Professional Documents
Culture Documents
tmpDA91 TMP
tmpDA91 TMP
Abstract
Climate change scenarios predict increases in the frequency and duration of ENSO-related droughts for parts of
South-East Asia until the end of this century exposing the remaining rainforests to increasing drought risk. A
pan-tropical review of recorded drought-related tree mortalities in more than 100 monitoring plots before, during
and after drought events suggested a higher drought-vulnerability of trees in South-East Asian than in Amazonian
forests. Here, we present the results of a replicated (n = 3 plots) throughfall exclusion experiment in a perhumid
tropical rainforest in Sulawesi, Indonesia. In this first large-scale roof experiment outside semihumid eastern
Amazonia, 60% of the throughfall was displaced during the first 8 months and 80% during the subsequent
17 months, exposing the forest to severe soil desiccation for about 17 months. In the experiments second year, wood
production decreased on average by 40% with largely different responses of the tree families (ranging from 100 to
+100% change). Most sensitive were trees with high radial growth rates under moist conditions. In contrast, tree
height was only a secondary factor and wood specific gravity had no influence on growth sensitivity. Fine root
biomass was reduced by 35% after 25 months of soil desiccation while fine root necromass increased by 250%
indicating elevated fine root mortality. Cumulative aboveground litter production was not significantly reduced in
this period. The trees from this Indonesian perhumid rainforest revealed similar responses of wood and litter
production and root dynamics as those in two semihumid Amazonian forests subjected to experimental drought. We
conclude that trees from paleo- or neotropical forests growing in semihumid or perhumid climates may not differ
systematically in their growth sensitivity and vitality under sublethal drought stress. Drought vulnerability may
depend more on stem cambial activity in moist periods than on tree height or wood specific gravity.
Keywords: drought sensitivity of growth, fine root mortality, Paleotropis, perhumid climate, roof experiment, soil desiccation,
tree mortality
Received 20 June 2013 and accepted 28 August 2013
Introduction
Climate warming in the Tropics may lead to more
variable climates with larger extremes in the decades
and centuries to come (Dai, 2013). For most tropical
regions, a temperature increase by 24 K or more has
been predicted for the next hundred years (Christensen
et al., 2007). The effects of increasing temperatures may
be linked to an increasing evaporative demand of the
atmosphere, increases or decreases in precipitation,
changes of radiative forcing and regional wind systems.
The multitudes of factors that change in concert make
predictions about the response of natural and managed
tropical ecosystems difficult.
Correspondence: Christoph Leuschner, tel: +49 551 39 5718,
fax: +49 551 39 5701, e-mail: cleusch@gwdg.de
1482 G . M O S E R et al.
of El Ni~
no-Southern Oscillation (ENSO) events should
continuously increase (Latif et al., 1998; Guilyardi, 2006;
Hansen et al., 2006; Merryfield, 2006; Philip & van
Oldenborgh, 2006; Christensen et al., 2007; Rosenzweig
et al., 2007; Paeth et al., 2008; Sheffield & Wood, 2008).
Interannual rainfall variation in South-East Asia is largely influenced by the ENSO system (Sheffield &
Wood, 2008) and severe droughts as in 1997/98 occur
most often in association with extreme ENSO events.
For example, on Mount Kinabalu (Borneo), monthly
rainfall was reduced in the 1998 drought to <60 and
<10 mm at two elevations with a rainless period lasting
for more than 90 d (Aiba & Kitayama, 2002). IPCC
scenarios predict for SE-Asia a drought of 46 months
to happen all 15 years and forcast a high probability of
a 12-month drought once in a century (Sheffield &
Wood, 2008). Drought may thus represent a major stressor for tropical forests not only in semihumid but also
in humid and per-humid climates.
During the last 10 years, the response of tropical
trees and forests to extended drought periods has
received increasing attention (Marengo et al., 2008;
Phillips et al., 2009, 2010; Anderson et al., 2010; Lewis
et al., 2011). Observational studies on drought effects
have been conducted in the tropical forests of the
Americas, Africa and South-East Asia (Aiba & Kitayama, 2002; Slik, 2004; van Nieuwstadt & Sheil, 2005;
Nishimura et al., 2007; Allen et al., 2010; Meir & Woodward, 2010; Phillips et al., 2010; Itoh et al., 2012). The
consequences of the severe ENSO drought of 1997/98
for tree mortality rate and the impact of subsequent forest fires have been intensively studied in particular in
Borneo and Sumatra (Siegert et al., 2001; van Nieuwstadt & Sheil, 2005; Fredriksson et al., 2007). One finding
is that trees of different size and position in the canopy
are differently susceptible to water shortage. For example, Aiba & Kitayama (2002) found increased mortality
of small- to medium-diameter trees together with
reductions in stem diameter increment of large trees
(DBH >100 cm) in stands at 700 m elevation on Mt. Kinabalu, Borneo, while others reported the highest mortality rates in large-diameter trees (Slik, 2004; van
Nieuwstadt & Sheil, 2005; Itoh et al., 2012). A pan-tropical survey indicated a particularly high droughtinduced mortality of large trees and, to a lesser extent,
of trees with low wood density (Phillips et al., 2010).
A mechanistic understanding of how drought may
affect tropical forests can only be achieved by rainfall
manipulation experiments with controlled water
supply. Wu et al. (2011) and Beier et al. (2012) provide
recent reviews of the results of precipitation
manipulation experiments, and Smith et al. (2009) offer
a conceptual framework for understanding the impacts
of altered resource supply on ecosystem dynamics.
