Genetic variation in growth, stem straightness, pilodyn and dynamic modulus of elasticity in secondgeneration progeny tests of Acacia mangium at three sites in Vietnam

New Forests (2015) 46:577–591
DOI 10.1007/s11056-015-9484-6

Genetic variation in growth, stem straightness, pilodyn
and dynamic modulus of elasticity in second-generation
progeny tests of Acacia mangium at three sites in Vietnam
Phi Hong Hai1 • La Anh Duong1 •
Nguyen Quoc Toan2 • Trieu Thi Thu Ha2

Received: 30 July 2014 / Accepted: 23 April 2015 / Published online: 28 April 2015
Ó Springer Science+Business Media Dordrecht 2015

Abstract 164 open-pollinated families of Acacia mangium from six different genetic
groups were tested in three second-generation progeny tests planted at Tuyen Quang and
Ba Vi in northern Vietnam and Bau Bang in the south. All trees were measured to estimate
individual heritabilities and genetic correlations for growth traits, stem straightness and
pilodyn in the three trials, and dynamic modulus of elasticity (MoEd) of standing trees was
only assessed in Tuyen Quang. There were significant differences between families for
growth traits, stem straightness, pilodyn penetration and predicted MoEd. Heritabilities of
growth traits, stem straightness, pilodyn and dynamic modulus of elasticity were low to
moderate (h2 = 0.11–0.30). The coefficient of additive genetic variation for DBH, pilodyn
and MoEd were moderate at age 3 or 4 years (CVa = 4.9–9.4 %). Genetic correlations
between stem straightness, pilodyn and growth traits were favourable but weak, while
those between growth traits and dynamic modulus of elasticity were weak and unfavourable. The substantial coefficients of additive genetic variation and significant heritabilities for all traits indicate that it should be possible to use a selection strategy that
combines improvements in growth, stem straightness, and wood quality for A. mangium in
Vietnam. The site–site genetic correlations between the two northern trials and Bau Bang
site were low for growth traits, indicating that G 9 E effects are of practical importance
for growth and different deployment populations will be required for different sites.
Keywords

Acacia mangium Á Genetic variation Á Growth Á Pilodyn Á Wood stiffness

& Phi Hong Hai

1

Department of Planning and Sciences, Vietnamese Academy of Forest Sciences,
Duc Thang, Bac Tu Liem, Ha Noi, Viet Nam

2

Institute of Forest Tree Improvement and Biotechnology, Vietnamese Academy of Forest Sciences,
Duc Thang, Bac Tu Liem, Ha Noi, Viet Nam

123


578

New Forests (2015) 46:577–591

Introduction
Acacia mangium Willd. was introduced into Vietnam in the 1980s. The species has
become important commercially (Nghia 2003; Turnbull et al. 1997), especially in
northern Vietnam, because it displays adaptability to a wide range of site conditions,
produces straight stems, and grows faster than alternative short-rotation plantation species such as A. auriculiformis and Eucalyptus urophylla. It produces pulp logs and small
sawlogs on rotations as short as 7 years. At present, A. mangium occupies about
500,000 ha of the total area of forest plantation of over 2 million ha. This makes A.
mangium the most important acacia species in Vietnam. However, the productivity of A.
mangium plantations in Vietnam is moderate compared to rest of South-East Asia
(Harwood and Nambiar 2014), averaging 15 m3 ha-1 yr-1 (Nghia 2003), although it can

be higher in the south (Kha 2003).
The natural distribution of A. mangium includes northern Queensland (QLD, Australia),
Western Province, Papua New Guinea (PNG) and Indonesia (West Papua, adjacent to PNG
and outlying populations in the Aru Islands and Ceram). Commonly used seed material in
Vietnam plantations includes origins from provenances in both QLD and PNG (Kha 2003).
Provenances from PNG consistently show better growth than those from QLD or local
commercially-sold seed (Harwood and Williams 1992; Kha 2003; Nghia 2003). Within
QLD, provenances from the Cairns region (16°–18°S) were slower-growing than those
from Claudie R.—Iron Range in far north QLD (FNQ) (11°–13°S).
A breeding program for A. mangium in Vietnam commenced in 1996. In the first
generation, 150 families from 10 provenances of A. mangium were tested in two
progeny tests in northern and southern Vietnam, which were selectively thinned to
convert them to seedling seed orchards. At age 3 years, significant site by family
interactions for growth between the northern and southern sites were reported (Dao
2012). 150 superior individual trees in the southern trial were selected using an index
that combined growth rate and stem straightness, clonally propagated and established in
a clonal seed orchard in southern Vietnam. A study in the first generation progeny test
in northern Vietnam showed that narrow-sense heritabilities for growth traits ranged
from 0.12 to 0.33 at 11 years (Dao 2012), while heritabilities were higher (0.21–0.40)
for wood basic density, cellulose content, shrinkage, collapse, stiffness and strength.
Age-age genetic correlations were strong for growth traits measured at 3, 5, 7 and
11 years. To continue the breeding program, second-generation progeny tests were
established using individual seedlots collected from superior trees in the northern
progeny test and the clonal seed orchard.
Acacia mangium is known to have exceptionally low allelic diversity compared to other
forest trees, as assessed by isozyme analysis (Butcher et al. 1998). If there were correspondingly low levels of quantitative genetic variation in production traits, little genetic
gain could be made from breeding, beyond the gains of initial provenance selection.
Therefore, the aim of this paper was to determine the prospects for ongoing genetic
improvement of A. mangium in Vietnam for solid wood production. The specific questions
addressed are: (1) How strong is the genetic variation and degree of genetic control for

growth traits, stem straightness, pilodyn and dynamic stiffness of A. mangium from the
second generation progeny tests at Tuyen Quang, Ba Vi and Bau Bang?; (2) How strong
are the genotype by environment interactions for growth?; (3) How large are the predicted
gains from the breeding of A. mangium?

