuir
Baitlaws
Wauchope
Locharbriggs
Elibank
N. England
Finsthwaite
Castle Howard
Sand Hutton
SD 363865
SE 705700
SE 689582
54 16
54 07
54 00
2 58
0 55
0 56

VAR IAT IO N AM O N G SEED SO UR C ES O F SILVER BIR C H
435
Appendix 3
Seed source locations at the University of Edinburgh experiment (the lots were bought in commercially and the
exact locations are not all known)
Seed source
Location details
Berwickshire, Borders
Berwick
Newtyle Hill
Banchory
Cawdor
Newtyle Hill, Dunkeld, Perthshire (see Appendix 2)
Banchory, Aberdeenshire
Cawdor Estate, Moray
Glenlivet
Blackmount
Loch Creran
Cluanie
Glenlivet, Moray
Blackmount Estate, Bridge of Orchy, Argyll
Loch Creran, Appin, Argyll (see Appendix 2)
Cluanie, Glen Moriston, Inverness-shire
Unknown location in south Norway
Norway