I N D O N E S I A N R A I N F O R E S T D R O U G H T E X P E R I M E N T 1483
in Borneo than in Amazonia, and differences in edaphic
factors, rooting depth and biogeography complicate
direct comparison.
To close this knowledge gap, we set up a replicated
throughfall exclusion experiment with three subcanopy
roofs of 1600 m2 area each in a perhumid, pre-montane
tropical rainforest in Sulawesi, Indonesia (Sulawesi
Throughfall Exclusion Experiment, STEE) for studying
drought responses at the tree and ecosystem levels in a
tropical forest, which is not exposed to longer regular
dry periods. The desiccation started in 2007 and was
maintained for 25 months until summer 2009. This
article presents results on the trees drought response
with a focus on aboveground (wood growth and litter
production) and belowground productivity and carbon
cycling (fine root production and mortality). We
wanted to answer three questions, first, how variable is
the drought response among the trees of a species-rich
tropical perhumid forest; second, what is the role of
species identity (or phylogeny) in comparison to tree
morphology (height and wood specific gravity) as a
determinant of drought sensitivity, and third, which
productivity component (production of wood, leaves or
fine roots) is the most drought-sensitive. We tested the
hypotheses (i) that the trees of this aseasonal perhumid
rainforest are on average more sensitive to drought
than the trees of the two existing Amazonian throughfall reduction experiments, where extended dry periods
are regularly occurring, and that (ii) the trees drought
sensitivity increases with tree size but decreases with
increasing wood specific gravity.
Study site
The Sulawesi Throughfall Exclusion Experiment was
established in 2006 in a natural pre-montane rainforest in
Pono Valley, about 3 km NE of the village of Toro in Lore Lindu National Park in Central Sulawesi, Indonesia (0129.6S,
12003.4E; see map in Fig. 1). Six study plots of 40 m 9 40 m
size on level terrain at 1050 m elevation were selected in close
proximity to each other in a forest section with no signs of
human disturbance (as evidenced by the absence of stumps
and charcoal in the soil, and the presence of rattan palms that
are extracted first). The plots were spread in a stratified
random design over an area of about 5 ha on a small plateau
in overall rugged terrain to guarantee comparability of site
conditions. The maximum distance between the first and the
last plot was ca. 250 m. The climate of the study area is perhumid with a long-term mean annual rainfall of 1894 mm and a
mean annual temperature of 21.3 C (WorldClim, 2006). During the study period, the mean annual rainfall was somewhat
higher with 2672 mm (3003 mm in the first and 2222 mm in
the second year) and a mean annual temperature of 20.8 C
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
Experimental design
Three of the six plots with an area of 0.16 ha were selected by
random (# 2, 3 and 6) and equipped with subcanopy roofs
(40 m 9 40 m) to cover the whole plot (total roof area:
0.48 ha; Fig. 1). Frames made of bamboo poles of
5 m 9 0.5 m size were placed at 1.23.0 m height above the
ground on a wooden roof construction and covered by
transparent and UV-insensitive polyethylene sheet to collect
the throughfall and channel it out of the plots with the help of
gutters (Fig. 1). The captured throughfall water percolated
into the soil at a distance of more than 3 m beyond the plot
boundary trenching in downslope direction. The efficiency of
the 40 cm deep trenching was controlled continuously by
weekly TDR measurements of soil moisture conducted with
high spatial resolution. Accordingly, the plot edges were not
moister than the interior (see Fig. S1 in the Appendix). Around
all six plots (roof and control), the 40 cm deep trenches were
covered by plastic foil for inhibiting root growth into the plots
and uptake of water from outside. According to excavations,
the soil profile to 40 cm contained ca. 80% of the fine root
biomass in the profile (Hertel et al., 2009). Gaps in the roofs
adjacent to tree stems were closed with smaller frames that fitted into the openings. Stem flow was not excluded from the
plots due to the large number of stems covered by the roofs
(in total more than 200 with DBH >10 cm); stemflow in a
nearby forest stand equalled <1% of gross precipitation and
thus contributed only marginally to soil infiltration (Dietz
et al., 2006). All measurements on trees and soil were conducted in a central core zone (30 m 9 30 m) of each plot surrounded by a 5 m wide buffer zone to exclude edge effects.
1484 G . M O S E R et al.
Fig. 1 Map of Indonesia with the island of Sulawesi, in detail the Toro and Pono Valleys, the relative position of the six study plots
(40 m 9 40 m) and the meteorological tower in a patch of the Pono Valley forest with surface relief of the terrain and photographs of
the roof construction. The three desiccation plots (#2, 3 and 6) carry a below-canopy roof construction for throughfall displacement.
Solar panels were mounted on the 16 m high meteorological tower in a natural forest gap.
dry months (100 mm) and it might have reduced soil water
availability to a similar extent as the exceptional 12-month dry
periods predicted to occur once per century in future time
(Sheffield & Wood, 2008). We decided to extend the throughfall exclusion period to 17 months, that is, longer than
predicted by climate change scenarios, because we wished to
balance the mitigating effect of constantly high air humidity
during our experiment. Natural droughts are characterized by
the simultaneous occurrence of dry soil and dry air, a situation
that cannot be simulated in field experiments.