123


New Forests (2015) 46:577–591

579

Materials and methods
Material
A total of 164 families were used to establish the second-generation progeny tests in the
current study. Families were from 6 different genetic groups, including four groups
originating from PNG and FNQ, which were collected from the first-generation progeny
tests or clonal seed orchards, as well as two infusion groups from seed orchards in the
Philippines and Australia. Some of families come from sib (sister) trees descended from
the same mother (original seed tree in QLD or PNG) in the first-generation trials and the
clonal seed orchard. Numbers of families in each genetic group, families common to pairs
of sites and families across the three trials are shown in Table 1.

Location and trial description
The second-generation progeny tests were established at Ba Vi in Hanoi Province and Son
Duong, Tuyen Quang Province, both in northern Vietnam and Bau Bang in Binh Duong
province in southern Vietnam. Site conditions, experimental designs, site preparation and
fertilizer applications are described in Table 2. The trials were planted during the rainy
seasons of 2008 or 2009, in September at Ba Vi and Tuyen Quang, and August in Bau
Bang. Planting materials were 4-month-old seedlings, raised in polythene bags in a soilbased potting mixture. The tests at Ba Vi and Tuyen Quang used 10 replicates and 3-tree

plots, but the test at Bau Bang only used 6 replicates and 2-tree plots. All tests used rowcolumn designs generated by the computer program CycDesigN (Williams et al. 2002),
providing 2-dimensional incomplete blocking (rows and columns) within replicates.

Table 1 Number of A. mangium families in six genetic groups, families common to pairs of sites and
families in each progeny test
Genetic group

Total number of
families in group

Number of families in each site
Ba Vi

Tuyen Quang

Bau Bang

CSO-PNG

84

60

44

CSO-FN

11

5

7

9

SSO-PNG

40

23

14

31

SSO-FN

9

4

5

8

IN-SSO-FN

5

5

100

IN-SSO-PNG

15

15

164

112

70

Number of families common
to Bau Bang

50

35

Number of families common
to Tuyen Quang

54

Total families tested

Number of families tested
in all three sites

52

28

CSO, Families collected from Vietnam clonal seed orchard; SSO, families collected from Vietnam 1st
opened-pollinated seed orchard; IN-SSO, Families collected from international seed orchard; PNG, FNQ,
family group originating from PNG and FNQ

123


580

New Forests (2015) 46:577–591

Table 2 Description of A. mangium progeny tests

Latitude

Ba Vi–Ha Noi

Tuyen Quang

Bau Bang–Binh Duong

21°070 N

21°490 N

11°320 N

0

0

Longitude

105°26 E

105°13 E

105°560 E

Altitude

60 m

100 m

50 m

Soil type

Ferralitic clay loam with
heavy lateritization

Ferralitic clay loam

Sandy alluvium

Annual rainfall
(mm)

1680

1641

1917

Annual average
temperature (°C)

23.2

22.9

26.2

Site preparation

Slash burned and ripped

Slash burned and
holes dug

Slash burned and ploughed

Planting time

September 2008

September 2009

August 2009

Fertilizer (per tree)

3 kg cattle
manure ? 0.2 kg NPK

3 kg cattle
manure ? 0.2 kg
NPK

0.5 kg micro-organic
fertilizer ? 0.2 kg NPK

Replicates

10

10

6

Rows/replicate

16

10

10

Columns per
replicate

7

7

10

Trees per plot

3

3

2

Spacing

4m92m

4m92m

4m92m

Design

Assessments
Total tree height (HT), diameter at breast height (DBH) and stem straightness (STR) of all
trees were recorded annually to 4 years at Ba Vi and 3 years at the other two trials.
The stem straightness was scored using a scale with 5 classes:
1.
2.
3.
4.
5.

for a very crooked stem with [2 serious bends (serious bends occur when the main
stem is bent to the right or left with more than 20° along its axis);
for crooked stem with 2 serious bends;
for slightly crooked stem with 1 serious bend and/or [2 small bends (small bends
occur when the main stem is bent to the right or left with \10° along its axis);
for almost straight stem with 1–2 small bends and
for a perfectly straight stem.

Stem volume over bark of each tree was calculated for the last measure of each trial
using the following formula:
VOL ¼ p  HT  DBH 2 =4  0:45=10
where VOL is stem volume (dm3/tree), DBH (cm), HT (m) and 0.45 is the form factor. The
form factor of 0.45 was the same as that from a study of A. mangium in Sumatra, Indonesia
(Hardiyanto and Nambiar 2014).

123


New Forests (2015) 46:577–591

581

Pilodyn penetration was measured for all trees in the tests by using a 6 J Forest Pilodyn,
by removing a small section of the bark at 1.3 m above the ground and taking two readings
for each tree, one from the east side and one from the north. Acoustic measurements at the
Tuyen Quang trial were made using a FAKOPP microsecond timer (Ross 1999). This
instrument measures transmission of sound waves between a transmitting and a receiving
probe. Probes were positioned at 0.1 m and 2.0 m above the ground. Stress waves are
generated by tapping the transmitting probe with a hammer. Sound transit time is converted to sound flight velocity through the outer stem wood using the distance between the
two probes. This velocity is used to predict dynamic modulus of elasticity (MoEd in GPa)
according to:
MoEd ¼ Green density  velocity2
where green density is assumed to be a constant at 1060 kg m-3 for A. mangium (Moya
and Mun˜oz 2010).

