基盤研究(B)(2):森林の潜在的生産力と撹乱体制を考慮した生態的ゾーニング手法の開発 (課題番号：14360090)
4.4. Effects of light and microsite conditions on tree size of 6-year-old Cryptomeria japonica
planted in a group selection opening
・Nobuya Mizoue・
・Satoshi Ito・
・Akio Inoue・
・Kotaro Sakuta・
・Hiroyuki Okada
Toru Kohama・
Abstract
We examined the extent to which direct and indirect measures of light and microsite condition could explain variation
in the size of 6-year-old Cryptomeria japonica trees planted in a group selection opening of about 0.4 ha on a steep
slope at Shiiba, Miyazaki Prefecture, southern Japan. We first examined directly measured light conditions from the
gap light index (GLI) and directly measured soil thickness (ST). We then examined the relationship between tree size
and indices obtained from indirect measurements. For light, we used the position of residual trees to estimate a
vertical competition index (VCI). For microsite, we used a digital elevation model (DEM) with a 2-m resolution to
calculate maximum slope (MS) and average slope (AS) in the area immediately surrounding trees. The multiple linear
regression using GLI and ST explained about 44% of variation in tree height (H) or diameter at the base (D), while
simple regression using only GLI explained about 35%. The contribution of ST was about half of GLI. The multiple
regressions using VCI, MS and AS did not explain any more variation than using VCI alone (R2 of about 0.26). GLI
was highly correlated with VCI (r=-0.8), while ST had a weak correlation with MS (r=0.35) and no correlation with
AS. We conclude that the heterogeneity in microsite conditions on steep slopes is an important factor determining tree
growth within a group opening in addition to the light gradient caused by edge effects. However, use of a DEM clearly
provides limited value in expressing variation in microsite condition for predicting tree size within small group openings,
while the indirect index of light condition (VCI) may be useful in practical forest management.
Key words competition index・edge effect・DEM・GLI・soil thickness
Introduction
In many countries, there is an increasing trend to
convert even-aged plantation forests into uneven-aged
stands (e.g. Cameron et al. 2001). In Japan, about 40%
of the forested area (ca. 10 million hectares) is covered
with plantation forests, mainly even-aged conifer stands
(Fujimori 2001), and forest management policy for these
simple conifer plantations has shifted away from largescale clearcutting towards uneven-aged methods (Nyland
1996).
Recently, the group selection system has received
considerable attention as an alternative for uneven-aged
stand management, as a compromise between the
environmental effects of clearcuts and the productivity
penalties associated with single-tree selection (York et
al. 2004). However, there are few examples of where the
group selection system has been used in conifer
plantations in Japan (Fujimoto 1984; Yamashita et al.
2005). Therefore there is a lack of basic information on
the growth of planted trees in group selection openings,
and the factors affecting the growth of regenerated trees
have not been well documented.
In group selection openings, the existence of residual
trees surrounding the openings causes gradients in
resource availability for the regenerated trees. York et al.
(2003) examined the effects of light and water availability
on the height of 3-year-old trees planted in group openings
ranging in size from 0.1 to 1 ha. Of three species
examined, sensitivity to light and water availability was
highest for Giant sequoia and lowest for Douglas-fir, and
only light was a significant predictor of ponderosa pine
performance. Their study suggests that light is a critical
resource affecting tree growth planted in group selection
openings.
In Japan, most conifer plantations have been
established on steep slopes with complicated topography,
resulting in heterogeneity of microsite conditions for
regenerated trees. Geomorphic factors such as slope
aspect and slope steepness can substantially modify the
local environment of plants by altering microclimate
conditions and soil development at a small-scale
(Oberhuber and Kofler 2000). Therefore, we can assume
that in addition to light resources microsite conditions
have a significant affect on the growth of regenerated
trees on steep slopes.
The objective of this study was to examine the effects
of light and microsite conditions on tree size of 6-yearold Cryptomeria japonica planted in a group selection
opening on a steep slope. To assess light and microsite
conditions, we first examined two directly measured
indices, GLI (gap light index) and soil thickness. In
practical forest management, the intensive field surveys
77
4.4. 群状伐区におけるスギ植栽木の成長と光環境・微地形
needed to get data such as GLI and soil thickness would
be too time-consuming and impractical, yet there is a need
to improve growth prediction with such information; more
easily collected indices are needed. Therefore, we also
examined the effects of light and microsite indices
calculated from residual tree data (position and height)
and a digital elevation model (DEM).
