Plant Ecology 140: 139–157, 1999.

© 1999 Kluwer Academic Publishers. Printed in the Netherlands.

139

Structure, composition and species diversity in an altitude-substrate

matrix of rain forest tree communities on Mount Kinabalu, Borneo

Shin-ichiro Aiba1 & Kanehiro Kitayama2

1 Graduate School of Environmental Earth Science, Hokkaido University, Sapporo 060, Japan (Present address:
Faculty of Science, Kagoshima University, Kagoshima 890, Japan); 2 Japanese Forestry and Forest Products
Research Institute, PO Box 16, Tsukuba Norin Kenkyu Danchi, Ibaraki 305, Japan

Received 29 January 1998; accepted in revised form 2 October 1998

Key words: Environmental matrix, Forest structure, Tree species diversity, Tropical rain forest, Ultrabasic
vegetation

Abstract
We studied forest structure, composition and tree species diversity of eight plots in an environmental matrix of four
altitudes (700, 1700, 2700 and 3100 m) and two types of geological substrates (ultrabasic and non-ultrabasic
rocks) on Mount Kinabalu, Borneo. On both substrate series, forest stature, mean leaf area and tree species
diversity (both ≥ 4.8 cm and ≥ 10 cm diameter at breast height [dbh]) decreased with altitude. The two forests
on the different substrate series were similar at 700 m in structure, generic and familial composition and tree
species diversity, but became dissimilar with increasing altitude. The decline in stature with altitude was steeper
on the ultrabasic substrates than on the non-ultrabasic substrates, and tree species diversity was generally lower
on ultrabasic substrates than on non-ultrabasic substrates at ≥ 1700 m. The forests on non-ultrabasic substrates
at higher altitudes and those on ultrabasic substrates at the lower altitudes were similar in dbh versus tree height
allometry, mean leaf area, and generic and familial composition at ≥ 1700 m. These contrasting patterns in forest
structure and composition between the two substrate series suggested that altitudinal change was compressed on
the ultrabasic substrates compared to the non-ultrabasic substrates. Tree species diversity was correlated with
maximum tree height and estimated aboveground biomass, but was not with basal area, among the eight study
sites. We suggest that forests with higher tree species diversity are characterized by greater biomass allocation to
height growth relative to trunk diameter growth under more productive environment than forests with lower tree
species diversity.

Introduction cally corresponds to a temperature gradient (Kitayama

Tropical high mountains provide a spectrum of rain
forest communities that are entirely composed of ever-
green species from the foot to the forest limit. This
unique condition enables us to evaluate the effects
of environmental conditions on the forest communi-
ties under controlled life-form, i.e., evergreen trees,
along the altitudinal gradient. By contrast, on temper-
ate mountains in northern Hemisphere, the dominant
life-form changes from broad-leaved deciduous (from
broad-leaved evergreen on warm temperate moun-
tains) to coniferous evergreen trees towards the forest
limits (Ohsawa 1993). An altitudinal gradient basi-
1992; Pendry & Proctor 1996), although it is also af-
fected by other environmental factors, especially by
humidity (Marrs et al. 1988; Kitayama 1995). It has
been well documented that both species diversity and
forest stature decrease with increasing altitude on trop-
ical mountains, and that other physiognomic features
such as leaf characteristics also systematically change
with altitude (Yamada 1977; Whitmore 1984; Ohsawa
et al. 1985; Gentry 1988; Proctor et al. 1988; Weaver
& Murphy 1990; Nakashizuka et al. 1991; Lieberman
et al. 1996). Decreasing availability of nutrients, espe-
cially nitrogen, with increasing altitude was suggested
to be a cause for these changes (Grubb 1977).
140
It is well known that tropical rain forest commu-
nities vary dramatically between different soil and/or
substrate types at similar altitudes (Austin et al. 1972;
Tanner 1977; Proctor et al. 1983). However, only a
few studies have examined the interactive effects of al-
titude and substrate on tree communities (e.g., Tracey
1982; Aplet & Vitousek 1994). Vegetation on soils de-
rived from ultrabasic rocks is distinctly different from
the usual zonal vegetation (Proctor & Woodell 1975;
Roberts & Proctor 1992). Yamanaka (1959) summa-
rized differences between climatic climax vegetation
and ultrabasic vegetation in western Japan. Wilson
et al. (1990) documented altitudinal change in plant
species diversity on both ultrabasic and schist sub-
strates in New Zealand. The only detailed account for
ultrabasic vegetation in tropical Southeast Asia is that
of Proctor et al. (1988). They studied the vegetation
on Mount Silam, a small mountain entirely formed by
ultrabasic substrates in North Borneo, and found that
the lowland forests are unexpectedly large-statured
while the montane forests are stunted. However, they
could not isolate the effects of altitude from those of
substrates because there was no control of vegetation
on non-ultrabasic substrate. Mount Kinabalu (4095 m
above sea level [asl]), the highest mountain in South-
east Asia between the Himalayas and New Guinea, is
an ideal place to study the effects of altitude and sub-
strate on plant communities, because primary forests
are preserved on both non-ultrabasic and ultrabasic
substrates side by side from ca. 700 m asl to the forest
limit (ca. 3500 m asl). We designed an array of study
plots consisting of a matrix of four altitudes and two
types of substrates.
The purpose of this paper is to describe forest
structure, floristic composition and tree species diver-
sity in this environmental matrix, which enables us to
assess the effects of altitude and substrates indepen-
dently by comparing the two altitudinal sequences on
ultrabasic versus non-ultrabasic substrates. We then
discuss possible determinants of the patterns found in
aboveground vegetation.
Methods
Study sites and the environment
On the eastern and southern slopes of Mount Kinabalu
(4095 m, 6◦ 050 N, 160◦330 E), Sabah, Malaysian Bor-
neo, a matrix of eight sites at four altitudes on two
geological substrate series (non-ultrabasic and ultra-
basic) was selected for study (Figure 1 and Table 1).
Seven of these sites (except the 17N plot, see Table 1
for plot abbreviation) were exactly the same sites as
the previous soil study (Kitayama et al. 1998), al-
though their altitudes were amended according to the
last land survey. The main massif and adjacent area
are gazetted as a state park and primary forests are
preserved inside. We selected sites on gentle slopes
(11–27◦), which are thought to be old-growth and
mostly of mature phase of forest regeneration cycles.
Selecting two sites on the different substrate series at
exactly the same altitude was not always possible due
to heterogeneous distribution of the substrates and the
precipitous topography.
The geological substrates of Mount Kinabalu are
dominated by Tertiary sedimentary rocks below ca.
3000 m and granite above ca. 3000 m, and ultra-
basic rocks distributed as patches up to ca. 3000 m
(Figure 1, Collenette 1964; Jacobson 1978). Vegeta-
tion and soils on non-ultrabasic substrates represent
a typical zonation of non-volcanic tropical mountains
(Kitayama 1992): four altitudinal zones represent low-
land, lower montane, upper montane and subalpine
forests, respectively. Vegetation of the whole range
of Mount Kinabalu Park was mapped by Kitayama
(1991).
The climate of Mount Kinabalu is humid tropical
with little yearly fluctuation. Mean annual air tem-
perature is 27.5 ◦ C at 0 m with a mean lapse rate of
0.0055 ◦ C m−1 (Kitayama 1992). The daily tempera-
ture difference diminishes from 8.5 ◦ C at 0 m to 4.5 ◦ C
at 3100 m. Rainfall shows irregular patterns both
within and between years, and prolonged dry spells
occasionally occur and result in severe droughts (Ki-
tayama 1996a). Mean annual rainfall is ca. 2500 mm
at 10 m, and increases with altitude probably up to
ca. 2800 m, above which an upper limit of a fre-
quent cloud belt marks a drier subalpine environment
(Kitayama 1992). We are monitoring climate with
data loggers at four altitudes (from November 1995
to October 1996 at 700, 1700 and 2700 m, and from
January to October 1996 at 3300 m); observed tem-
perature agreed well with that predicted from the lapse
rate reported by Kitayama (1992), but annual rainfall
obtained from the short-term records did not follow
the above-mentioned rainfall pattern across altitude
(Table 1).
Soil analysis demonstrated substantial differences
between the two substrates (Table 2; Kitayama et al.
1998; K. Kitayama, unpublished data). The amount of
organic carbon in top soils was consistently higher on
non-ultrabasic substrates than on ultrabasic substrates.
 141
Table 1. Outline, climate and vegetation zones of the eight study plots in a matrix of two substrate types and four altitudes on
Mount Kinabalu.