Statistical analysis
Single-site analysis
Some of individuals with obviously hybrids morphology in the tets were excluded from
data sets before single-site analysis. Stem straightness was not normally distributed. It was
assumed that this trait was controlled genetically by an underlying polyfactorially-determined liability scale (Falconer and Mackay 1996), and that the given scores were caused
by imposed thresholds. Prior to analysis class scores were therefore transformed into
asymptotic ‘normal scores’ (Gianola and Norton 1981) in order to adjust for non-adequate
or variable spacing of classes and to improve the efficiency of subsequent analyses
(Ericsson and Danell, 1995).
The statistical analysis was based on individual tree observations according to the linear
mixed model (1):
y ¼ XB m þ XP p þ ZW w þ ZN n þ ZT t þ ZF f þ e
with y ¼ ðy01 ; y02 ; . . .; y0n Þ0 , m ¼ ðm01 ; m02 ; . . .; m0n Þ0 ,
w02 ; . . .; w0n Þ0 , n ¼ ðn01 ; n02 ; . . .; n0n Þ0 , f ¼ ðf 01 ; f 02 ; . . .; f 0n Þ0 ,

ð1Þ

p ¼ ðp01 ; p02 ; . . .; p0n Þ0 ,
e ¼ ðe01 ; e02 ; . . .; e0n Þ0 , X

w ¼ ðw01 ;
¼ RÈ XBi ,
X ¼ RÈ XPi , ZW ¼ RÈ ZWi , ZN ¼ RÈ ZNi , ZT ¼ RÈ ZTi and ZF ¼ RÈ ZFi , RÈ denotes the
direct sum, and i the number of traits from 1 to n, y is the vector of individual observations
for the different traits, m is the vector of fixed effect of replicate, p is the vector of fixed
effect of genetic group, w is the vector of random row within replicate effect, n is the
vector of random column within replicate effect, t is the vector of random effect of plot for
assessments at age 3 and age 4, f is the vector of random family within genetic group
effects, and e is the vector of random residuals. XB ; XP ; ZW ; ZN ; ZT and ZF are incidence
matrix relating m, p, w, n, t and f to y. The data analyses were implemented using ASReml
software (Gilmour et al. 2006).
Assuming a multivariate normal distribution (MND), the expected mean and covariance
were:

123


582

New Forests (2015) 46:577–591

3 2
WI
0

0
w
6n7 6 0
N

I
0
6 7 6
7 6
0
TI
V6
6 t7¼6 0
4f5 4 0
0
0
0
0
0
e
2

0
0
0
FA
0

3
0

0 7
7
0 7
7
0 5
RI

ð2Þ

where 0 is a null matrix, I is an identity matrix of order equal to the total number of rows,
columns, plots, genetic,
È
Éand residuals, respectively,
È É and Èis theÉ direct (Kronecker)
È
É product
operation. W ¼ rwi wj , N ¼ frni ni g, T ¼ rti tj , F ¼ rfi fi , and R ¼ rei ej are the
row, column, plot, family and residual variance–covariance matrices between trait i and j,
denoting variance when i = j. A is the additive genetic relationship matrix. To ensure that
the variance–covariance matrix was positive definite, restrictions were in some cases applied to the parameters. The significance of seed source effects was assessed using F-tests.

Genetic parameters
Age-age and trait–trait genetic correlations and heritabilities were simultaneously estimated based on multivariate REML analysis using model (2). Family variance (r2f ),
phenotypic variance (r2P ), plot variance (r2t ) and environmental variance (r2e ) for different
traits and ages were estimated using ASReml. The estimated variance components were
used to calculate the narrow-sense heritabilities for the characters under consideration.
Open-pollinated families in the progeny test came from open-pollinated parent trees in
seed orchards or wild stands. There is some relatedness among the families within firstgeneration seed orchards. Therefore, some degree of inbreeding (about 10 %) was expected, as a result of relatedness among families, particularly in the seed orchards the
coefficient of relationship within families was assumed to be 0.33, making heritability
values more conservative than if a value of 0.25 was assumed (Gilmour et al. 2006). The

additive genetic variance (r2a ), total phenotypic variance (r2P ) and individual-tree heritability (h^2 ) estimates were calculated as follows (Squillace 1974):
r2a ¼ 3r2f
r2P ¼ r2f þ r2t þ r2e and
h^2 ¼

r2f

r2a
þ r2t þ r2e

Coefficient of additive variation (CVa), additive genetic correlation (^
ra ) and phenotypic
correlation (^
rP ) between traits or between ages were estimated as:
100 ra
X
ra a
r^a ¼ 1 2
ra1 ra2
rP1 P2
r^P ¼
rP1 rP2

CVa ¼

where X is the phenotypic mean, ra1 a2 and rP1 P2 are the genotypic and phenotypic covariance between two traits, respectively. ra1 , ra2 and rP1 , rP2 are the genotypic and
phenotypic standard deviations of trait 1 and trait 2, respectively. Standard errors of the
estimates of heritabilities, genotypic and phenotypic correlations were calculated using a

123

New Forests (2015) 46:577–591

583

standard Taylor series approximation implemented in the ASReml program (Falconer and
Mackay 1996).
Predicted gain from genetic selection was calculated according to Mullin et al. (1992) as
GY ¼ in;N h^2Y rPY
where Gy is the predicted selection gain, and in,N is the intensity of selection (10 %) based
on selection of n genotypes from N tested. Values for in,N were derived from Becker
(1992).