(m)
N
40
20
Material and Methods
0
Study site and field surveys
The study site is located on a steep slope with an
average inclination of 38° at about 540 m to 600 m a.s.l.
at Shiiba in Miyazaki Prefecture in southern Japan (32°
27' N, 131°13' E). This site is owned by Sumitomo
Forestry Co., Ltd. The mean annual temperature and
annual rainfall are about 14 ! and 3,000 mm,
respectively. In 1998 at this site, group selection cuttings
of about 0.4 ha (about 80 m by 50 m) were conducted in
a 45-year-old even-aged stand of C. japonica. Then in
1999 C. japonica seedlings were planted (ca. 3,000
cuttings per hectare) in the openings.
In 2003, when the residual and planted trees were 51
and 5 years old, respectively, we measured the height,
DBH, and position of the residual trees surrounding the
opening. In 2004, we measured the height, diameter at
ground level (base), and position of all 685 of the 6-yearold trees planted within the opening. The canopy of the
young trees was not closed at the time of the field survey.
The position of all measured trees is shown in Fig. 1.
Table 1 presents a summary of data from the tree
measurements and light and microsite conditions.
Intensive measurement of light and microsite conditions
Of the 685 young planted trees, we selected a sample
of 93 trees for measuring light and microsite conditions
along two roughly south-north and east-west aligned
transects as shown in Fig. 1. To measure light conditions,
gap light index (GLI) was determined from color
hemispherical photographs taken at each of the sample
trees in still and overcast sky conditions using a digital
camera (Coolpix 990, Nikon Corporation, Tokyo, Japan)
with an exclusive fish-eye lens (Fish-eye converter FCE8, Nikon Corporation, Tokyo, Japan). The camera was
78
-20
-40
-20
0
20
(m)
Fig 1. Tree positions in the study site. Large symbols indicate the residual trees and small symbols indicate planted trees. The solid small symbols indicate the sample trees where GLI and soil
thickness were measured.
Table 1. General description of the survey site.
Mean
Residual trees
H (cm)
D (cm)
Planted trees
H (cm)
D (cm)
H (cm)
D (cm)
Light and microsit
GLI (%)
ST (cm)
VCI
AS (°)
MS (°)
SD
Number/ha
825
n
161
17.4
25
4.7
10.5
2.9
5.5
2.7
4.5
0.7
1.6
0.8
1.8
93
93
685
685
68.6
72.5
3.4
1
40.9
13.5
17.4
5.4
0.2
7.7
93
93
685
685
685
1713
mounted on a tripod at a height of about 1.0 m above the
ground and leveled with a bubble level. Image size was
set to XGA and resolution was set at Normal (Inoue et al.
2004). The digital images obtained from the hemispherical
photographs were analyzed using an image-processing
program LIA for Win32, available on the Internet at http:/
/hp.vector.co.jp/authors/VA008416. The data from each
processed image were used to calculate GLI over the
growing season (April to December) for this geographic
province using the following equation:
基盤研究(B)(2):森林の潜在的生産力と撹乱体制を考慮した生態的ゾーニング手法の開発 (課題番号：14360090)
Bdif + Bdir
× 100
GLI =
Adif + Adir
where Bdif and Bdir are the cumulative values of
diffusive light and direct light below the canopy during
the growing season, and Adif and Adir are the
corresponding values above the canopy.
There are many aspects of microsite condition that
could affect water availability and rooting capacity of the
planted trees, soil properties and micro-topography for
example. In this study we examined soil thickness because
it is closely related to water availability (Enoki et al. 1996).
Soil thickness at 4 or 5 points near the base of each sample
tree was measured using a soil auger (1 m length and 10
mm diameter). Soil thickness more than 1 m was recorded
as 1 m. Soil thickness (ST) for each sample tree was
calculated as the average of 4 or 5 measurements.
Indirect estimation of light and microsite condition
Of the resources limiting growth of individual trees in
forest stands, light is often assumed to be the most
important (Brunner and Nigh 2000). Mitsuda et al. (2002)
proposed a vertical competition index (VCI) as a surrogate
index for light conditions. VCI is calculated from the
difference in tree height and distance between the subject
tree and its competitors. We used a modified version of
the Mitsuda et al. (2002) index as proposed by Yamashita
et al. (2005):
 Hj + Zj − Zi 
 VCI = ∑ ATAN 
i =1
 DISTij 
n
(1)
where VCI is vertical competition index, ATAN is the
inverse tangent, DISTij is the distance between the subject
tree i and competitor tree j, Hj is the height of competitor
tree j, Zi is the elevation at the base of the subject tree i, Zj
is the elevation at the base of competitor tree j and n is
the total number of competitors. Increasing VCI values
represent an increasing degree of shading by residual
trees.