Common Sub- Abbre- Exact Plot Slope Aspect Estimated Annual Saturation Mean Vegetation
altitude stratea viation altitute area (◦ ) temperature rainfall deficit PAR zoned
(m) (m) (ha) ◦
( C)b c
(mm) (kPa) c −2 −1
(µmol m s ) c

700 N 07N 650 1.00 19 N85E 23.9 2509 0.522 361 lowland
U 07U 700 1.00 11 N80E 23.7
1700 N 17N 1560 0.50 17 S55W 18.9 2714 0.258 286 lower montane
U 17U 1860 0.20 24 S30E 17.3
2700 N 27N 2590 0.25 20 S20E 13.3 2085 0.147 254 upper montane
U 27U 2700 0.20 22 S5W 12.7
3100 N 31N 3080 0.20 27 S 10.6 3285e 0.222e 307e subalpine
U 31U 3050 0.06 19 S45W 10.7
a N, non-ultrabasic substrates (Tertiary sedimentary rocks at 700–2700 m, granite at 3100 m); U, ultrabasic substrates.
b Mean annual temperature estimated from mean lapse rate obtained by Kitayama (1992).
c Annual rainfall, mean saturation deficit and daily mean PAR (photosynthetically active radiation) calculated from the mea-
surement from November 1995 to October 1996 at 700, 1700 and 2700 m and from January to October 1996 at 3300 m.
d Vegetation zone on non-ultrabasic substrates after Kitayama (1992).
e Actually at 3300 m.

The content of inorganic nitrogen (NH4 + NO3 ) de-
creased with altitude on both substrate series, but the
difference between two substrate series was unclear.
The amount of exchangeable magnesium is two- to
four-fold higher on the ultrabasic than on the non-
ultrabasic series, the difference reflecting the high
amount of primary mineral magnesium contents in
the ultrabasic rocks. An earlier analysis of soil ni-
trogen mineralization suggested that environmental
factors affecting in-situ net nitrogen transformation
rates seemed to be contrasting between the two sub-
strate series (Kitayama et al. 1998); besides common
effects of decreasing temperature across increasing al-
titudes, elevated soil moisture due to poorer soil water
permeability more strongly reduced the rates in the
cloud belt (ca. 2100–2700 m) on the non-ultrabasic
series, while phosphorus deficiency seemed to limit
the rates more strongly on the ultrabasic series. These
results suggest that ultrabasic soils are poorer in bio-
logically available phosphorus but are better drained
than non-ultrabasic soils.
Plot censuses
Permanent plots of rectangular shape on horizontal
planes, consisting of 10 m × 10 m subplots, were
measured out with a surveying compass. Plot area was
differently chosen among sites because of different
tree density, stature and species diversity. The point
without any stem irregularities at around breast height
(ca. 1.3 m above ground) was selected along the trunk
for girth measurement, and trunk diameter at breast
height (dbh, cm) was calculated from the girth. But-
tressed trees were measured at well above (ca. 50 cm)
the protrusions. All trees ≥ 4.8 cm dbh (15 cm girth)
were measured in the six higher plots, but in the 07N
and 07U plots all trees ≥ 10 cm dbh were enumerated
within 1-ha plots while trees 10 cm > dbh ≥ 4.8 cm
were measured in two 10 m by 100 m transects (20
subplots, 0.2 ha in area). Multiple stems were sepa-
rately recorded and these stems except for the largest
one were defined as sprouting stems. Specimens (most
were infertile) were collected from all trees in the 07N
and 07U plots and from those trees that could not be
identified in situ in other plots. Tree censuses were
conducted from September 1995 to February 1996.
Specimens taken were identified at the Herbarium,
Forest Research Centre, Sabah Forestry Department at
Sandakan (nomenclature follows its index, see Appen-
dix 1 for authorities) from November 1995 to March
1996, and were stored in the Sabah Parks Herbarium
at the Kinabalu Park Headquarters. Bornean plants are
still being revised, thus it is difficult to identify to
species for some taxa, especially Syzygium (Eugenia
sensu lato, Myrtaceae) and Lauraceae. The plants that
could not be identified to species were distinguished
as separate species based on morphology of vegetative
characters (leaves, twigs and barks).
From October to December 1996, the plots were
revisited, and species identification was checked. On
this occasion, a 10 m wide transect was laid within
each plot (100 m long for the 07N and 07U plots
142
Figure 1. Map showing the location of the eight study sites on Mount Kinabalu, Borneo. Distributions of geological substrates for south-west
part of the Kinabalu Park from Jacobson (1978).
that were identical to one of the two transects where
smaller trees were measured for dbh, 60 m long for
the 17N plot, and 50 m long for the other plots except
at 31U where a portion of 20 m × 20 m was selected).
For the trees ≥ 2 m and < 12 m in height within these
transects, tree height, defined as a vertical distance
from trunk base to the highest leaves, was measured
directly by a measuring pole. The height of the trees
≥ 12 m high was estimated by measuring angles to
tree top, the top of a pole of known height standing
by the tree, and the ground level.
Leaf area of each species was estimated by a for-
mula: leaf area = 2/3 lamina length × width, based
on a pressed typical mid-crown leaf of the largest tree
(a leaflet in case of compound leaves). Each species
was then classified to one of the six leaf-size categories
(Webb 1959, see Table 4); megaphylls ≥ 1600 cm2 in
area were absent at our plots.
Data analysis
The relationship between dbh, D (cm), and tree height,
H (m), was regressed by an expanded allometric equa-
tion, which incorporates an asymptotic height term
(Ogawa 1969):
1 1 1 Ws = 0.02903(D2H )0.9813,
= h
+ ∗, (1)
H AD H
where A (m cm−h ), h (dimensionless) and H ∗ (m) are
regression constants. Trees that had lost most of their
crowns or were inclined ≥ 45◦ were excluded from
this analysis. Departure from linearity of the relation-
ship on ln-ln plots was examined by ANOVA using
the sum of squares of residuals (RSS) when data were
regressed by a simple allometric equation (degree of
freedom [df] = N −2) and RSS when regressed by (1)
(df = N − 3) (Aiba & Kohyama 1997). Between-site
differences in the relationship were tested by ANOVA
using RSS when data from two sites were pooled to
be regressed by (1) (df = N1 + N2 − 3) and RSS
when data from two sites were separately regressed by
(1) (df = N1 + N2 − 6). A sequential Bonferroni test
was applied to control the group-wide type I error rate
associated with multiple comparison (Rice 1989).
Aboveground biomass (kg m−2 ) was estimated us-
ing two published empirical allometric relationships.
One is obtained from a warm-temperate evergreen oak
forest in southern Japan (Nagano 1978):
D0.1 = 0.941D + 0.734
and
W = 0.0303D0.1 2 H, (2)
where D0.1 (cm) is the trunk diameter at 1/10 of the
tree height and W (kg) is the aboveground dry mass
of the tree. The other is reported for a tropical low-
land dipterocarp forest in East Kalimantan, Indonesian
Borneo (Yamakura et al. 1986):
Wb = 0.1192Ws1.059
and
Wl = 0.09146(Ws + Wb )0.7266, (3)
where Ws , Wb and Wl (kg) are the dry mass of the
trunk stem, branches and leaves of the tree, respec-
tively, and W = Ws + Wb + Wl . The two estimates
 143
Table 2. Nutrient contents of top soils up to 15 cm depth of the eight study plots (Kitayama et al. 1998). Values are means of
four composites, each of which are made of ten soil cores (37 mm diameter) taken systematically along transects laid out at
each study site.