The fallen fine litter on the plastic-lined panels was
removed fortnightly and evenly spread out on the ground
below the roofs. We were not able to conduct a true pretreatment reference measuring phase prior to roof closure due to
the substantial logistic and technical problems (including a
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
169.1 23.0a
148.0 16.0
126.9 17.5a
76.5 10.3
67.1 7.1
4.8 0.3
325.4 43.7
4.8 0.2
13.4 0.4
13.9 4.2
870.8 20.4
866.7 46.4
4.7 0.2
4.5 0.2
20.9 3.8
20.9 0.3
0.12
1.13
5.9 0.1
5.8 0.1
6.6 0.2
6.3 0.3a
5.8 0.03
425
431.7 43.8a
396.0 23.3
456.5 9.3
435.0 5.6
0.09
6.6 0.2
0.1 0.02
0.13
5.8 0.1
0.29 0.41
0.11
5.8 0.1
6.1 2.9
413.8 23.6
Total
Mean SE
0.4 0.4
Control
Mean SE
1.1 0.6
Roof
Mean SE
1.8 1.1a
R
6
1.1
419
363
C
5
1.8
C
4
350
R
3
513
413
0
R
2
3.8
C
1
Trees/ha
0.14
height m
0.5 0.3a
m2/ha
Tree
m2/ha
Basal area
285.5 30.3
a
a
a
a
a
a
a
57.7 7.8a
245.6 33.1a
4.8 0.6a
14.3 1.0a
875.0 23.9a
4.3 0.4a
20.9 0.8a
477.3 43.8a
63.8
264.9
5.9
13.3 0.5
881
4.8
19.6 0.5
450
56.0
87.1
363.8
232.0
5.1
4.6
13.3 0.5
14.3 0.5
800
844
4.5
5.0
21.2 0.5
20.4 0.5
425
438
67.0
278.7
4.7
16.3 0.6
913
4.6
22.3 0.4
563
86.3
42.3
173.0
360.4
4.8
3.7
13.3 0.5
12.8 0.4
956
831
3.5
4.6
21.1 0.4
20.8 0.5
419
Mg/ha
m2/ha
Basal area
Tree height m
Trees/ha
m2/ha
Basal area
Tree height m
Trees/ha
444
Mg/ha
140.7
192.9
123.2
92.1
147.9
191.2
Mg/ha
Carbon pool
BGB
AGB
All Trees
Trees 10 cm DBH
Basal area
Roof/Control
#
Stranglers
Plot
Fig. 2 Climate diagram after Walter & Lieth (1967) with the
monthly means of air temperature (lower line) and precipitation
sums (upper line) for the period March 2007 to August 2009
recorded at a 16 m high tower in a forest gap close to the study
site. Hatched areas indicate humid periods, black areas perhumid conditions with monthly precipitation sums >100 mm
(compressed scale above 100 mm). Also given are elevation,
mean annual temperature and mean precipitation total, length
of observation period, and the coordinates of the station.
Table 1 Stand structure, above- (AGB) and belowground biomass (BGB) and total biomass carbon pool of the six forest plots in Pono Valley, Central Sulawesi, Indonesia
(means 1 SE for n = 3 plots each). AGB was estimated with allometric regressions and BGB derived from root : shoot ratios given in the literature (see Methods section). Tree
height is the mean of all trees in the respective DBH class, basal area of the plot total. Different letters indicate significant differences between control and roof plot means
(P < 0.05)
I N D O N E S I A N R A I N F O R E S T D R O U G H T E X P E R I M E N T 1485
1486 G . M O S E R et al.
calibration was done for the 10-20 and 40-75 cm soil layers
due to very homogenous soil texture). The hydrological and
physiological measurements began on March 27, 2007, in the
roof plots and on May 31, 2007, in the control plots (the delay
was caused by a lightning strike in March 2007).
The soil moisture measurements were used to calculate the
relative amount of extractable water (REW) in the profile
using the equation given by Brda et al. (2006) and Granier
et al. (2007). REW is calculated from daily data on the fractional volumetric soil water content, the soil water content at
which all plant-available water is assumed to be extracted
(corresponding to a matric potential of 1.5 MPa in the considered soil depth), and the maximum water content recorded
during the study period in that depth. The volumetric soil
water content at 1.5 MPa was obtained from laboratoryderived soil water retention curves (van Genuchten, 1980)
established with a pressure plate apparatus (van Straaten
et al., 2011). REW varies between 1 (field capacity or highest
soil water content recorded) and 0 (at 1.5 MPa). According
to Granier et al. (2007), temperate trees typically experience
drought stress when REW drops below a threshold value of
0.4. The soil water retention curves were further used for
deriving soil matric potential values from the continuous soil
moisture measurements conducted in these soil depths. The
spatial heterogeneity of soil moisture was monitored with 100
TDR measurements per plot conducted at weekly intervals in
a regular grid (8.75 m grid net width with 2.5 m distance to
the plot edge) with a mobile TDR probe by inserting the metal
bars from the forest floor vertically 30 cm into the soil.
trunk diameter (DBH in cm), H is total tree height (in m), and
q is wood specific gravity (in g cm3). The literature data on
wood density often refer to wood samples measured under
fresh weight conditions (i.e. at 12 or 15% moisture). For transforming these data to wood specific gravity on a dry weight
basis, we corrected the literature data for water content
following Chave et al. (2006). In cases where ranges of wood
density values were given, we used median values. In species
for which wood density data were not available, the average
across all species in that genus growing in the corresponding
habitat was used (see Baker et al., 2004; Slik, 2006). In the few
cases (n = 10) where a tree individual could not be identified
at least to the genus level or where a literature record was not
available, we used the plot average.