Across-site analysis
Genetic correlations between sites were estimated based on multivariate REML analysis,
by treating measurements from different sites as different traits based on model (2). In the
across-site analysis R ¼ RÈ Rn where Rn is the individuals in trial n. All off-diagonal
elements were assumed to be zero for combinations of traits measured in different trials.
The aim of this analysis was to test the significance of genotype by environment
(G 9 E) interactions. Log likelihood ratio tests were used to test if the correlations were
significantly different from one, and also to test if the correlations between different pairs
of trials differed significantly.

Results
Differences between genetic groups for the growth traits (HT, DBH and VOL) were
significant only in the Ba Vi test. In contrast, significant differences among genetic groups
were found for the quality traits of STR and pilodyn at all three tests, and MoEd at Tuyen
Quang (Table 3). Trees descended from the PNG provenances grew slightly faster than
trees from the FNQ provenances at Ba Vi. The infusion families of international seed

orchards grew slower than the selected families from seed orchards in Vietnam. However,
families originating from FNQ had higher density (lower pilodyn) and higher dynamic
modulus of elasticity than those from PNG. The best growth was recorded at Bau Bang,
where mean annual increment of DBH and HT was 4.3 cm/year and 4.3 m/year respectively, followed by Tuyen Quang (2.6 cm; 2.5 m) and Ba Vi (2.0 cm; 1.8 m). Ranking of
families (data not presented) indicated that the best performing families in the three sites
were among those selected from the clonal seed orchard at Bau Bang.
Heritability estimates for year 3 or 4 growth traits, pilodyn and dynamic modulus of
elasticity ranged from low to moderate at all sites (Table 4). The heritabilities of growth
traits in the Ba Vi test at 4 years were higher than those at 3 years in the other two sites.
Heritabilities for stem straightness were low at all sites. The calculated coefficient of
additive variation (CVa) was moderate for all studied traits at the three sites, ranging from
4.9 % to 9.4 % for HT and DBH. CVa for DBH was higher than for HT at all sites.
Similarly, CVa ranged from 6.1 to 7.3 % for STR, Pilodyn and MoEd (Table 4). Predicted
gain from selection of the best 10 % of trees varied from 12.4 to 15.6 % for growth traits,
8.7–10.3 % for stem straightness and 10.8–11.8 % for pilodyn at three tests and 11.3 % for
MoEd at Tuyen Quang test.
Within each site, positive genetic correlations were observed among all the combinations of traits that were assessed, except for correlation between DBH and MoEd in the
Tuyen Quang test (Table 5). Stem straightness (STR) showed consistent positive genetic

123


584

New Forests (2015) 46:577–591

Table 3 Genetic group means for studied traits at age 3–4 years in the 2nd progeny test
Genetic
group

Survival
(%)

DBH
(cm)

HT
(m)

VOL
(dm3/tree)

STR
(score)

Pilodyn
(mm)

MoEd
(GPa)

At Ba Vi (4 year old)
CSO-PNG

82.4

8.3

7.2

20.1

3.8

14.3

CSO-FNQ

90.0

8.1

7.1

19.1

3.5

13.0

SSO-PNG

85.0

8.1

7.2

18.2

3.6

13.8

SSO-FNQ

84.2

7.9

7.0

17.9

3.4

12.3

IN-SSO-PNG

83.2

8.0

7.1

16.2

3.5

15.1

IN-SSO-FNQ

81.7

7.3

6.9

15.5

3.8

13.4

F. probability

**

***

***

***

***

***

At Tuyen Quang (3 year old)
CSO-PNG

72.8

8.9

7.6

23.6

3.8

14.0

15.7

CSO-FNQ

78.6

8.3

7.2

20.1

3.5

11.5

18.3

SSO-PNG

76.7

7.8

6.9

16.5

3.6

13.3

16.8

SSO-FNQ

75.0

7.2

6.6

13.7

3.3

10.7

18.8

F. probability

ns

ns

ns

ns

***

***

***

At Bau Bang (3 year old)
CSO-PNG

75.0

13.2

13.2

99.1

3.4

14.4

CSO-FNQ

73.0

13.0

12.7

94.3

3.7

13.1

SSO-PNG

78.4

13.0

12.7

93.9

3.6

14.3

SSO-FNQ

74.2

13.2

12.5

87.8

3.3

12.9

F. probability

ns

ns

ns

ns

***

***

ns not significant
* F-probability \ 0.05; *** F-probability \ 0.001

correlations with DBH. The estimates obtained were from 0.25 to 0.37 at three sites. The
positive correlations between DBH and pilodyn, and the negative correlation between
DBH and MoEd were non-significant.
The log likelihood ratio test showed that the genetic correlations for growth traits
between Tuyen Quang and Bau Bang, and between Ba Vi and Bau Bang, were different
from 1, but the correlations between Tuyen Quang and Ba Vi did not differ significantly
from 1. Family-by-site interactions between the two northern sites and Bau Bang were
significant (p \ 0.001) for Pilodyn (Table 6). Genetic correlations for STR between pairs
of sites were all positive and non-significant (p [ 0.05).