As in the study by Yamashita et al. (2005) we
determined the search range for competitors affecting a
subject tree based on the coefficients of determination of
simple linear regression models of tree size against VCI.
We calculated relative distance (rDISTij) defined as DISTij
divided by the elevation difference between the base of
the subject tree i and the tree tip of the competitor j. We
then incrementally changed rDISTij from 0 to 1 in steps
of 0.1, and determined the search range for effective
competitors from the value that gave the regression model
with the maximum adjusted R2.
To obtain an indirect measure of microsite condition,
we used the slope convexity/concavity and maximum
slope angle at each tree position, both of which may be
related to soil thickness (Tsai et al 2001). Using the
elevation measurements taken at each tree position, we
generated a digital elevation model (DEM) with 2 m
resolution based on the Akima interpolation method
(Akima 1978). We calculated average slope (AS) from
the cell containing the subject tree relative to its eight
neighboring cells as an index of convexity/concavity
(Blaszczynski 1997). In calculating average slope,
neighboring cells at a higher elevation than the subject
cell were expressed as positive values, those at a lower
elevation, negative values. Thus, positive average slope
values indicated concavity and negative values indicated
convexity. The maximum slope (MS) for each cell
containing a tree was determined as the largest absolute
slope angle of the eight values used to calculate AS.
Regression analysis
We first examined the effects light condition (GLI or
VCI) alone on tree height (H) and basal diameter (D) using
Table 2. The regression models used in this
study. Models 1 to 4 are based on indices developed from intensive field survey data (n=93)
and models 5 to 8 are based on indirectly estimated indices (n=685).
Model
1
2
3
4
5
6
7
8
Equation
H = aGLI + b
D = aGLI + b
H = aGLI + bST + c
D = aGLI + bST + c
H = aVCI + b
D = aVCI + b
H = aVCI + bMS + cAS + d
D = aVCI + bMS + cAS + d
H, tree height; D, diameter at the base; GLI, gap
light index; ST, soil thickness; VCI, vertical competition index; MS, maximum slope; AS, average slope; a, b, c, and d, parameters
79
4.4. 群状伐区におけるスギ植栽木の成長と光環境・微地形
simple linear regression. We then determined whether
microsite condition (ST or AS + MS) explained any further
variation when used together with light condition with
multiple linear regression. We also examined whether
there were correlations between the directly measured
environmental factors and the corresponding indirect
measures (GLI vs VCI; ST vs MS or AS) as a check on the
reliability of the indirect indices.
Results
Table 3. Results of the regression analysis.
Model R a2
1 0.35
2 0.358
3 0.433
4
0.447
5
6
7
0.262
0.259
0.266
8
0.252
Variable
GLI
GLI
GLI
ST
GLI
ST
VCI
VCI
VCI
MS
AS
VCI
MS
AS
SRC
0.584
0.298
0.59
0.306
-0.506
-0.073
-0.013
-0.492
-0.078
-0.013
p
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
0.027
0.7
<0.001
0.002
0.7
Discussion
All variable names are given in Table 1.Ra2, adjusted
R2, SRC, standardized regression coefficient.
It is known that within a group selection opening, light
is the critical factor influencing growth of regenerated
0.30
0.30
(a)
Adjusted R2
Table 3 shows the results of the simple and multiple
regression analyses. GLI only explained 35% and 36%
of variation in H and D, respectively. The multiple linear
regression models using both GLI and ST improved the
adjusted R2 (0.43 and 0.45 for H and D, respectively).
Standardized regression coefficients (SRC) showed that
the contribution of ST to the model was equivalent to
about half of that of GLI.
Fig.2 shows the relationships between search range,
rDISTij, and R2 for the regressions of H and D versus
VCI. The R2 values were largest at rDISTij =0.4 for H
and 0.5 for D. For calculating VCI, we therefore adapted
a search range for rDISTij of ≤ 0.4 for H and ≤0.5 for D.
VCI only explained 26% of variation in H and 25% in
D. The multiple linear regression models using VCI, MS
and AS did not improve the coefficient of determination.
MS had a significant negative influence on both H and
D, but the contribution of MS to the models was relatively
small (about 14% of that of VCI). AS was not a significant
variable in the regressions.
Fig. 3 and Fig. 4 show the relationships between the
directly measured environmental factors and their
corresponding indirect index. GLI was highly correlated
with VCI (r=-0.80, p<0.001) (Fig.3), and ST had a weak
correlation with MS (r=-0.348, p<0.001) and no
correlation with AS (p=0.71) (Fig.4).