Two groups of tree species required a specific calculation
procedure for estimating AGB due to their specific growth
forms. (i) The abundant Fagaceae Castanopsis acuminatissima
Blume builds clusters of root suckers that sprout from a
common compact stem basis of up to 2 m height (Soepadmo, 1972). For each stem with a DBH 2 cm, we recorded
diameter and height separately above this common basis
and AGB was then calculated for every stem. (ii) Stranglers
of the genus Ficus L. (Moraceae) with a free-standing treelike structure are characterized by a large hole amidst of
the vertical axis that was originally filled by the trunk of
the host tree. The aboveground biomass of these stranglers
was estimated by calculating the AGB for both the strangler and the original host tree and subsequently subtracting
the latter from the former. These calculations were
conducted by Culmsee et al. (2010); contact this reference
for details.
Annual aboveground tree biomass production (ANPP; in
Mg dry mass ha1 yr1) was calculated from stem
increment data obtained from the repeated reading of
dendrometer tapes (UMS, Munich, Germany) that were
installed in December 2006 on 20 tree individuals per
study plot. The 20 trees in a plot were chosen by random
from all trees with a DBH >10 cm based within the core
zone of the plots; they belonged to different taxa and varied in DBH from 10 to 150 cm. The stem diameter increment recorded in the 120 trees in the first and second year
of the experiment (June 2007May 2008, and June 2008
May 2009 respectively) were extrapolated to all other tree
individuals of the covered families in the plots based on
DBH and by using family averages of increment. For the
24 (from 42) families not covered by the measurements,
such as Rubiaceae and most families with less than 1%
contribution to stand basal area, we calculated with increment rates that were proportional to their wood density in
the respective treatment, adopting a procedure described
in Hertel et al. (2009). The aboveground wood production
in the first and the second year of the experiment was
then calculated from basal area increment applying the
allometric regression equation used also for biomass estimation as described above. The mortality of trees in the
plot (for individuals with DBH 10 cm) was documented
in the 25-months period and mortality rates were calculated for every plot (see Sheil & May, 1996; Nepstad et al.,
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
I N D O N E S I A N R A I N F O R E S T D R O U G H T E X P E R I M E N T 1487
2007; da Costa et al., 2010). Mortality was also taken into
account in the productivity calculation.
True LAI (LAItrue) values were derived from hemispherical photographs taken before roof closure in 1 m height at
30 points per plot located at regular grid positions. CanEye software (INRA, Avignon, France) was used for image
analysis and the effective LAI (LAIeff) was calculated for a
zenith angle of 57.5 after Bonhomme & Chartier (1972) as
follows:
LAItrue LAIeff =CI
with the clumping index (CI) being calculated for each photo
after Lang & Xiang (1986).
For estimating annual leaf mass production, 12 litter
traps (75 cm 9 75 cm aperture) were installed in every plot
in March 2007. The litter was collected every 2 weeks and
sorted into leaves and other fine litter components (mostly
flowers, fruits and small twigs); litter dry mass was
obtained by drying at 70 C for 48 h. We assumed that
annual leaf litter production was equal to annual leaf production in both the first and the second year of the experiment.
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
Statistical analyses
The nonparametric U-test after Mann and Whitney was used
for detecting significant differences between roof and control
plots for the parameters above- and belowground biomass
and above- and belowground productivity. ANOVA was used to
test for significant treatment effects on leaf litter production
(log-transformed data). For identifying important factors
influencing the degree of radial growth reduction from the
first to the second year, we calculated linear mixed effects
models with six biomass or growth-related traits as fixed variables and plot and tree nested in plot as random factors. The
statistical analyses were done with the software package R
(R Development Core Team 2009, version 3.0.0).
1488 G . M O S E R et al.
Results
I N D O N E S I A N R A I N F O R E S T D R O U G H T E X P E R I M E N T 1489
(a)
(b)
(c)
(d)
(e)
Fig. 3 Daily rainfall amounts (a), relative air humidity (RH) and atmospheric vapour pressure deficit (VPD; measured at 16 m height
in a forest gap; b,c), and volumetric soil water content (VWC) expressed as volume fraction in 00.5 and 0.53.0 m soil depth (d,e). For
RH and VPD, daily means and the extremes are given. The first phase of the desiccation (8 months duration) with about 60% of
throughfall reduction and the second phase with ca. 80% throughfall reduction (17 months) are indicated.
Table 2 Annual net primary production (above- and belowground compartments) (in Mg dry mass ha1 yr1) during the first
and second year of the desiccation experiment in the control and roof plots (means 1 SE of n = 3 plots each). Different letters indicate significant differences between control and roof plots for the respective year (P < 0.05). The wood increment of the Ficus stranglers could not be determined
First year
Fraction
Control
Leaves
Other fine litter fractions
Stems and branches
Total aboveground
Fine roots
Stand total
6.19
1.19
5.84
13.21
1.76
14.97
Second year
Roof
0.63a
0.05a
0.22a
0.83a
0.28a
0.73a
5.93
1.17
5.41
12.50
1.96
14.46
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
Control
0.21a
0.18a
0.52a
0.52a
0.29a
0.43a
5.83
1.78
6.87
14.48
0.86
15.34
Roof
0.56A
0.05A
0.38A
0.45A
0.19A
0.35A
5.21
1.68
3.95
10.83
0.79
11.62
0.16A
0.15A
0.49B
0.24B
0.19A
0.41B
1490 G . M O S E R et al.
more than 40% compared to the control (4.0 vs.