Plots Organic Total Inorganic nitrogen Soluble Extractable Extractable

carbon nitrogen (g m−2 )c phosphorus calcium magnesium
(kg m−2 )a (g m−2 )b NH4 NO3 (g m−2 )d (g m−2 )e (g m−2 )e

Non-ultrabasic substrates
07N 3.38 241 0.45 1.20 0.18 2.0 3.6
17Nf 2.08 154 1.27 0.04 0.53 5.2 7.0
27N 3.02 159 0.49 0.05 0.36 1.1 5.6
31N 4.85 344 0.28 0.48 0.35 41.6 4.5

Ultrabasic substrates
07U 2.96 256 1.09 0.90 0.14 3.5 10.4
17U 1.61 135 1.25 0.03 0.04 29.7 13.3
27U 1.81 185 0.39 0.05 0.09 17.4 22.8
31U 2.86 213 0.42 nd 0.07 29.8 22.6
a Determined on fresh soil by the Walkley-Black wet digestion method.
b Digested by the micro Kjeldahl procedure with concentrated sulfuric acid and determined colorimetrically.
c Extracted with 1.5 N KCl solution and determined by the Burkard SFA-2 autoanalyzer.
d Extracted with hydrochloric-ammonium fluoride solution and determined colorimetrically.
e Extracted with 1.5 N KCl solution and determined by atomic absorption spectrometry.
f K. Kitayama, unpublished data.

using the two sets of equations for the Kinabalu plots
were very similar to each other. The regression be-
tween estimates by (2) and those by (3) among the
eight sites was: [estimate by (2)] = 0.989 [estimate by
(3)] + 0.024 (r 2 = 1.00, p < 0.001). For simplicity
we use estimates from (2) in this paper.
We calculated two indices of species diversity, i.e.,
Fisher’s α index and Shannon–Wiener index (using
stem number and natural logarithm), and Pielou’s in-
dex of evenness for each of the eight plots (Pielou
1975). We also calculated percentage similarity (PS)
between plots, defined as the sum of the smaller rel-
ative basal area of the taxa common to two plots.
Cluster analyses were conducted for the eight plots
by the unweighted pair-group method using arith-
metic average (UPGMA) with a measure of distance
between plots: (1–PS/100). Species and plots were or-
dinated by detrended correspondence analysis (DCA)
based on relative basal area of species at the eight
plots, using a debugged version of CANOCO with
strict convergent criteria (ter Braak 1988; Oksanen &
Minchin 1997). Resultant axis 1 (eigenvalue 0.984)
represented altitudinal trend irrespective of substrates:
in case of the ordination of plots, 07U had the lowest
value (0.00) and 31U had the highest value (17.65).
None of the subsequent axes (eigenvalues 0.599) gave
interpretable meanings. DCA was also applied to rela-
tive basal area of families, and scores on emerged axis
1 (eigenvalue 0.734) was used to order the families
according to altitudinal trend.
Results
Forest structure
Maximum tree height and the proportion of trees with
buttresses decreased with altitude (Table 3). Stem
density (≥ 4.8 cm dbh) and percentage of sprouting
stems increased with altitudes. These changes oc-
curred more steeply on ultrabasic substrates. Basal
area and stem density (≥ 10 cm dbh) increased with
altitude on the non-ultrabasic series but peaked at the
middle altitude on ultrabasic series at the range of al-
titudes investigated. Estimated aboveground biomass
and maximum trunk diameter showed similar patterns
as maximum tree height, but the 27N plot had a larger
value than the 17N plot (and also 27U > 17U for maxi-
mum trunk diameter). Apparent aboveground biomass
density (kg m−3 ), defined as aboveground biomass
divided by maximum tree height (cf., Kira & Shidei
1967), was correlated with basal area (Spearman rank
correlation, r = 0.76, p < 0.05 for trees ≥ 4.8 cm
dbh) and was relatively constant (0.62–1.50 kg m−3
for trees ≥ 4.8 cm dbh) compared with ten- and 15-
144

Figure 2. Frequency distribution of trunk diameter at breast height of stems in the eight study sites; stems < 5 cm dbh were excluded. Plots
sharing the same letters did not differ at P < 0.05.

fold differences in maximum tree height and biomass,

respectively. Thus, increasing biomass across the en-
vironmental gradients coincides with increasing forest
stature.
Frequency distribution of trunk diameter of all the
eight plots consistently showed that larger stems had
lower frequency (Figure 2), following negative power
function models (a straight line on double logarith-
mic scale) at the 07N and 07U plots and negative
exponential models (a straight line on semilogarith-
mic scale) at the other plots when the regressions by
the two models were compared in terms of coeffi-
cients of determination. On both substrate series, the
density of trees of smaller diameter increased with
altitude, and the skewness of the distributions be-
came smaller. These changes across altitude occurred
more steeply on the ultrabasic substrate than on the Figure 3. The relationship between trunk diameter at breast height
non-ultrabasic substrate. Diameter distributions were (dbh, cm) and tree height (m) fitted to expanded allometric equa-
not different between the two substrate series at 700 tions, [tree height] = 1/A{[dbh]h } + 1/H ∗ where H ∗ corresponds
and 1700 m, but were different at 2700 and 3100 m to an asymptotic tree height, at each of the eight study plots. Broken
lines represent the non-ultrabasic plots and solid lines represent the
(p < 0.05, Kolmogorov–Smirnov two-sample tests ultrabasic plots. Data for 07N (open circles) and 31U plots (filled
with Bonferroni correction). circles) are shown. Plots sharing the same letters did not differ at
P < 0.05. See Table 1 for abbreviations of the plots.
 145
Table 3. Structural features of the eight study plots for stems ≥ 10 cm dbh (for stems ≥ 4.8 cm dbh in parentheses).

Plots Basal Stem Max. Max. Estimated Sprouting Buttressed

area density tree dbh aboveground stems stems
(m2 ha−1 ) (ha−1 ) height (cm) biomass (%) (≥ 0.5 m high)
(m) (kg m−2 ) (%)

Non-ultrabasic substrates
07N 34.0 (36.2) 464 (1064) 46.8 133.8 48.1 (49.1) 1.7 5.8
17N 36.4 (40.0) 778 (1730) 30.0 75.9 28.0 (29.5) 0.3 3.6
27N 49.1 (53.5) 984 (2116) 20.6 81.7 29.5 (30.8) 6.0 0.0
31N 57.2 (64.0) 1945 (3665) 15.0 47.7 20.7 (22.3) 4.6 0.0
Ultrabasic substrates
07U 38.3 (40.7) 510 (1175) 65.4 195.7 54.2 (55.2) 0.2 8.2
17U 42.0 (49.9) 1420 (3535) 22.6 51.4 21.3 (24.0) 0.4 0.0
27U 32.6 (41.5) 1355 (3775) 14.2 51.9 10.8 (12.7) 4.4 0.0
31U 12.2 (25.1) 917 (4383) 6.1 22.3 1.9 (3.8) 10.5 0.0

The relationships between dbh and tree height
showed no significant departure from linearity on ln-
ln plots for five plots (Figure 3, ANOVA, p > 0.1),
while those for 07U, 17N and 27N plots showed sig-
nificant departure from linearity (p < 0.05). On each
of the substrates, trees at higher altitudes were shorter
for the same dbh. The regressed lines from the two dif-
ferent substrates were not different from one another
at 700 m (ANOVA, p > 0.1), but were different
at the higher altitudes (p < 0.001, p < 0.05
after Bonferroni correction): the forests on ultraba-
sic substrates were more stunted than those on the
non-ultrabasic substrates of the equivalent elevation.
When the two substrates at different altitudes were
compared, the regressed lines did not differ between
31N and 27U (p > 0.1) and the difference between
27N and 17U was not significant at p < 0.05 af-
ter Bonferroni correction. The disharmonic altitudinal
trends between basal area and aboveground biomass
(or maximum tree height) can partly be explained by
the altitudinal change in tree allometry: trees at the
lower altitudes had more slender trunks than trees of
comparable dbh at the higher altitudes.
Mean leaf area and proportion of species with
compound leaves decreased with altitude on both sub-
strates (Table 4). Mean leaf area did not differ between
an ultrabasic site at a given altitude and the non-
ultrabasic sites at the higher altitudes (07U and 17N,
07U and 27N, 17U and 27N, and 27U and 31N,
Scheffé tests, p > 0.05), but mean leaf area always
differed between a non-ultrabasic site at a given al-
titude and the ultrabasic sites at the higher altitudes
(p < 0.05).
Floristic composition
At the lowest altitude (700 m) Dipterocarpaceae
(mostly Shorea spp.) dominated the forests (Table 5).
At the higher altitudes Myrtaceae and Podocarpaceae
became dominant. These two families include sev-
eral genera that are most abundant at any of the six
higher plots, i.e., Leptospermum, Syzygium, Tristan-
iopsis, Dacrycarpus and Dacrydium. To illustrate the
relative importance of these genera on each plot, their
relative basal area and maximum dbh, together with
those of the three most abundant genera, are shown
in Figure 4. The genus Syzygium was prevalent be-
neath the emergent Shorea spp. and Tristaniopsis sp.
occurred as an occasional canopy tree at 700 m. These
two myrtaceous genera (Syzygium and Tristaniopsis)
became canopy dominants at the middle altitudes
(1700–2700 m on non-ultrabasic substrate and 1700 m
on ultrabasic substrate), where Dacrycarpus imbri-
catus and Dacrydium pectinatum were co-dominants.
At the higher altitude (3100 m on non-ultrabasic sub-
strate and 2700 m on ultrabasic substrate), Dacrycar-
pus kinabaluensis and Dacrydium gibbsiae were more
dominant than Syzygium spp. and Tristaniopsis sp.,
and Leptospermum recurvum appeared with relatively
small abundance. At 3100 m on ultrabasic substrates,
there existed almost pure stands of Leptospermum re-
curvum, which are similar as those in the summit
region (>3500 m asl) on non-ultrabasic substrates
(Kitayama 1991).
146

Table 4. Percentage of species with compound leaves, and of species in six leaf-size categories (size range in parentheses,
cm2 ) of the study plots for trees ≥ 4.8 cm dbh; leaf area of each species was measured as 2/3 lamina length × width, based
on a single typical mid-crown leaf. Between-plot differences were examined by Scheffé tests for log-transformed data; plots
sharing the same letters did not differ at P < 0.05.