Discussion
Growth performance
Comparisons of growth rate between sites showed that VOL at Ba Vi were lower at age of
4 years, having \20 % of the growth at Tuyen Quang and 85 % of the growth at Bau
Bang. In a series of species and provenance trials, Kha (2003) and Nghia (2003) also
reported that A. mangium at Tuyen Quang and Bau Bang performed better than Ba Vi. At
Tuyen Quang, site conditions are more favourable for A. mangium than at Ba Vi, with

123


New Forests (2015) 46:577–591

585

Table 4 Mean values, narrow-sense heritability (h^2 ), additive coefficient of variation (CVa) and predicted
gain (GY %) based on a selection intensity of 1.755 (10 % selected) for studied traits at different ages in
each site
h^2

CVa

GY %

8.12

0.26 ± 0.05

8.9

11.6

7.12

0.22 ± 0.05

4.9

10.8

19.53

0.25 ± 0.05

3.48

0.15 ± 0.04

7.0

10.3

13.94

0.25 ± 0.05

7.1

12.3

7.85

0.27 ± 0.06

9.4

15.4

7.57

0.24 ± 0.06

6.1

12.8

18.12

0.29 ± 0.06

Trial

Age

Trait

Unit

Mean

Ba Vi

4

DBH

cm

4

HT

m

4

VOL

dm3/tree

4

STR

Score

4

Pilodyn

mm

3

DBH

cm

3

HT

m

3

VOL

dm3/tree

3

STR

Score

3

Pilodyn

mm

Tuyen Quang

Bau Bang

3.26

0.10 ± 0.04

7.0

9.9

17.52

0.21 ± 0.05

6.1

10.8

7.82

3

MoEd

GPa

0.15 ± 0.05

7.3

11.8

3

DBH

cm

13.0

0.14 ± 0.09

9.3

13.3

3

HT

m

12.9

0.30 ± 0.09

7.7

12.4

3

VOL

dm3/tree

95.6

0.17 ± 0.08

3

STR

Score

3

Pilodyn

mm

3.54

0.11 ± 0.09

6.4

8.7

13.68

0.22 ± 0.05

7.4

11.7

deep, fertile soil (Nghia 2003). Also, the soil and climate are more favourable at Bau Bang
(the south of Vietnam) than at Ba Vi (the north), there being deep soil and light soil texture
as well as no temperature limitation in the winter months (Nghia 2003). The soil at Ba Vi

was a yellow ferralitic, clay loam with strong laterisation evident in the profile, acidic (pH
3.5–4.5), and infertile, with low levels of phosphorus and potassium (Kha 2003).
The natural provenances were not tested in the present study. Number of families in
each genetic group is not similar, some groups have 14–60 families, but some groups only
have 4–9 families (Table 1). However, the results at age three/four years as reported here
clearly showed that trees derived from PNG provenances were noted not only for fast
growth but also for high frequency of single-stemmed trees with a good clear bole, as
shown by high scoring for stem straightness (Table 3) and low frequency of upright
branches on the lower part of the stem (data not presented). Therefore, additional imports
and selections focusing on these provenances are recommended to augment the genetic
base of these provenances already established in the second-generation progeny tests in
Vietnam. Consistent differences in growth performance among natural provenances of A.
mangium have been demonstrated in many trials across a number of countries (Harwood
and Williams 1992). Provenances from the south west of Western Province, PNG, and
adjacent Western Papua display the fastest growth, followed by Claudie River from FNQ
(13°S) and then provenances from further south in Queensland (16°–18°S), with outlying
provenances from the Indonesian island of Ceram, and Piru in Western Papua, growing the
slowest. The significant genetic variation in pilodyn and dynamic modulus of elasticity
demonstrated in our study was also reported in other studies in A. mangium (Dao 2012;
Thinh et al. 2011) and A. auriculiformis (Aggarwal et al. 2002; Firmanti et al. 2007; Kumar
et al. 1987; Mahat 1999). Mossman (QLD Cairns Region) was the best provenance of A.

123


586
Table 5 Genetic (^
ra ), phenotypic correlations (^
rP ) and standard errors of correlations within
parenthesis between growth traits

and quality traits (straightness,
pilodyn and dynamic stiffness in
the progeny tests at Ba Vi, Tuyen
Quang and Bau Bang)

New Forests (2015) 46:577–591

r^a

Trait–trait

r^P

In Ba Vi test
DBH versus VOL

0.99 ± 0.01

0.97 ± 0.001

HT versus VOL

0.97 ± 0.03

0.75 ± 0.009

DBH versus STR

0.25 ± 0.15

0.14 ± 0.02

DBH versus Pilodyn

0.31 ± 0.14

0.05 ± 0.02

In Tuyen Quang test
DBH versus VOL

0.99 ± 0.006

0.96 ± 0.002

HT versus HT

0.86 ± 0.07

0.67 ± 0.02

DBH versus STR

0.37 ± 0.23

0.10 ± 0.03

DBH versus Pilodyn

0.23 ± 0.15

0.06 ± 0.03

-0.31 ± 0.24

-0.06 ± 0.04

DBH versus MoEd
In Bau Bang test

Table 6 Genetic correlations
between sites for height, diameter, volume, stem straightness and pilodyn

DBH versus VOL

0.99 ± 0.02

0.96 ± 0.003

HT versus VOL

0.99 ± 0.05

0.79 ± 0.017

DBH versus STR

0.33 ± 0.03

0.02 ± 0.08

DBH versus Pilodyn

0.27 ± 0.17

0.03 ± 0.02

Trait
HT

Site

Ba Vi

Tuyen Quang

0.68 ± 0.17***

Tuyen Quang

0.75 ± 0.17***

Tuyen Quang

0.67 ± 0.18***

Tuyen Quang

0.25 ± 0.31ns

Tuyen Quang
Ba Vi

0.44 ± 0.41ns
0.48 ± 0.43ns

Ba Vi
Pilodyn

0.16 ± 0.37ns
0.08 ± 0.39ns

Ba Vi
STR

0.12 ± 0.32ns
0.11 ± 0.32ns

Ba Vi
VOL

0.01 ± 0.03ns
0.23 ± 0.39ns

Ba Vi
DBH

Bau Bang

0.71 ± 0.02***

0.65 ± 0.04***
0.63 ± 0.05***

mangium for mechanical properties (Ani and Lim 1993; Hazani 1994; Shanavas and
Kumar 2006).