6.9 Mg ha1 yr1, P < 0.05) and by about 27% relative
to the first year. Similar relative growth reductions
were found when the two most abundant tree families,
Fagaceae and Sapotaceae, were excluded. In contrast,
leaf litter mass and the remaining fine litter fraction
were not significantly reduced (Table 2). ANPP was by
25% lower in the roof than the control plots in the second year (P < 0.05).
The temporal fluctuation in leaf litter fall from May
2007 to January 2009 was similar in the roof and control
plots; significant treatment differences (positive or negative) occurred only in a minority of sampling intervals
with a higher significance in the second year and the
rewetting phase (Fig. 4). However, the annual amount
of leaf litter was not significantly reduced (P > 0.05;
Table 2).
The monitoring of stem diameter increment in the 60
trees of the three roof plots with variable species identity, tree height and DBH allowed searching for biometric, growth-related and phylogenetic factors that might
determine the drought sensitivity of radial stem
growth. We used the relative reduction in basal area
increment from the first (mild desiccation) to the second year (strong desiccation) as a measure of growth
sensitivity in the 60 investigated roof plot trees of 45
species. Contrary to expectation, wood specific gravity
and tree height had no significant influence on the magnitude of growth reduction in the experiments second
year (Table 3). The most influential factor was the relative basal area increment in the 1st year (P < 0.001),
followed by DBH (P = 0.017), that is, basal area increment per existing basal area (or specific cambial activity) and the size of the stem cambium (stem diameter)
determined the extent of growth reduction and not tree
height. Interestingly, only relative stem growth rate,
but not the absolute amount of wood production in the
first year (P = 0.131; Table 3), determined the desiccation sensitivity of radial growth.
Unit
dAIC
3
g cm
m
kg tree1
kg tree1
yr1
cm
%
L-ratio
1.748
0.519
0.097
0.276
0.252
2.519
1.903
2.277
0.615
0.113
0.168
0.131
3.741
28.431
5.741
30.431
0.017
<0.001
*log transformed.
Fig. 4 Time course of leaf litter fall during the desiccation experiment (June 2007May 2009) in the roof and control plots (means SE
of n = 3 plots). Significant differences between the treatments are marked with asterisks (P < 0.01*; P < 0.001**; P < 0.0001***).
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
I N D O N E S I A N R A I N F O R E S T D R O U G H T E X P E R I M E N T 1491
(a)
(b)
Fig. 5 (a) Percental contribution of the 16 most abundant families to the stand total of aboveground biomass (AGB; calculated from
diameter at breast height, tree height and wood density for all trees per plot). Data from the 1st census before the start of the desiccation
(numbers in parentheses behind the families indicate the sample size for the mean of the six plots). (b) Absolute change in annual wood
production in the 16 most abundant families in the three roof plots from the first (60% throughfall displacement) to the second year of
the desiccation (80% displacement; calculated from AGB and repeated dendrometer tape readings; negative change = growth reduction, positive change = growth increase). Numbers in parentheses behind the families indicate the number of trees per family in the
three roof plots equipped with dendrometers. Only families with more than 1% contribution to AGB are considered*; aboveground
wood production of stranglers (Moraceae) was not determined. *Families with less than 1% contribution to stand basal area are given
with the total number of trees in all six plots (nall) and the number of dendrometer trees in the roof plots (nden-roof) as (nall/nden-roof):
Aceraceae (2/0), Apocynaceae (2/0), Araliaceae (1/0), Arecaceae (7/0), Chrysobalanaceae (1/0), Cyatheaceae (18/1), Dracaenaceae
(14/0), Escalloniaceae (16/0), Gesneriaceae (5/0), Himantadraceae (3/0), Juglandaceae (1/0), Leguminosae/Mimosoideae (17/0),
Magnoliaceae (5/0), Melastomataceae (2/0), Monimiaceae (18/0), Myrsinaceae (6/0), Oleaceae (4/0), Pandanaceae (13/0), Rhizophoraceae (1/0), Rutaceae (21/1), Sabiaceae (8/0), Sapindaceae (18/0), Staphyleaceae (9/0), Sterculiaceae (2/0), Theaceae (2/0), Verbenaceae
(1/0).
cantly (by 56%; P < 0.01) in the roof plots and by 24% in
the control plots (P < 0.05). The ingrowth core
approach gave a fine root production estimate of ca.
1.76 Mg ha1 yr1 in the control and ca. 1.96 Mg ha1
yr1 in the roof plots in the experiments first year. In
the dry second year, the estimated productivity was
0.86 and 0.79 Mg ha1 yr1 in the control and roof
plots, respectively (difference between years significant
at P < 0.05, but no significant difference between the
treatments) (Table 2).
Discussion
1492 G . M O S E R et al.
Fig. 6 Comparing the vertical fine root biomass and necromass distribution to 120 cm soil depth before (February 2007) and immediately after the desiccation (June 2009). Given are means 1 SE, *** indicates a significant difference in fine root biomass between roof
and control for a distinct soil depth and sampling time.
ments at Tapaj
os and Caxiuan~
a, STEE was conducted
as a replicated experiment (n = 3 at the plot level)
enabling statistical testing of the roof effect and thereby
increasing the rigour of analysis. Working in a
perhumid environment with more than 2500 mm of
rain per year required a very effective rainfall displacement system that was able to reduce the throughfall by
ca. 80%. The operation of this construction at a remote
site was very labour-intensive and forced us to terminate the desiccation phase after 25 months, which was
long enough to produce a marked growth reduction
but did not lead to elevated tree mortality. The discussion of the results thus focuses on desiccation effects on
important above- and belowground productivity
components, fine root mortality and carbon allocation
shifts in a sublethal pre-dieback phase of the drought
response.