Plots n Compound Lepto- Nano- Micro- Noto- Meso- Macro- No Mean

leaves phyll phyll phyll phyll phyll phyll data area
(< 0.25) (0.25–2.3) (2.3–20) (20–45) (45–180) (180–1600) (cm2 )

Non-ultrabasic substrates
07N 163 18.0 0.6 0.0 14.7 33.7 45.4 3.7 1.8 59.3 a
17N 121 6.7 1.7 1.7 36.4 36.4 23.1 0.8 0.0 35.3 bd
27N 23 0.0 0.0 8.7 60.9 26.1 4.3 0.0 0.0 18.5 cdg
31N 24 0.0 8.3 12.5 54.2 16.7 8.3 0.0 0.0 14.7 eg
Ultrabasic substrates
07U 161 14.0 0.0 1.9 23.6 41.6 28.6 2.5 1.9 46.9 abc
17U 43 4.8 4.7 11.6 48.8 20.9 11.6 2.3 0.0 26.6 def
27U 26 3.8 11.5 11.5 65.4 7.7 3.8 0.0 0.0 11.8 fg
31U 9 0.0 33.3 11.1 44.4 11.1 0.0 0.0 0.0 5.9 e

Figure 4. Relative basal area, maximum dbh and number of species per genera ≥ 10 cm dbh of 6 genera that were most abundant at any of the
eight study plots. Other genera among the three most abundant genera at each site are also shown.
 147
Table 5. Relative basal area (%) of families for trees ≥ 10 cm dbh of the eight study plots (figures
in bold-type are the three most abundant families at each plot). Families are arranged by scores on
axis 1 in DCA and by relative basal area when the scores are the same.

Family 07N 07U 17N 17U 27N 27U 31N 31U DCA
scores

Trigoniaceae – 1.3 – – – – – – −0.55

Ctenolophonaceae – 0.8 – – – – – – −0.55
Simaroubaceae – 0.6 – – – – – – −0.55
Verbenaceae – 0.4 – – – – – – −0.55
Icacinaceae – 0.2 – – – – – – −0.55
Araliaceae – 0.1 – – – – – – −0.55
Sterculiaceae 0.3 4.5 – – – – – – −0.530
Ulmaceae 0.0 0.4 – – – – – – −0.528
Bombacaceae 0.3 2.9 – – – – – – −0.52
Burseraceae 1.0 6.3 – – – – – – −0.51
Moraceae 0.6 2.8 – – – – – – −0.50
Thymelaeaceae 1.6 3.4 – – – – – – −0.471
Annonaceae 1.3 2.5 – – – – – – −0.465
Dipterocarpaceae 22.7 36.9 0.1 – – – – – −0.45
Leguminosae 3.8 4.3 – – – – – – −0.44
Ebenaceae 0.5 0.4 – – – – – – −0.415
Lecythidaceae 0.4 0.4 – – – – – – −0.413
Chrysobalanaceae 0.4 0.1 – – – – – – −0.37
Euphorbiaceae 6.3 2.1 0.2 – – – – – −0.33
Ixonanthaceae 8.0 – – – – – – – −0.31
Hypericaceae 1.2 – – – – – – – −0.31
Olacaceae 0.8 – – – – – – – −0.31
Tiliaceae 0.2 – – – – – – – −0.31
Anisophyllaceae 0.2 – – – – – – – −0.31
Melastomataceae 0.1 – – – – – – – −0.31
Anacardiaceae 1.6 0.3 0.1 – – – – – −0.27
Loganiaceae 3.5 – 0.1 0.1 – – – – −0.20
Sapindaceae 1.7 0.7 0.5 – – – – – −0.10
Lauraceae 7.7 4.4 3.4 – – – – – 0.00
Myristicaceae 1.0 1.4 1.1 – – – – – 0.12
Meliaceae 4.4 2.2 3.5 – – – – – 0.23
Flacourtiaceae 0.4 – 0.2 – – – – – 0.33
Sapotaceae 7.2 3.4 9.9 0.1 – – – – 0.49
Proteaceae 0.1 – – 0.1 – – – – 1.09
Compositae 0.4 – 0.6 – 0.4 – – – 1.12
Polygalaceae 1.5 1.3 – 6.9 – – – – 1.13
Rubiaceae 0.7 0.1 0.5 1.6 – – – – 1.15
Fagaceae 6.9 1.6 15.2 5.6 8.0 – – – 1.20
Guttiferae 0.7 1.0 5.4 3.5 – – – – 1.26
Araucariaceae – 3.5 – 19.2 – – – – 1.39
Myrsinaceae 0.1 0.3 – 1.8 0.3 – – – 1.45
Caprifoliaceae – – 0.1 – – – – – 1.50
Illiciaceae – – 0.1 – – – – – 1.50
Chloranthaceae – – 1.0 0.1 – – – – 1.718
Rhizophoraceae – – 1.7 2.2 – – – – 1.719
Celastraceae – – 1.3 0.7 0.4 0.3 – – 1.79
Elaeocarpaceae 0.1 – 1.4 – 2.5 – – – 1.88
Rutaceae 0.1 – – 3.0 – – – – 1.90
148
Table 5 continued.

Family 07N 07U 17N 17U 27N 27U 31N 31U DCA
scores

Magnoliaceae 2.2 0.6 0.6 0.3 18.3 1.6 2.3 – 1.93

Cunoniaceae – – 0.1 3.6 – 0.2 – – 2.07
Myrtaceae 7.1 8.0 32.2 37.3 40.0 33.8 26.3 86.6 2.28
Oleaceae 0.2 0.1 0.3 – 9.2 – – – 2.40
Clethraceae – – 0.3 – 2.6 – – – 2.55
Phyllocladaceae – – 3.2 0.1 7.9 0.6 1.5 1.4 2.58
Rosaceae – – – – 0.6 – – – 2.85
Myricaceae – – 0.4 0.1 – 0.1 0.8 – 3.42
Aquifoliaceae – 0.1 0.1 1.6 8.1 0.5 5.9 – 3.47
Podocarpaceae – – 14.5 10.9 – 51.4 36.3 12.1 3.88
Daphniphyllaceae – – – 0.2 – – 0.6 – 4.12
Theaceae – 0.2 1.9 1.2 – 7.4 14.9 – 4.26
Symplocaceae 0.1 – – – 1.8 1.3 6.4 – 4.43
Escalloniaceae – 0.1 0.1 – – 2.8 5.1 – 4.73
Indeterminate 2.8 0.8 0.2 – – – – – –

Percentage similarities in terms of genus and fam-

ily among the eight sites were 16.6% and 33.0%,
respectively, on average. Dendrograms drawn from
the percentage similarity indicates that the two forests
on different substrate series at 700 m were similar,
and that the two forests at the same altitude become
increasingly dissimilar between substrate series with
increasing altitude (Figures 5a and 5b). The forest on
non-ultrabasic substrates at higher altitudes was simi-
lar to forest on ultrabasic substrates at a lower altitude:
the pair of 31N and 27U was the most similar among
all possible pairs from the eight sites, and 27N was
more similar to 17U than to 27U.
Species that were among the 15 most abundant
at any of the eight plots are shown in Appendix 1.
Most species had very restricted distributions (Fig-
ure 6): for stems ≥ 10 cm dbh, 311 species (81.8%)
appeared at only one site (358 species [79.2%] for
stems ≥ 4.8 cm dbh). These species were defined as
‘restricted species’ in this paper. The percentage of
restricted species decreased with altitude, and was not
different between the two substrate types (Table 6, G
test, p > 0.05 except at 1700 m for stems ≥ 4.8 cm
dbh where p = 0.01). Reflecting such high percent-
age of restricted species, percentage similarities in
terms of species composition between plots were very
low, ranging from 0% to 43.3% with average 5.0% Figure 5. Dendrograms showing similarities between the eight plots
(Figure 5c). at different taxonomic levels for stems ≥ 10 cm dbh. (a) Family, (b)
genus, (c) species. See Table 1 for abbreviations of the plots.
 149
Table 6. Number of families, genera and species per plot, Fisher’s diversity index, Shannon–Wiener diversity index,
Pielou’s evenness index and percentage of restricted species that are absent from other plots for stems ≥ 10 cm dbh (for
stems ≥ 4.8 cm dbh in parentheses) at the eight study plots.