0.25
0.25
0.20
0.20
0.15
0.15
0.10
0.10
0.05
0.05
0.00
0.00
0.0
0.2
0.4
0.6
rDISTij
0.8
1.0
(b)
0.0
0.2
0.4
0.6
0.8
1.0
rDISTij
Fig. 2. Relationship between the search range rDISTij for competitors and adjusted R2 for the regressions
of H (a) and D (b) on VCI.
80
基盤研究(B)(2):森林の潜在的生産力と撹乱体制を考慮した生態的ゾーニング手法の開発 (課題番号：14360090)
100
r = -0.80
(p<0.001)
GLI （ %）
80
60
40
20
0
0
2
4
6
8 10
VCI
12
14
16
Fig. 3. Relationship between light condition indices
obtained from intensive direct field measurement
(GLI) and indirect estimation based on the height
and position of residual trees (VCI).
120
120
r =-0.35
(p<0.001)
100
80
80
ST (cm)
ST (cm)
(p=0.71)
100
60
60
40
40
20
20
0
0
0
10
20
30 40 50
MS (degree)
60
70
-20 -15 -10
-5 0
5 10
AS (degree)
15
20
Fig. 4. Relationship between microsite condition indices obtained from intensive direct
field measurement of soil thickness (ST) and indirect estimation from a DEM (MS or AS).
trees, so the potential loss of productive growing space
for regenerated trees occurs at the perimeter of openings
(York et al. 2003; Coates 2000, Malcolm et al. 2001).
Yamashita et al. (2005) clearly revealed the negative
effects of shading by residual trees on the height growth
of 18- and 19-year-old Cryptomeria japonica planted in
about 0.1 ha openings on a flat site. The current study is
the first to examine the extent to which microsite
conditions can explain variation in the growth of
Cryptomeria japonica within openings on a steep slope,
as compared with the contribution of light condition.
This study found that adding soil thickness (ST) to light
condition (GLI) in the regression improved the
explanation of variation in tree size (Table 3). This
indicates that shallower soil reduced tree growth because
of less water availability (Enoki et al. 1996). The R2 of
our multiple regression models using both GLI and ST
was of a similar magnitude to that for the best model
(R2=0.47) obtained for giant sequoia by York et al. (2003),
who examined the effect of the percent of total transmitted
radiation, as measured by hemispherical photography, and
pre-dawn water potential on tree height of 3-year-old trees
of three species planted in group selection openings in a
Sierran mixed conifer forest, California. In addition, the
fact that ST contributed to explaining around half as much
variation as GLI (Table 3) suggests that in addition to
81
4.4. 群状伐区におけるスギ植栽木の成長と光環境・微地形
light gradient, the heterogeneity in microsite conditions
is an important factor determining tree growth within
group openings on steep slopes.
In actual forest management, indirect methods of
assessing site productivity using topographic maps and
DEMs are more practical than the intensive direct
measurement of soil physical and soil chemical properties
(Teraoka et al. 1991; Mitsuda et al. 2001). We attempted
to estimate light and microsite conditions indirectly using
2-m resolution DEM and residual-tree position data. Our
findings of the effective search range of residual-tree
competitors for estimating light condition index (VCI)
was within the range (0.3 to 0.6) obtained by Yamashita
et al (2005), even though there were large differences
between the two studies with respect to topography and
age of both the regenerated and residual trees.
Interestingly, there was a strong correlation between GLI
and VCI (Fig. 3). These results indicate that VCI is a very
useful indirect index of light condition for explaining
variation in tree size in group selection openings.
The explanatory power of our multiple regression
models using the indirect measures were less than those
obtained using direct measurements from intensive field
survey, i.e., from about 44% to about 26% (Table 3). It
was also found that the indices of microsite condition,
MS and AS did not contribute to the multiple regressions
(Table 3). This can be explained by the lack of correlation
between the directly measured ST and the surrogate
measures, MS and AS (Fig. 4). We can thus say that there
is a limitation in using DEM to express microsite
condition for predicting tree size within a group selection
opening.
In routine forest management, availability of a 2-m
resolution DEM would be unlikely. DEMs with cells of
about 30 m are more commonly available and can be used
to estimate site productivity effectively at a strategic forest
management planning mesoscale (ca. 1 km2-100 km2)
(Mitsuda et al. 2003). The results of our study imply that
a DEM, even at high resolution, cannot be used to explain
variation in tree growth in small openings of about 0.4
ha in size. The individual residual-tree positions used for
estimating the light condition index VCI in this study are
also not likely to be available in routine forest
management. However we can simulate variation in VCI
within an opening that has already existed or will be
established if a basic description of the residual stand,
such as stand density and average height, is available.
Such a simple simulation of light condition may help
82
foresters to design the size and shape of group selection
openings and to select appropriate species for planting in
the openings.