The tree mortality rates in the two treatments of STEE
(2.6 and 2.1% per year) were in the range of values
reported from other tropical moist forests without
excessive drought stress (e.g. Condit et al., 1995). Mortality did not increase in the roof plots in the 25-month
soil desiccation period indicating that our experiment
did not last long enough for driving sensitive trees
towards dieback. In support of this assumption, Brando
et al. (2008) observed a significantly higher tree mortality in the roof plot of the Tapaj
os experiment only in
the experiments third year with peak mortality rates as
high as 9.5% yr1; in the Caxiuan~
a experiment, significantly elevated dieback occurred only in the 5th year
(da Costa et al., 2010). While in natural droughts, most
mortality effects occurred within 9 months of the
drought (Phillips et al., 2010), a lag phase of 24 years
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
I N D O N E S I A N R A I N F O R E S T D R O U G H T E X P E R I M E N T 1493
Unexpected was the about 50% higher (nonleaf) fine
litter production in the experiments second year in
Pono. Since it was observed in both the roof and the
control plots, the increase likely is an expression of an
irregular mass flowering and fruiting event in this aseasonal forest, as it was observed in 2008 in particular in
the abundant Lithocarpus trees in the plots. The plus in
fruit production measured in the second year nearly
equalled the minus in fine root production in the roof
and control plots in that year pointing at a shift in allocation priorities that was independent from soil desiccation.
The principal comparability of the Pono forest and
the two Amazonian forests in terms of the ANPP
drought response astonishes, given the large differences in species composition and precipitation regimes.
While the reduction in REW was roughly comparable
in the three experiments, the Pono forest was exposed
to a shorter desiccation period and thus was lacking
certain tree reactions. It appears that the mean response
of tropical rainforest trees to sublethal drought stress is
not that different in the Neo- and Paleotropics, even
though large differences in the species response may
exist (see below). This leads us to the rejection of our
first hypothesis on the higher drought sensitivity of
trees in perhumid forests as compared to preadapted
species from semihumid forests (e.g. Phillips et al.,
2010). Clearly, this hypothesis was derived from field
observations on tree mortality during natural droughts
and did not consider growth responses of nondying
trees. Nevertheless, drought-induced growth sensitivity
and eventual dieback of a tree should be related
because the reasons for reductions in growth and overall tree vitality, for example, hydraulic failure, carbon
starvation or pathogen attack (McDowell et al., 2008)
should essentially be the same. Our results also do not
support the assumption that trees in South-East Asian
tropical forests should be more drought-sensitive than
Amazonian trees.
Extrapolating the results of throughfall manipulation experiments to the natural situation with regularly and irregularly occurring droughts must be
done with care. A main shortcoming of experiments
with subcanopy roofs is that only the soil is desiccated but not the atmosphere. This has the inevitable
consequence that the physiological effect of a natural
drought must be larger than that of a simulated
drought at equal rainfall reduction, because tree
water status and cambial growth is controlled not
only by soil moisture availability but also by the
atmospheric vapour pressure deficit and radiative
forcing (K
ocher et al., 2012). Furthermore, natural
drought periods are usually associated with higher
radiation loads than normal, which either can
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
1494 G . M O S E R et al.
nutrients towards root growth under conditions of
water shortage is predicted by the functional balance
hypothesis (Thornley, 1972a,b; Bloom et al., 1985) but is
not supported by our root biomass and root production
data. Similarly, Metcalfe et al. (2010) found no indication of a greater relative investment of total NPP in root
growth in the Caxiuan~
a forest upon drought exposure.
Clearly, our root data cover only the uppermost 1.2 m
(or 3 m) of the profile and ignore the deeper subsoil.
However, we have no quantitative information on
deep-reaching sinker roots that were found to extend to
more than 8 m in the deeply weathered forest soils of
semihumid eastern Amazonia (Nepstad et al., 1994).
For the very deep soils of Tapaj
os National Forest, soil
hydrological modelling indicated that layers below
5.5 m depth had an increasing importance for water
uptake only in the second and third year of the desiccation adding up to 12% of total water uptake, but that
the overall contribution of these layers remained
remarkably small and declined to near zero in the following years (Markewitz et al., 2010). While at least the
dominant tall Fagaceae Castanopsis acuminatissima may
possess sinker roots, in the rugged mountainous terrain
of Central Sulawesi, it is unlikely that the weathering
front and maximum rooting depth are as deep-reaching
as in parts of the Amazonian lowlands.
From the Amazonian experiments, it was concluded that tree mortality increases when the water
resources in the deep soil are depleted. Due to
lacking data on root water absorption in STEE, some
hints on the role of deep roots for stand water
uptake may be obtained from subsoil moisture
dynamics. Apparently, significant amounts of water
were extracted from deeper soil layers (0.53.0 m)
only from February to April 2009 after a rainless
period in December 2008 and January 2009 when
the topsoil matric potential approached 1.5 MPa in
the roof plots and root water uptake shifted downwards in the profile. The high soil matric potential
(>0.1 MPa) at 0.53.0 m depth in this period suggests that deep-reaching roots should continuously
have had access to easily available moisture despite
severe topsoil desiccation. However, when considering that 80% of the fine root biomass is located in
the top 40 cm of the profile and this part of the root
system suffered dieback and may mostly have lost
its water uptake function in the dry soil, it is
astonishing how the trees survived with a greatly
reduced fine root surface area without shedding
leaves. Uptake studies with tracers have to show
whether deep roots can fully compensate such losses
in absorbing surface area and how nutrient supply
is affected by topsoil desiccation (e.g. Kreuzwieser &
Gessler, 2010).