Plots Number Number Number Fisher’s Shannon–Wiener Pielou’s Percentage

of familiesa of generaa of species diversity diversity evenness of restricted
index index index species
(nit)

Non-ultrabasic substrates
07Nb 40 (41) 80 (88) 148 (163) 75.1 (75.0) 4.53 (4.61) 0.907 (0.905) 74.3 (66.3)
17N 31 (42) 51 (68) 84 (121) 32.9 (38.3) 3.93 (4.18) 0.886 (0.872) 78.6 (78.5)
27N 13 (18) 13 (19) 15 (23) 3.5 (4.9) 2.02 (2.23) 0.745 (0.712) 46.7 (30.4)
31N 10 (14) 14 (19) 17 (24) 3.6 (4.8) 2.36 (2.59) 0.833 (0.816) 35.3 (20.8)
Ultrabasic substrates
07Ub 37 (38) 84 (91) 148 (161) 68.8 (69.0) 4.54 (4.62) 0.909 (0.910) 75.7 (70.2)
17U 22 (27) 29 (39) 32 (43) 9.3 (10.1) 2.81 (3.10) 0.811 (0.825) 62.5 (58.1)
27U 11 (17) 14 (22) 15 (26) 3.4 (5.2) 2.21 (2.65) 0.817 (0.814) 20.0 (19.2)
31U 3 (6) 4 (8) 4 (9) 1.0 (1.8) 0.39 (0.57) 0.282 (0.258) 0.0 (0.0)
a Indeterminate species at the family and genus level were excluded from the calculation of number of families and genera,
respectively.
b Stems 10 cm > dbh ≥ 4.8 cm were enumerated in parts of plots (0.2 ha).

strates but occurred at different altitudes, 22 species

Figure 6. Frequency distribution of species that occurred in differ-
ent number of plots. Shaded bars for stems ≥ 4.8 cm dbh; open bars
for stems ≥ 10 cm dbh.
Among the species ≥ 4.8 cm dbh that appeared at
more than two sites on the same substrate, two species
(Dacrydium gibbsiae at 2700–3100 m and Syzygium
bankense (Hassk.) Merr. & Perry at 700–1700 m) were
restricted to ultrabasic substrates (pattern A), while
11 species to non-ultrabasic substrates (pattern B).
Among the 26 species that were common to both sub-
had the upper limit of the distributions at the higher al-
titudes on non-ultrabasic than on ultrabasic substrates
and/or had the lower limit of distributions at lower al-
titudes on ultrabasic than on non-ultrabasic substrates
(pattern C). By contrast, only four species had the
upper limit at higher altitudes on ultrabasic than on
non-ultrabasic substrates and/or had the lower limit
at lower altitudes on non-ultrabasic than on ultrabasic
substrates (pattern D). This suggests that lower plots
on ultrabasic substrates provide similar habitats as the
higher plots on non-ultrabasic substrates.
Species with low stem density that do not appear
within a plot may be recorded when the plot area is ex-
tended. To remove the effects of such sampling errors
for rare species, we also examined distributional pat-
terns of species after discarding the data of rare species
at each plot with varying cut-off threshold of stem
density (ha−1 ). The results suggest that the contrast
between the pattern C and D is free from sampling
errors. For example, two and 11 species showed the
distributional pattern A and C respectively after dis-
carding the data of stem density < 50 ha−1 , while no
species showed the pattern B and D.
Tree species diversity
A total of 380 species were recorded among the 2614
stems ≥ 10 cm dbh (452 species among the 5077 stems
≥ 4.8 cm dbh) in the eight plots. Both species-area and
150
species-stem number curves suggested that species
number had reached a plateau at 2700–3100 m on non-
ultrabasic substrates and at 1700–3100 m on ultrabasic
substrates (Figure 7). Although more species will be
found if the plot area is expanded, the overall trend in
species diversity across the eight sites was represented
by the present plot sizes. Fisher’s α and Shannon–
Wiener indices were correlated with species number
per plot among the eight sites (Table 6, Spearman rank
correlation, r = 0.95 and r = 0.99, respectively, for
stems ≥ 10 cm dbh, both p < 0.001). Species diver-
sity decreased with altitude on both substrate series.
When the two substrate types were compared, species
diversity was similar at 700 m, lower at the ultraba-
sic sites at the higher altitude, with the exception at
2700 m that corresponds to a cloud belt. Pielou’s index
of evenness showed a correlated pattern across sites
as the two diversity indices (Spearman rank correla-
tion, p < 0.05). Numbers of genera and families
per plot showed similar trends across sites as species
number per plot. Mean number of lower taxa per
higher taxon decreased with altitude (Table 6). This
was related to the altitudinal reduction in the number
of multi-specific genera (Table 7).
We examined correlations between tree species
diversity and forest structure among the eight sites ir-
respective of their substrate types for trees ≥ 10 cm
dbh. Species diversity indices (both Fisher’s α and
Shannon-Wiener) were positively correlated with
maximum height and biomass (Spearman rank cor-
relation, p < 0.05), but were not correlated with
basal area and stem density among the eight sites
(p > 0.1).
Discussion
Structural characteristics and species diversity of the
altitudinal sequence of rain forests on both non-
ultrabasic and ultrabasic substrates of Mount Kinabalu
agreed with the general trends for tropical mountains
of the world (Grubb 1977; Whitmore 1984; Gentry
1988). There were correlated changes among species
diversity, leaf size, forest stature and biomass, all de-
creasing with altitude. However, these changes were
accelerated on the ultrabasic substrate. We conclude
that the altitudinal sequence of forests on ultraba-
sic substrates was compressed compared to that on
non-ultrabasic substrates. This compression is simi-
lar to the telescoping of the altitudinal zonation of
vegetation on smaller mountains compared to larger
mountains (‘Massenerhebung’ effect, Grubb 1971).
The causes of these altitudinal changes remain to be
determined. That the magnitude of the changes was
greater on ultrabasic than on non-ultrabasic substrates
suggests that decreasing temperature was not the sole
cause. Thus temperature and substrate properties had
interactive effects on forests at the higher altitudes.
Nutrient deficiency is one of the possible reasons
for stunting on the ultrabasic substrate (Proctor &
Woodell 1975; Bruijnzeel et al. 1993) and at high al-
titudes (Grubb 1977; Tanner 1977; Marrs et al. 1988).
Indeed, soil analyses suggested that in-situ nitrogen
availability decreased with altitude on both substrate
series of Mount Kinabalu, and the availability was
always lower on ultrabasic than on non-ultrabasic sub-
strates except the cloud zone (Kitayama et al. 1998).
However, nitrogen alone cannot explain the greater
vegetation changes on the ultrabasic than on non-
ultrabasic substrates because the in-situ nitrogen avail-
ability more abruptly declined on the non-ultrabasic
series. Besides nitrogen, the availability of soil phos-
phorus is consistently lower on the ultrabasic series
above 1700 m and thus may limit the vegetation.
Another problem for plants alongside lack of
warmth and nutrient deficiency may be water de-
ficiency. Limited hydraulic conductance of stems
(Cavelier 1996) and occasional droughts (Kitayama
1996a) are known to occur at high altitudes, although
climate data in Table 1 do not show direct evidence
of drought and we do not have data on soil water
potential. Ultrabasic soils generally have high water
permeability, which can cause water deficiency during
droughts (Proctor & Woodell 1975). The soils on ul-
trabasic substrates of Mount Kinabalu contained less
organic carbon and were probably better drained than
those on non-ultrabasic substrates. Thus, water defi-
ciency may also explain the stunting of the forest on
ultrabasic substrate compared with non-ultrabasic sub-
strate. We observed that the highest shoots of trees
in stunted forests were often dead, presumably due
to desiccation. Such damage to meristems should pre-
vent growth of trees. Frequent sprouting in the stunted
forests may partly be the response to the damage.
Smaller leaf size at higher altitudes may also reflect
limited water supply (Cavelier 1996; but see Buckley
et al. 1980).
Forest structure and some soil chemical properties
are similar between the two substrate series at 700 m.
We speculate that a faster nutrient cycling due to the
warmer climate may in part compensate inherent geo-
chemical differences between the substrates at 700 m.
 151
Table 7. Number of species (number of restricted species that are absent from other plots in parentheses)
≥ 4.8 cm dbh in the genera that include 8 or more congeneric species in the eight study plots.