Heritabilities and additive coefficients of variation
Individual heritabilities reported here for growth traits, pilodyn and MoEd in all three tests
ranged from low to moderate (from 0.17 to 0.30). Stem straightness at all sites showed low
heritability (0.10–0.15). Sib trees contributing progenies in Vietnam descended from the
same mother (original seed tree in QLD or PNG) are more closely related than is assumed
in normal heritability calculations (it is normally assumed that progenies are unrelated
through recent maternal descent) and this would tend to reduce our estimated heritabilities.
However, growth trait heritabilities are similar to heritabilities reported from other openpollinated family trials of A. mangium (Arnold and Cuevas 2003; Dao 2012; Nirsatmanto

123


New Forests (2015) 46:577–591

587

and Kurinobu 2002) and of other Acacia species (Arnold and Cuevas 2003; Dunlop et al.
2005; Hai et al. 2008b; Luangviriyasaeng and Pinyopusarerk 2002).
The additive coefficient of variation (CVa) for growth traits, pilodyn and MoEd at all
sites reported in this study were [5 %. At age 3 and 4 years, these CVa of growth traits
were generally highest at Tuyen Quang and Bau Bang. This may have been caused by
genetic variation being better expressed at an early age in more favourable environments
with a higher growth rate (Cotterill and Dean 1990). Also, greater competition would have

developed at both sites because of the much faster growth there, and this may have
accentuated differences between the families. Furthermore, the Bau Bang site was very
uniform, and intensive mechanical weed control with a tractor-plough was maintained after
establishment. These factors would reduce the environmental variance and thereby increase individual heritabilities and CVa of growth traits.
Low to moderate heritabilities, significant differences among genetic groups and
relatively high CVa suggest that considerable selection responses could be expected for
growth and wood quality traits, following selection of the best individuals for establishing
new clonal seed orchards to meet improved seed requirements for operational planting
targets in Vietnam. Typically, progeny trials can be selectively thinned to cull poorer trees
within families having slow growth and/or poor stem form, and completely remove the
poorest families, converting them into seedling seed orchards (SSOs). Another option is to
establish large areas of unpedigreed seed production areas (SPAs) using a broad and
improved genetic base, such as a mix of equal quantities of seed from 50 or more unrelated
superior trees in second-generation SSOs. These SPAs can either be dedicated solely to
seed production, or managed for both wood and seed production. In this second approach,
early selective thinning is undertaken, to promote development of heavy crowns on the
retained, phenotypically superior trees. Once the retained trees reach harvestable size,
clear-felling of the entire stand for a commercial wood harvest is timed to coincide with
ripening of a heavy seed crop deriving from a heavy, general flowering, which is then
collected from the felled trees. This approach was successful when tried on a pilot scale in
Sarawak, Malaysia in the early 2000s (Harwood, unpublished data). In principle, it should
produce, at low cost and with limited scientific inputs, large quantities of improved seed,
without compromising timber production. However, unpedigreed SPAs represent a dead
end from a breeding perspective, as all pedigree information is lost and no information on
genetic parameters is obtained. The other strategy is clonal family forestry, which is to
collect seed from the very best families in the second-generation tests and multiply these
seedlots via CFF (clonal family forestry (White et al. 2007) which is suited to tree species
such as A. mangium where it is difficult to propagate tested individual clones because of
maturation problems.

Correlations between traits
DBH had a slightly negative correlation with MoEd in Tuyen Quang test and non-significant correlations with pilodyn in all three tests. This is consistent with the underlying
non-significant negative relationship between growth rate and basic density or modulus of
elasticity in previous studies of A. mangium (Dao 2012; Khasa et al. 1995). In other
hardwood species, such as Eucalyptus species, Q. petrea, Q. rubra, Petersianthus
macrocarpa and C. spruceanum, non-significant correlations between growth traits and
basic density, and shrinkage parameters have been reported (Chafe 1994; Nepveu 1984;
Sotelo Montes et al. 2007).

123


588

New Forests (2015) 46:577–591

At age 3 and 4 years, the trait–trait correlations between STR and growth traits were
consistently positive with a range from 0.25 to 0.37, indicating that larger diameters are
associated with straighter stems. In an early evaluation of first-generation progeny tests in
Vietnam, Dao (2012) found that HT and DBH were not generally correlated to stem
straightness. Positive correlations between growth and stem straightness were also reported
from low to high in other tropical species, such as A. auriculiformis, A. crassicarpa, A.
nilotica, Eucalyptus grandis and E. camaldulensis (Arnold and Cuevas 2003; Gapare 2003;
Mahmood et al. 2003; Ginwal and Mandal 2004; Hai et al. 2008a, b).