I N D O N E S I A N R A I N F O R E S T D R O U G H T E X P E R I M E N T 1495
in the stem xylem in this species (Schuldt et al., 2013).
Investigations on the carbohydrate supply and hydraulic functioning of trees from tropical perhumid forests
that were exposed to soil desiccation have to reveal the
underlying mechanisms leading to the registered wood
growth reduction.
The Sulawesi Throughfall Exclusion Experiment
(STEE) is only the third experiment on the drought
response of tropical forests, the first in a perhumid
climate and also the first outside eastern Amazonia.
Even though the desiccation lasted only for
25 months, that is, shorter than in the Amazonian
experiments, and the climatic, edaphic and phylogenetic conditions were largely different, we found
remarkably similar growth and allocation responses
among the trees in Sulawesi and Amazonia pointing
at common elements in the drought response of neoand paleotropical trees. On the other hand, STEE
revealed marked phylogenetic differences in the
responsiveness at the community level as where
observed in the Amazonian experiments. Our results
point to a higher growth sensitivity of trees with fast
diameter growth when moisture is not limiting; however, sensitivity was not higher in trees with lighter
wood. The greater suppression of fast-growing species
should lead in the long run to a forest community
with dominance of slower growing species, either
through competitive exclusion in a drier climate with
altered competitive hierarchies, or through higher
drought-induced mortality. We are less optimistic than
earlier authors about the potential for identifying precise soil moisture or rainfall minimum thresholds
beyond which tropical forests should encounter catastrophic dieback. More than in the Amazonian experiments, STEE has revealed the limitations of
throughfall exclusion experiments where dry soil
meets a relatively moist atmosphere that cannot be
manipulated. Continuous soil desiccation for 2 years
(Sulawesi) to 67 years (Amazonia) is far from the
reality in natural droughts that typically develop faster
than in the experiment because the atmosphere is dry
and cloudless. Rapid stress development with relatively short duration (as in natural droughts) and
more gradual stress development with long duration
(experiments) require different plant responses.
Acknowledgements
This study was part of the German-Indonesian collaborative
research project SFB 552: Stability of Rain Forest Margins in
Indonesia funded by the German Research Foundation
(DFG). The financial support is gratefully acknowledged. We
thank the inhabitants of Toro village for their hospitality,
Thomas Kl
uter and Hilmar M
uller and their team of local
workers for the roof construction and its maintenance, our
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
References
Aiba SI, Kitayama K (2002) Effects of the 1997-98 El Ni~
no drought on rain forests of
Mount Kinabalu, Borneo. Journal of Tropical Ecology, 18, 215230.
Allen CD, Macalady AK, Chenchouni H et al. (2010) A global overview of drought
and heat-induced tree mortality reveals emerging climate change risks for forests.
Forest Ecology and Management, 259, 660684.
Anderson LO, Malhi Y, Aragao LEOC, Ladle R, Arai E, Barbier N, Phillips O (2010)
Remote sensing detection of droughts in Amazonian forest canopies. New Phytologist, 187, 733750.
Baker TR, Phillips OL, Malhi Y et al. (2004) Variation in wood density determines
spatial patterns in Amazonian forest biomass. Global Change Biology, 10, 545
562.
Beier C, Beierkuhnlein C, Wohlgemuth T et al. (2012) Precipitation manipulation
experiments challenges and recommendations for the future. Ecology Letters, 15,
899911.
Bloom AJ, Chapin FS III, Mooney HA (1985) Ressource limitation in plants - an
economic analogy. Annual Review of Ecology and Systematics, 16, 363392.
Bonhomme R, Chartier P (1972) The interpretation and automatic measurement of
hemispherical photographs to obtain sunlit foliage. Israelian Journal of Agricultural
Research, 22, 5361.
Brando PM, Nepstad DC, Davidson EA, Trumbore SE, Ray D, Camargo P (2008)
Drought effects on litterfall, wood production and belowground carbon cycling in
an Amazon forest: results of a throughfall reduction experiment. Philosophical
Transactions of the Royal Society B-Biological Sciences, 363, 18391848.
Brda N, Huc R, Granier A, Dreyer E (2006) Temperate forest trees and stands under
severe drought: a review of ecophysiological responses, adaptation processes and
long-term consequences. Annals of Forest Science, 63, 625644.
Breshears DD, McDowell NG, Goddard KL, Dayem KE, Martens SN, Meyer CW,
Brown KM (2008) Foliar absorption of intercepted rainfall improves woody plant
water status most during drought. Ecology, 89, 4147.
Cassana FF, Dillenburg LR (2013) The periodic wetting of leaves enhances water relations and growth of the long-lived conifer Araucaria angustifolia. Plant Biology, 15,
7583.
Chave J, Andalo C, Brown S et al. (2005) Tree allometry and improved estimation of
carbon stocks and balance in tropical forests. Oecologia, 145, 8799.