In conclusion, both light and microsite conditions are
critical factors influencing tree growth in group selection
openings on steep slopes. However, it should be noted
that DEMs are of limited value in expressing microsite
condition for predicting tree size within group openings,
but VCI, an indirect index of light condition, may be
practical for routine forest management.
Acknowledgements
We are grateful to Prof. S. Kai for his support and
encouragement to our study and to Dr. Y. Mitsuda for his
kind suggestions regarding the statistical analysis. We also
wish to thank the members of the Laboratory of Forest
Planning in the University of Miyazaki and Kyushu
University and at the Laboratory of Forest Ecology and
Environments in the University of Miyazaki for their
assistance in the field work. A part of this study was
supported by a Grant-in-Aid for Scientific Research by
JSPS (Nos. 14560122, 14360090, 14760102).
Literature Cited
Akima A (1978) A method of bivariate interpolation and
smooth surface fitting for irregularly distributed data
points. ACM Transact. Math Software 4:145- 159
Blaszczynski (1997) Landform characterization with
geographic information systems. PE & RS 63: 183191
Brunner A, Nigh G (2000) Light absorption and bole
volume growth of individual Douglas-fir. Tree Physiol
20:323-332
Cameron AD, Mason WL, Malcolm DC (2001)
Transformation of Plantation Forests: Papers presented
at the IUFRO Conference held in Edinburgh, Scotland,
29 August to 3 September 1999. For Ecol Manage
151:1-5
Coates KD (2000) Conifer seedling response to northern
temperate forest gaps. For Ecol Manage 127:249-269.
Enoki T, Kawaguchi H, Iwatsubo G (1996) Topographic
variations of soil properties and stand structure in a
Pinus thunbergii plantation. Ecol Res 11:299-309
Fujimori T (2001) Ecological and silvicultural strategies
for sustainable forest management. Elsevier.
Amsterdam.
基盤研究(B)(2):森林の潜在的生産力と撹乱体制を考慮した生態的ゾーニング手法の開発 (課題番号：14360090)
Fujimoto K (1984) Application of group-selection system
to even-aged sugi (Cryptomeria japonica D. Don) stand
(In Japanese). Mem Fac Agric Ehime Univ 29:4-113
Inoue A, Yamamoto K, Mizoue N, Kawahara Y (2004)
Effects of image quality, size and camera type on forest
light environment estimates using digital hemispherical
photography. Agric For Meteorol 126:89-97
Malcolm DC, Mason WL, Clarke GC (2001) The
transformation of conifer forests in Britain –
regeneration, gap size and silvicultural systems. For
Ecol Manage 151: 7-23
Mitsuda Y, Yoshida S, Imada M (2001) Use of GISderived environmental factors in predicting site indices
in Japanese larch plantations in Hokkaido. J For Res
6:87-93
Mitsuda Y, Ito S, Takata K (2002) Effects of competition
and cooperative interaction among neighboring trees
on tree growth in a naturally regenerated even-aged
Larix sibirica stand in considering height stratification.
J For Res 7: 185-191
Mitsuda Y, Ito S, Matsushita K, Fukuzato K, Taniguchi
Y, Mizoue N, Shimizu O (2003) ESC (Ecological Site
Classification) based DSS for Japanese forest landscape
suffering frequent rainstorms. Decision support for
multi purpose forestry CD-Proceedings (2003.4): 111
Nyland RD (1996) Silviculture; Concepts and
applications. McGraw-Hill. Boston
Oberhuber W, Kofler W (2000) Topographic influences
on radial growth of Scots pine (Pinus sylvestris L.) at
small spatial scales. Plant Ecol 146:231-240
Teraoka Y, Masutani T, Imada M (1991) Estimating site
index of sugi and hinoki from topographical factors
on maps for forest management. (in Japanese with
English abstract) Sci Bull Fac Agric Kyushu Univ 45:
125-133
Tsai C, Chen Z, Duh C, Horng F (2001) Prediction of
soil depth using a soil-landscape regression model: a
case study on forest soils in southern Taiwan. Proc.
Natl Sci Counc 25:34-39
Yamashita K, Mizoue N, Ito S, Inoue A, Kaga H (2005)
Variation in tree height of 18- and 19-year-old
Cryptomeria japonica planted in group selection
openings. J For Res (submitted)
York RA, Battles JJ, Heald RC (2003) Edge effects in
mixed conifer group selection openings: tree height
response to resource gradients. For Eco: Manage
179:107-121
York RA, Heald RC, Battles JJ, York JD (2004) Group
selection management in conifer forests: relationship
between opening size and tree growth. Can J For Res
34:630-641
83
84