Genera 07N 07U 17N 17U 27N 27U 31N 31U Total

Elaeocarpus 1 (1) – 4 (4) – 2 (2) – 1 (1) – 8

Garcinia 1 (1) 6 (6) 5 (4) 1 (0) – – – – 11
Lithocarpus 6 (2) 2 (0) 12 (8) 1 (1) 1 (0) – – – 16
Litsea 7 (4) 3 (1) 6 (5) – 1 (1) – – – 14
Shorea 4 (1) 6 (3) 1 (1) – – – – – 8
Syzygium 11 (10) 15 (13) 15 (14) 5 (1) 2 (1) 2 (0) 3 (0) 2 (0) 47
Xanthophyllum 4 (2) 10 (8) – 1 (1) – – – – 13

Figure 7. Species accumulation curves for stems ≥ 10 cm dbh in the eight plots. (a) Species-area curves; (b) species-stem number curves. Each
curve follows the order of enumeration. See Table 1 for abbreviations of the plots.

However, it appears that the inherent geochemical
differences are more directly reflected in the vegeta-
tion as air temperature decreases and consequently the
vegetation becomes more contrasting between the sub-
strates. Additionally, the presence of species adapted
to nutrient and/or water shortage may explain the large
stature of the lowland ultrabasic vegetation. The ultra-
basic site at 700 m showed predominance of Shorea
laevis, which has small leaves and characterises low-
land vegetation on skeletal or dry soils on ridges
(Ashton 1982). It also contained more smaller-leaved
species than the non-ultrabasic site, as indicated by
smaller modal leaf area, and was characterised by
large individuals of conifers Agathis borneensis and
Dacrycarpus sp. although the latter was not enumer-
ated within the plot. These small-leaved species may
be able to grow efficiently on ultrabasic soils, because
the small leaf area could be related to adaptation to
water and nutrient shortage (Coomes & Grubb 1996).
This adaptation may apply also to upland sites in
general.
Inspite of the well-known term ‘serpentine en-
demism (Proctor & Woodell 1975)’, poor taxonomic
endemism of tree species on ultrabasic rocks has been
reported for Malaya and Solomon Islands (Whitmore
1984), and for Mount Silam, Sabah, ca. 200 km south
east of Mount Kinabalu (Proctor et al. 1988). Mei-
jer (1965) has noted only one endemic tree species
(Dacrydium gibbsiae) on ultrabasic soils of Kinabalu.
On the basis of a more complete database of the
Kinabalu flora, however, Beaman & Beaman (1990)
attributed high species diversity of Kinabalu’s vascular
flora including all life forms partly to the endemism on
ultrabasic soils. In the present study we did not find ev-
152
idence for higher endemism of tree species on ultraba-
sic than on non-ultrabasic substrates. Rather, on both
substrates, most species had very restricted distribu-
tions, and this pattern was also found in multi-specific
genera.
The lowland forests at 700 m on Kinabalu had rel-
atively high species diversity in comparison with other
lowland forests in North Borneo (Nicholson 1965;
Fox 1973; Newbery et al. 1996). The lower mon-
tane forest at 1700 m on non-ultrabasic rock of Mount
Kinabalu harbours one of the richest species diversity
among similar montane forests in the world (Yamada
1977; Gentry 1988; Proctor et al. 1988; Nadkarni
et al. 1995; Lieberman et al. 1996). These species-
rich forests on Kinabalu contain many multi-specific
genera (Table 7). By contrast, Nadkarni et al. (1995)
found at most only five species per genus within a
4-ha plot at Monteverde, Costa Rica, another species-
rich montane forest. We hypothesize that high tree
species diversity on Mount Kinabalu is owing to the
presence of multi-specific genera. The steep altitu-
dinal gradient and the heterogeneous distribution of
the different substrates on Mount Kinabalu should
have facilitated species diversification and provided
refugia for many rare species. The rich flora of Male-
sia should serve as the prerequisite for such species
diversification (Kitayama 1996b).
On Kinabalu, basal area was not correlated with
biomass and maximum tree height, which in turn were
correlated each other. We interpreted this as reflect-
ing the fact that tree allometry changes from slender
trunks to robust trunks with increasing altitude and
from non-ultrabasic to ultrabasic substrates. Probably
disproportionately greater biomass is allocated to stem
diameter growth than to height growth with increas-
ing altitude. This change across altitude seems to be
the general trend on tropical mountains (Edwards &
Grubb 1977; Tanner 1980; Ash 1987). Aboveground
net primary productivity decreases with increasing al-
titude on tropical mountains (Weaver & Murphy 1990;
Raich et al. 1997). Short-term records (1.7 to 1.9
years) of litterfall and tree growth on Mount Kinabalu
also indicate that aboveground net primary productiv-
ity decreases with altitude and is lower on ultrabasic
than on non-ultrabasic substrates (K. Kitayama &
S. Aiba, unpublished data). Thus, the allocation of
biomass seems to coincide with primary productiv-
ity: the forests with higher primary productivity are
characterized by greater biomass allocation to height
growth than those with lower primary productivity.
Tree species diversity decreases with increasing alti-
tude, and therefore decreases with decreasing primary
productivity. We suggest that high allocation of bio-
mass to height growth under productive environment
may be related to high species diversity. Duivenvo-
orden (1996) found a positive correlation between
canopy height and tree species diversity in Amazonian
lowland rain forests. Further studies are necessary to
clarify causes and effects behind this correlation.
Acknowledgements
This study is a contribution to the TEMA, a core
research of IGBP-GCTE. We thank the following per-
sons: Datuk L. Ali, F. Liew, J. Nais and R. Repin of
the Sabah Parks for support in every aspect of our
study; K. M. Wong, J. B. Sugau, J. T. Pereira and
the staffs of the Herbarium, Forest Research Cen-
tre, Sabah Forestry Department, for the assistance in
specimen identification; and P. Akau, K. Kimura, T.
Shumiya and many other persons who assisted in the
fieldwork. We also thank P. S. Ashton, D. F. R. P.
Burslem, K. Homma, T. Kohyama and two anony-
mous reviewers for comments on the manuscript. This
work was supported by a grant from the Environ-
ment Agency of Japan to K.K. and supplemented by
a Research Fellowship of the Japan Society for the
Promotion of Science for Young Scientists to S.A.
References
Aiba, S. & Kohyama, T. 1997. Crown architecture and life-history
traits of 14 tree species in a warm-temperate rain forest: signifi-
cance of spatial heterogeneity. J. Ecol. 85: 611–624.
Aplet, G. & Vitousek, P. M. 1994. An age-altitude matrix analysis
of Hawaiian rain-forest succession. J. Ecol. 82: 137–147.
Ash, J. 1987. Stunted cloud-forest in Taveuni, Fiji. Pac. Sci. 41:
191–199.
Ashton, P. S. 1982. Dipterocarpaceae. Flora Malesiana Ser. I 9: 237–
552.
Austin, M. P., Ashton, P. S. & Greig-Smith, P. 1972. The ap-
plication of quantitative methods to vegetation survey. III. A
re-examination of rain forest data from Brunei. J. Ecol. 60:
308–324.
Beaman, J. H. & Beaman, R. S. 1990. Diversity and distribution pat-
terns in the flora of Mount Kinabalu. Pp. 147–160. In: Baas, P.,
Kalkman, K. & Geesink, R. (eds), The plant diversity of Malesia.
Kluwer Academic Publishers, Dordrecht.
Bruijnzeel, L. A., Waterloo, M. J., Proctor, J., Kuiters, A. T.
& Kotterink, B. 1993. Hydrological observations in montane
rain forests on Gunung Silam, Sabah, Malaysia, with special
reference to the ‘Massenerhebung’ effect. J. Ecol. 81: 145–167.
Buckley, R. C., Corlett, R. T. & Grubb, P. J. 1980. Are the xero-
morphic trees of tropical upper montane rain forests drought-
resistant? Biotropica 12: 124– 136.