Genotype by environment interactions
To manage GxE interactions, the best families could be selected for specific sites to
maximize deployment gains (Libby and Rauter 1984). This would involve identifying
different plantation regions, representing homogenous environmental zones within which
selections are deployed and used in further breeding. Significant genotype by environment

(G 9 E) interaction effects were found in growth traits between the two northern sites and
Bau Bang; i.e. the most promising families were not the same at both sites. Figure 1a–c
show plots of family Best Linear Unbiased Prediction (BLUP) values for DBH in pairs of
sites. The figures illustrate that many families contribute to the interaction, and the

(b)
1.0

BLUP values of DBH at Ba Vi

BLUP values of DBh at Ba Vi

(a)
0.5
0.0
-0.5
-1.0

ra=0.75

-1.5
-1.5

-1.0

-0.5

0.0

0.5

1.0

2
1
0
-1

ra=0.11

-2
-3
-1.5

BLUP values of DBH at Tuyen Quang

-1.0

-0.5

0.0

0.5

1.0

BLUP values of DBH at Bau Bang

BLUP values of DBH at Tuyen
Quang

(c)
2
1
0
-1
-2
-3
-1.5

ra=0.12
-1.0

-0.5

0.0

0.5

1.0

BLUP values of DBH at Bau Bang

Fig. 1 Clonal BLUP of 28 the same families for diameter at breast height at a Ba Vi and Tuyen Quang,
b Ba Vi and Bau Bang, and c Tuyen Quang and Bau Bang

123


New Forests (2015) 46:577–591

589

tendency was the same for all traits. Therefore, G 9 E effects were clearly of practical
importance for breeding program of A. mangium, separate seed orchards should be
established for northern and southern Vietnam. In further research, progeny tests should be
also set up in central Vietnam to determine G 9 E differences between that region and the
north and south of the country. Significant G 9 E interactions were also reported in A.
mangium in Zaire (Libby and Rauter 1984) and Vietnam (Dao 2012), in A. auriculiformis
in Vietnam (Hai et al. 2008a) and A. mearnsii in South Africa (Dunlop et al. 2005).
The low genotypic correlations between northern and southern Vietnam in this study
could be explained by differences in both soil and climatic conditions. The experimental
sites in Tuyen Quang and Ba Vi are typical of most hill-sites in northern Vietnam with
yellow ferralitic clay loam soils, while the soil at Bau Bang is sandy alluvium, and light in
texture. The soils are fertile at Tuyen Quang and Bau Bang, but infertile and heavily
laterized at Ba Vi (Kha 2003). In addition, lower temperature as well as reduced light in
winter months reduce growth of the trees at Tuyen Quang and Ba Vi.

Conclusions
In second-generation progeny tests of A. mangium in Vietnam, significant differences
between families were found for growth traits, stem straightness, pilodyn penetration and
predicted MoEd. Heritabilities of the studied traits ranged from low to moderate in all sites.
The CVa for DBH, pilodyn and MoEd were high after age 3 years. Genetic correlations
between quality traits (pilodyn and dynamic modulus of elasticity) and growth traits were
weak and favourable or unfavourable with large standard errors. The substantial coefficients of additive genetic variation and significant heritabilities for all traits indicate that it
should be possible to use a selection strategy that combines improvements in growth, stem,
and wood quality for A. mangium in Vietnam. The genetic correlations between Tuyen
Quang and Bau Bang site were low for all studied traits, indicating that G 9 E effects are
of practical importance for growth and that different deployment populations are required
for different sites.

Acknowledgments The second-generation progeny tests used in this study were established by the Institute of Forest Tree Improvement and Biotechnology in collaboration with CSIRO Sustainable Ecosystems
and CSIRO Plant Industry. The authors acknowledge Dr. Chris Harwood for his editorial assistance and
comments on the manuscript, and staff in the Institute of Forest Tree Improvement and Biotechnology in
Hanoi, Ba Vi station, An Hoa pulp mill and the Forest Science Institute of Southern Vietnam who worked on
establishment, maintenance of the trials and data collection over several years. This study was funded by a
national project named ‘‘Improvement of Acacia crassicarpa and Acacia mangium for plantation’’.

References
Aggarwal PK, Chauhan SS, Karmarkar A (2002) Variation in growth strain, volumetric shrinkage and
modulus of elasticity and their inter-relationship in Acacia auriculiformis. J Trop For Prod 8:135–142
Ani S, Lim SC (1993) Variation in specific gravity of five-year-old Acacia mangium from the Batu Arang
Plantation, Selangor, Malaysia. J Trop For Sci 6:203–206
Arnold R, Cuevas E (2003) Genetic variation in early growth, stem straightness and survival in Acacia
crassicarpa, A. mangium and Eucalyptus urophylla in Bukidnon Province, Philippines. J Trop For Sci
15:332–351
Becker WA (1992) Manual of quantitative genetics. Academic Enterprises, London

123


590

New Forests (2015) 46:577–591

Butcher PA, Moran GF, Perkins HD (1998) RFLP diversity in the nuclear genome of Acacia mangium.
Heredity 81:205–213
Chafe SC (1994) Relationship between shrinkage and specific gravity in the wood of Eucalyptus. Aust For
57:59–61
Cotterill PP, Dean CA (1990) Successful tree breeding with index selection. CSIRO Australia
Dao DN (2012) Genetic variation of growth traits and wood properties of Acacia mangium in provenance