Cavelier, J. 1996. Environmental factors and ecophysiological
processes along altitudinal gradients in wet tropical mountains.
Pp. 399–439. In: Mulkey, S. S., Chazdon, R. L. & Smith, A.
P. (eds), Tropical forest plant ecophysiology. Chapman & Hall,
New York.
Collenette, P. 1964. A short account of the geology and geological
history of Mt Kinabalu. Pro. Roy. Soc. B 161: 56–63.
Coomes, D. A. & Grubb, P. J. 1996. Amazonian caatinga and re-
lated communities at La Esmeralda, Venezuela: forest structure,
physiognomy and floristics, and control by soil factors. Vegetatio
122: 167–191.
Duivenvoorden, J. F. 1996. Patterns of tree species richness in rain
forests of the middle Caquetá area, Colombia, NW Amazonia.
Biotropica 28: 142–158.
Edwards, P. J. & Grubb, P. J. 1977. Studies of mineral cycling in a
montane rain forest in New Guinea. I. The distribution of organic
matter in the vegetation and soil. J. Ecol. 65: 943–969.
Fox, J. E. D. 1973. A handbook to Kabili-Sepilok Forest Reserve.
Sabah forest record no. 9. Borneo Literature Bureau, Kuchin.
Gentry, A. H. 1988. Changes in plant community diversity and
floristic composition on environmental and geographic gradients.
Ann. Mo. Bot. Gard. 75: 1–34.
Grubb, P. J. 1971. Interpretation of the ‘Massenerhebung effect’ on
tropical mountains. Nature 229: 44–45.
Grubb, P. J. 1977. Control of forest growth and distribution on wet
tropical mountains: with special reference to mineral nutrition.
Ann. Rev. Ecol. Syst. 8: 83–107.
Jacobson, G. 1978. Geology. Pp. 101–110. In: Kinabalu, summit of
Borneo. The Sabah Society, Kota Kinabalu.
Kira, T. & Shidei, T. 1967. Primary production and turnover of or-
ganic matter in different forest ecosystems of the Western Pacific.
Jap. J. Ecol. 17: 70–87.
Kitayama, K. 1991. Vegetation of Mount Kinabalu Park, Sabah,
Malaysia. Project paper. East-West Center, Honolulu.
Kitayama, K. 1992. An altitudinal transect study of the vegetation
on Mount Kinabalu, Borneo. Vegetatio 102: 149–171.
Kitayama, K. 1995a. Biophysical conditions of the montane cloud
forests of Mount Kinabalu, Sabah, Malaysia. Pp. 183–197. In:
Hamilton, L. S., Juvik, J. O. & Scatena, F. N. (eds), Tropical
montane cloud forests. Springer-Verlag, New York.
Kitayama, K. 1996a. Climate of the summit region of Mount Kin-
abalu (Borneo) in 1992, an El Niño year. Mountain Res. Devel.
16: 65–75.
Kitayama, K. 1996b. Patterns of species diversity on an oceanic
versus a continental island mountain: a hypothesis on species
diversification. J. Veg. Sci. 7: 879–888.
Kitayama, K., Aiba, S., Majalap-Lee, N. & Ohsawa, M. 1998. Soil
nitrogen meneralization rates of rain forests in a matrix of ele-
vations and geological substrates on Mount Kinabalu, Borneo.
Ecol. Res. 13, in press.
Lieberman, D., Lieberman, M., Peralta, R. & Hartshorn, G.
1996. Tropical forest structure and composition on a large-scale
altitudinal gradient in Costa Rica. J. Ecol. 84: 137–152.
Marrs, R. H., Proctor, J., Heaney, A. & Mountfield, M. D. 1988.
Changes in soils, nitrogen mineralization and nitrification along
an altitudinal transect in tropical rain forest in Costa Rica. J. Ecol.
76: 466–482.
Meijer, W. 1965. A botanical guide to the flora of Mt. Kinabalu. Pp.
325–366. In: Symposium on ecological research in humid tropics
vegetation. Government of Sarawak and UNESCO, Kuchin.
Nadkarni, N. M., Matelson, T. J. & Haber, W. A. 1995. Structural
characteristics and floristic composition of a Neotropical cloud
forest, Monteverde, Costa Rica. J. Trop. Ecol. 11: 481– 495.
153
Nagano, M. 1978. Dynamics of stand development. Pp. 21–32. In:
Kira, T., Ono, Y. & Hosokawa, T. (eds), Biological production in
a warm-temperate evergreen oak forest of Japan. JIBP synthesis
vol. 18. University of Tokyo Press, Tokyo.
Nakashizuka, T., Yusop, Z. & Nik, A. R. 1991. Altitudinal zonation
of forest communities in Selangor, Peninsular Malaysia. J. Trop.
For. Sci. 4: 233–244.
Newbery, D. McC., Campbell, E. J. F., Proctor, J. & Still, M.
J. 1996. Primary lowland dipterocarp forest at Danum Valley,
Sabah, Malaysia. Species composition and patterns in under-
storey. Vegetatio 122: 193–220.
Nicholson, D. I. 1965. A study of virgin forest near Sandakan, North
Borneo. Pp. 67–87. In: Symposium on ecological research in hu-
mid tropics vegetation. Government of Sarawak and UNESCO,
Kuchin.
Ogawa, H. 1969. An attempt at classifying forest types based on
the relationship between tree height and dbh. Pp. 3–17. In: Kira,
T. (ed.), Comparative study of primary productivity in forest
ecosystems. JIBP-PT-F progress reports for 1968 (in Japanese).
Ohsawa, M. 1993. Latitudinal pattern of mountain vegetation zona-
tion in southern and eastern Asia. J. Veg. Sci. 4: 13–18.
Ohsawa, M., Nainggolan, P. H. J., Tanaka, N. & Anwar, C.
1985. Altitudinal zonation of forest vegetation on Mount Kerinci,
Sumatra: with comparison to zonation in the temperate region of
east Asia. J. Trop. Ecol. 1: 193–216.
Oksanen, J. & Minchin, P. R. 1997. Instability of ordination results
under changes in input data order: explanations and remedies. J.
Veg. Sci. 8: 447–454.
Pendry, C. A. & Proctor, J. 1996. The causes of altitudinal zonation
of rain forests on Bukit Belalong, Brunei. J. Ecol. 84: 407–418.
Pielou, E. C. 1975. Ecological diversity. Wiley-Interscience, New
York.
Proctor, J., Anderson, J. M., Chai, P. & Vallack, H. W. 1983. Eco-
logical studies in four contrasting lowland rain forests in Gunung
Mulu National Park, Sarawak. I. Forest environment, structure
and floristics. J. Ecol. 71: 237–260.
Proctor, J., Lee, Y. F., Langley, A. M., Munro, W. R. C. & Nelson,
T. 1988. Ecological studies on Gunung Silam, a small ultrabasic
mountain in Sabah, Malaysia. I. Environment, forest structure
and floristics. J. Ecol. 76: 320–340.
Proctor, J. & Woodell, S. R. J. 1975. The ecology of serpentine soils.
Adv. Ecol. Res. 9: 255–366.
Raich, J. W., Russell, A. E. & Vitousek, P. M. 1997. Primary produc-
tivity and ecosystem development along an elevational gradient
on Mauna Loa, Hawaii. Ecology 78: 707–721.
Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:
223–225.
Roberts, B. A. & Proctor, J. 1992. The ecology of areas with ser-
pentinized rocks: a world view. Kluwer Academic Publishers,
Dordrecht.
Tanner, E. V. J. 1977. Four montane rain forests of Jamaica: a quan-
titative characterization of the floristics, the soils and the foliar
mineral levels, and a discussion of the interrelations. J. Ecol. 65:
883–918.
Tanner, E. V. J. 1980. Studies on the biomass and productivity in a
series of montane rain forests in Jamaica. J. Ecol. 68: 573–588.
ter Braak, C. J. F. 1988. CANOCO – a FORTRAN program for
canonical community ordination by [partial] [detrended] [canon-
ical] correlation analysis, principal components analysis and
redundancy analysis (version 2.1). TNO Institute of Applied
Computer Science, Wageningen.
Tracey, J. G. 1982. The vegetation of the humid tropical region of
North Queensland. CSIRO, Melbourne.
154
Weaver, P. L. & Murphy, P. G. 1990. Forest structure and pro-
ductivity in Puerto Rico’s Luiquillo mountains. Biotropica 22:
69–82.
Webb, L. J. 1959. A physiognomic classification of Australian rain
forests. J. Ecol. 47: 551–570.
Whitmore, T. C. 1984. Tropical rain forests of the Far East. 2nd ed.
Clarendon Press, Oxford.
Wilson, J. B., Lee, W. G. & Mark, A. F. 1990. Species diversity
in relation to ultramafic substrate and to altitude in southwestern
New Zealand. Vegetatio 86: 15–20.
Yamada, I. 1977. Forest ecological studies of the montane forest of
Mt. Pangrango, West Java. IV. Floristic composition along the
altitude. South East Asian Studies 15: 226–254.
Yamakura, T., Hagihara, A., Sukardjo, S. & Ogawa, H. 1986.
Aboveground biomass of tropical rain forest stands in Indonesian
Borneo. Vegetatio 68: 71–82.
Yamanaka, T. 1959. A phytosociological study of serpentine areas
in Shikoku, Japan. Res. Rep. Kochi Univ. 8: 1–47.
 155
Appendix 1. Relative basal area (%) of species ≥ 10 cm d.b.h. that were among the 15 most abundant at any of the eight plots.
Species are arranged by their scores on axis 1 in DCA, and by relative basal area for species with same score.

Species 07N 07U 17N 17U 27N 27U 31N 31U DCA scores

Shorea laevis Ridl. – 30.60 – – – – – – −0.34

Dacryodes costata (A.W. Benn.) – 4.16 – – – – – – −0.34
H.J. Lim
Agathis borneensis Warb. – 3.50 – – – – – – −0.34
Coelostegia griffithii Benth. – 2.87 – – – – – – −0.34
Heritiera simplifolia (Mast.) Kosterm. – 2.65 – – – – – – −0.34
Koompassia malaccensis (Maing.) – 2.57 – – – – – – −0.34
Benth.
Palaquium herveyi King & Gamble – 2.28 – – – – – – −0.34
Dysoxylum excelsum Bl. – 2.02 – – – – – – −0.34
Syzygium cf. tawahense Gaertner – 1.93 – – – – – – −0.34
Shorea gibbosa Brandis – 1.88 – – – – – – −0.34
Trigoniastrum hypoleum Miq. – 1.34 – – – – – – −0.34
Scaphium macropodum (Miq.) 0.04 1.65 – – – – – – −0.21
Beum.
Cyathocalyx magnificus Diels. 0.11 1.49 – – – – – – 0.10
Aquillaria malaccensis Lamb. 0.94 3.37 – – – – – – 1.01
Dialium indum L. 1.80 0.97 – – – – – – 2.96
Shorea parvistipulata Heim 4.54 2.14 – – – – – – 3.04
Shorea leprosula Miq. 7.31 0.36 – – – – – – 3.74
Hopea sp. 1 Roxb. 4.38 0.06 – – – – – – 3.82
Ixonanthes reticulata Jack 8.00 – – – – – – – 3.85
Shorea argentifolia Sym. 6.40 – – – – – – – 3.85
Castanopsis cf. javanica Spach 4.34 – – – – – – – 3.85
Norrisia major Soler. 3.50 – – – – – – – 3.85
Sandoricum koetjape (Burm. f.) Merr. 3.12 – – – – – – – 3.85
Tristaniopsis sp. 1 Brong. & Gris 2.99 – – – – – – – 3.85
Litsea cf. brachystachya Lam. 2.76 – – – – – – – 3.85
Cinnamomum parthenoxylon Meissner 2.36 – – – – – – – 3.85
Maglietia dolichogyna Dandy ex Noot. 2.23 – – – – – – – 3.85
Syzygium fastigiatum (Bl.) Merr. & 2.07 – – – – – – – 3.85
Perry
Payena microphylla (De Vriese) Pierre 6.69 0.09 7.01 – – – – – 6.49
Aglaia squamulosa King 0.11 – 2.60 – – – – – 7.47
Vernonia arborea Buch.-Ham. 0.38 – 0.63 – 0.39 – – – 7.51
Syzygium cf. pachysepalum Gaertner – – 8.22 – – – – – 7.57
Tristaniopsis sp. 2 Brong. & Gris – – 7.62 – – – – – 7.57
Tristaniopsis ‘clementis’ Brong. & – – 5.97 – – – – – 7.57
Gris
156
Appendix 1. continued.

Species 07N 07U 17N 17U 27N 27U 31N 31U DCA scores

Lithocarpus clementianus – – 4.35 – – – – – 7.57

(King ex Hook. f.) A. Camus
Syzygium chrysanthum Merr. & Perry – – 3.86 – – – – – 7.57
Syzygium napiforme (Koord. & Val.) – – 3.31 – – – – – 7.57
Merr. & Perry
Garcinia ‘ramiflora’ L. – – 2.71 – – – – – 7.57
Lithocarpus hatusimae Soepadmo – – 2.71 – – – – – 7.57
Palaquium gutta (Hook. f.) Baillon – – 1.92 – – – – – 7.57
Dacrycarpus imbricatus (Bl.) de Laub. – – 4.58 0.12 – – – – 7.59
Dacrydium pectinatum de Laub. – – 8.42 4.01 – – – – 7.73
Gynotroches axillaris Bl. – – 1.73 2.15 – – – – 7.85
Garcinia multinervia Meijer – – 1.43 2.58 – – – – 7.90
Eugenia subdecussata Duthie – – 0.49 2.28 – – – – 8.01
Tetractomia tetrandrum (Roxb.) Merr. 0.06 – – 3.04 – – – – 8.05
Tristaniopsis cf. ‘elliptica’ Brong. & – – – 31.41 – – – – 8.12
Gris
Agathis kinabaluensis de Laub. – – – 19.15 – – – – 8.12
Xanthophyllum tenue Chodat – – – 6.88 – – – – 8.12
Podocarpus gibbsii N.E. Gray – – – 6.76 – – – – 8.12
Weinmannia cf. blumei L. – – – 3.55 – – – – 8.12
Quercus lowii King – – – 3.30 – – – – 8.12
Lithocarpus rigidus Soepadmo – – – 2.27 – – – – 8.12
Syzygium cf. castaneum Gaertner – – – 2.05 – – – – 8.12
Rapanea forbesii Mez. – – – 1.80 – – – – 8.12
Ilex oppositifolia Merr. – – – 1.63 – – – – 8.12
Myrica javanica Bl. – – 0.35 0.09 – 0.12 0.78 – 12.02
Phyllocladus hypophyllus Hook. f. – – 3.18 0.11 7.91 0.58 1.47 1.41 12.37
Lithocarpus havilandii (Stapf) Barnett – – 0.74 – 7.96 – – – 13.05
Daphniphyllum glaucescens Bl. – – 0.21 – – 0.60 – – 13.58
Magnolia carsonii Dandy ex Noot. – – 0.05 – 18.33 – – – 14.08
Syzygium cf. punctilimbum Gaertner – – – – 39.90 – – – 14.12
Olea decussata (Heine) Kiew – – – – 9.20 – – – 14.12
Clethra pachyphylla Merr. – – – – 2.58 – – – 14.12
Elaeocarpus sp. 1 L. – – – – 2.50 – – – 14.12
Prunus oocarpa (Stapf) Kalkm. – – – – 0.57 – – – 14.12
Rapanea avenis (Bl.) D.C. – – – – 0.29 – – – 14.12
Symplocos deflexa Stapf – – – – 0.16 – – – 14.12
Ilex zygophylla Merr. – – – – 8.07 0.47 5.93 – 15.43
Microtropis wallichiana – – – – 0.42 0.31 – – 15.49
Wight ex. Thwaites
Symplocos pendula Wight – – – – 1.60 1.25 2.48 – 16.27
Schima brevifolia (Hook. f.) Stapf – – – 1.16 – 7.42 10.01 – 16.35
Syzygium kinabaluensis (Stapf) – – – – 0.12 – 6.80 – 17.04
Merr. & Perry
 157
Appendix 1. continued.

Species 07N 07U 17N 17U 27N 27U 31N 31U DCA scores

Ternstroemia lowii Stapf – – – – – – 4.19 – 17.09

Symplocos buxifolia Stapf – – – – – – 3.93 – 17.09
Syzygium houttuynii Merr. & Perry – – – – – 0.90 10.35 – 17.12
Polyosma hookeri Stapf – – – – – 2.80 5.07 – 17.19
Magnolia persuaveolens Dandy – – – – – 1.64 2.31 – 17.21
Dacrycarpus kinabaluensis (Wasscher) – – – – – 18.95 36.25 7.92 17.27
de Laub.
Syzygium steenisii Merr. & Perry – – – – – 4.62 2.28 – 17.28
Tristaniopsis ‘elliptica’ Brong. & Gris – – – – – 6.75 – – 17.37
Weinmannia clemensiae van Steenis – – – – – 0.24 – – 17.37
Dacrydium gibbsiae Stapf – – – – – 32.45 – 2.64 17.41
Leptospermum recurvum Hook. f. – – – – – 21.50 6.85 88.03 17.78
Other species 35.87 34.06 28.13 5.45 0 0 0.69 0 –