and first generation progeny tests. Dissertation, Vietnamese Academy of Forest Sciences
Dunlop RW, Resende MDV, Beck SL (2005) Early assessment of first year height data from Five Acacia
mearnsii (black wattle) sub-populations in South Africa using REML/BLUP. Silvae Genet 54:166–174
Ericsson T, Danell O (1995) Genetic evaluation, multiple-trait selection criteria and genetic thinning of
Pinus contorta var. latifolia seed orchards in Sweden. Scand J For Res 10:313–325
Falconer DS, Mackay TFC (1996) Introduction to quantitative genetics. Pearson Education Limited,
Singapore
Firmanti A, Komatsu K, Kawai S (2007) Effective utilization of fast-growing Acacia mangium willd.
Timber as a structural material. J Trop Wood Sci Technol 5:29–37
Gapare WJ (2003) Genetic parameter estimates for growth traits and stem straightness in a breeding seedling
orchard of Eucalyptus grandis. J Trop For Sci 15:613–625
Gianola D, Norton HW (1981) Scaling threshold characters. Genetics 99:357–364
Gilmour AR, Gogel BJ, Cullis BR, Welham SJ, Thompson R (2006) ASReml User Guide Release 2.0. VSN
International Ltd, London
Ginwal HS, Mandal AK (2004) Variation in growth performance of Acacia nilotica Willd. ex Del.
Provenances of wide geographical origin: six year results. Silvae Genet 53:264–269
Hai PH, Jansson G, Harwood C, Hannrup B, Thinh HH (2008a) Genetic variation in growth, stem
straightness and branch thickness in clonal trials of Acacia auriculiformis at three contrasting sites in
Vietnam. For Ecol Manag 255:156–167. doi:10.1016/j.foreco.2007.09.017
Hai PH, Jansson G, Harwood C, Hannrup B, Thinh HH, Pinyopusarerk K (2008b) Genetic variation in wood
basic density and knot index and their relationship with growth traits for Acacia auriculiformis A.
Cunn ex Benth in Northern Vietnam. NZ J For Sci 38:176–192
Hardiyanto EB, Nambiar EKS (2014) Productivity of successive rotations of Acacia mangium plantations in
Sumatra, Indonesia: impacts of harvest and inter-rotation site management. New For 45:557–575.
doi:10.1007/s11056-014-9418-8
Harwood CE, Nambiar EKS (2014) Productivity of acacia and eucalypt plantations in Southeast Asia. 2.
Trends and variations. Int For Rev 16:249–260
Harwood CE, Williams ER (1992) A review of provenance variation in growth of Acacia mangium. In
Carron LT, Aken KM (eds) Breeding technologies for tropical Acacias. ACIAR Proceedings No. 37,
pp 22–30

Hazani O (1994) Physical and mechanical properties of Acacia mangium Willd and A. auriculiformis A.
Cunn ex. Benth from different sites and provenances: Dissertation, University Putra Malaysia
Kha LD (2003) Chon tao giong va nhan giong cho mot so loai cay trong rung chu yeu o Viet Nam.
Agriculture Publishing House, Ha Noi
Khasa PD, Li P, Vallee G, Magnussen S, Bousquet J (1995) Early evaluation of Racosperma auriculiforme
and R. mangium provenance trials on four sites in Zaire. For Ecol Manag 78:99–113
Kumar P, Anathanarayana AK, Sharma SN (1987) Physical and mechanical properties of Acacia auriculiformis from Karnataka. Indian For 113:567–573
Libby WJ, Rauter RM (1984) Advantages of clonal forestry. For Chron 60:145–149
Luangviriyasaeng V, Pinyopusarerk K (2002) Genetic variation in second-generation progeny trial of Acacia
auriculiformis in Thailand. J Trop For Sci 14:131–144
Mahat MN (1999) Genetic variation of growth and selected wood properties of four years old Acacia
auriculiformis provenances at Serdang Selangor. Putra University of Malaysia
Mahmood K, Marcar NE, Naqvi MH, Arnold RJ, Crawford DF, Iqbal S, Aken KM (2003) Genetic variation
in Eucalyptus camaldulensis Dehnh. for growth and stem straightness in a provenance-family trial on
saltland in Pakistan. For Ecol Manag 176:405–416
Moya R, Mun˜oz F (2010) Physical and mechanical properties of eight fast-growing plantation species in
Costa Rica. J Trop For Sci 22:317–328
Mullin TJ, Morgenstern EK, Park YS, Fowler DP (1992) Genetic parameters from a clonally replicated test
of black spruce (Picea mariana). Can J For Res 22:24–36
Nambiar EKS, Harwood CE (2014) Productivity of acacia and eucalypt plantations in South-East Asia 1.
Biophysical determinants of production: opportunities and challenges. Int For Rev 16(2):225–248
Nepveu G (1984) Genetic control of wood density and shrinkage in three oak species (Quercus petraea,
Quercus robur and Quercus rubra). Silva Genet 33:110–115

123


New Forests (2015) 46:577–591

591

Nghia NH (2003) Phat trien cac loai Keo Acacia o Viet Nam. Agriculture Publishing House, Ha Noi
Nirsatmanto A, Kurinobu S (2002) Trend of within-plot selection practiced in two seedling seed orchards of
Acacia mangium in Indonesia. J For Res 7:49–52
Ross RR (1999) Using sound to evaluate standing timber. Int For Rev 1:43–44
Shanavas A, Kumar BM (2006) Physical and mechanical properties of three agroforestry tree species from
Kerala, India. J Trop Agric 44:23–30
Sotelo Montes C, Beaulieu J, Hernandez RE (2007) Genetic variation in wood shrinkage and its correlations
with tre growth and wood density of Calycophyllum spruceanum at an early age in the Peruvian
Amazon. Can J For Res 37:966–976. doi:10.1139/X06-288
Squillace AE (1974) Average genetic correlations among offspring from open-pollinated forest trees. Silvae
Genet 23:149–156
Thinh HH, Hai PH, Kien ND (2011) Selection, breeding and propagation of some main plantation tree
species in Vietnam. Agriculture Publishing House, Ha Noi
Turnbull JW, Midgley SJ, Cossalter C (1997) Tropical Acacias planted in Asia: an overview of recent
developments in Acacias planting. In: Turnbull JW, Crompton HR, Pinyopuserak K (eds) Recent
developments in Acacia planting. ACIAR Publishing, Ha Noi, pp 14–18
White TL, Adams WT, Neale DB (2007) Forest genetics. CABI Publishing, Massachusetts
Williams ER, Matheson AC, Harwood CE (2002) Experimental design and analysis for tree improvement.
CSIRO Publishing, Canberra

123