Chave J, Muller-Landau HC, Baker TR, Easdale TA, ter Steege H, Webb CO (2006)
Regional and phylogenetic variation of wood density across 2456 neotropical tree
species. Ecological Applications, 16, 23562367.
Christensen JH, Hewitson B, Buduioc A et al. (2007) Regional climate projections. In:
Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the
Fourth Assessment Report of the Intergovernmental Panel on Climate Change (eds. Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL)
pp. 847940. Cambridge University Press, Cambridge.
Clark DA, Brown S, Kicklighter DW, Chambers JQ, Thomlinson JR, Ni J (2001) Measuring net primary production in forests: concepts and field methods. Ecological
Applications, 11, 356370.
Condit R, Hubbell SP, Foster RB (1995) Mortality rates of 205 neotropical tree and
shrub species and the impact of a severe drought. Ecological Monographs, 65, 419
439.
da Costa ACL, Galbreith D, Almeida S et al. (2010) Effect of 7 yr of experimental
drought on vegetation dynamics and biomass storage of an eastern Amazonian
rainforest. New Phytologist, 187, 579.
Culmsee H, Pitopang R (2009) Tree diversity in sub-montane and lower montane primary rain forests in Central Sulawesi. Blumea, 54, 119123.
Culmsee H, Leuschner C, Moser G, Pitopang R (2010) Forest aboveground biomass
along an elevational transect in Sulawesi, Indonesia, and the role of Fagaceae in
tropical montane rain forests. Journal of Biogeography, 37, 960974.
Dai A (2013) Increasing drought under global warming in observations and models.
Nature Climate Change, 3, 5258.
Dietz J, H
olscher D, Leuschner C, Hendrayanto (2006) Rainfall partitioning in relation to forest structure in differently managed montane forest stands in Central Sulawesi, Indonesia. Forest Ecology and Management, 237, 170178.
1496 G . M O S E R et al.
Eller CB, Lima AL, Oliveira RS (2013) Foliar uptake of fog water and transport belowground alleviates drought effects in the cloud forest tree species, Drimys brasiliensis
Merryfield WJ (2006) Changes to ENSO under CO2 doubling in a multimodel ensemble. Journal of Climate, 19, 40094027.
Metcalfe DB, Meir P, Aragao LEOC et al. (2010) Shifts in plant respiration and carbon
use efficiency at a large-scale drought experiment in the eastern Amazon. New
Phytologist, 187, 608621.
Mokany K, Raison RJ, Prokushkin AS (2006) Critical analysis of root:shoot ratios in
terrestrial biomes. Global Change Biology, 12, 8496.
Nepstad DC, De Carvalho CJR, Davidson EA et al. (1994) The role of deep roots in
the hydrological and carbon cycles of Amazonian forests and pastures. Nature,
372, 666669.
Nepstad DC, Moutinho P, Dias-Filho MB et al. (2002) The effects of partial
throughfall exclusion on canopy processes, aboveground production, and biogeochemistry of an Amazon forest. Journal of Geophysical Research, 107, LBA
53-1LBA 53-18.
Nepstad DC, Tohver IM, Ray D, Moutinho P, Cardinot G (2007) Mortality of large
892898.
Godbold DL, Fritz HW, Jentschke G, Meesenburg H, Rademacher P (2003) Root turnover and root necromass accumulation of Norway spruce (Picea abies) are affected
by soil acidity. Tree Physiology, 23, 915921.
Goldsmith GR, Matzke NJ, Dawson TE (2013) The incidence and implications of
clouds for cloud forest plant water relations. Ecology Letters, 16, 307314.
Granier A, Reichstein M, Breda N et al. (2007) Evidence for soil water control on car-
trees and lianas following experimental drought in an Amazon forest. Ecology, 88,
22592269.
van Nieuwstadt MGL, Sheil D (2005) Drought, fire and tree survival in a Borneo rain
forest, East Kalimantan, Indonesia. Journal of Ecology, 93, 191201.
Nishimura TB, Suzuki E, Kohyama T, Tsuyuzaki S (2007) Mortality and growth of
trees in peat-swamp and heath forests in Central Kalimantan after severe drought.
Plant Ecology, 188, 165177.
bon and water dynamics in European forests during the extremely dry year: 2003.
Agricultural and Forest Meteorology, 143, 123145.
Guilyardi E (2006) El Ni~
no-mean state-seasonal cycle interactions in a multi-model
ensemble. Climate Dynamics, 26, 329348.
Hansen J, Sato M, Ruedy R, Lo K, Lea DW, Medina-Elizade M (2006) Global temperature change. Proceedings of the National Academy of Sciences of the USA, 103,
Oey DS (1990) Berat jenis dari jenis-jenis kayu berat Indonesia dan pengertian beratnya kayu
untuk keperluan praktek. FRPDC, Forestry Department, Bogor, Indonesia.
Paeth H, Scholten A, Friederichs P, Hense A (2008) Uncertainties in climate change
prediction: El Ni~
no-Southern Oscillation and monsoons. Global and Planetary
Change, 60, 265288.
Philip S, van Oldenborgh GJ (2006) Shifts in ENSO coupling processes under global
1428814293.
Hertel D, Moser G, Culmsee H, Erasmi S, Horna V, Schuldt B, Leuschner C (2009)
Below- and above-ground biomass and net primary production in a paleotropical
natural forest (Sulawesi, Indonesia) as compared to neotropical forests. Forest Ecology and Management, 258, 19041912.
Itoh A, Nanami S, Harata T et al. (2012) The effect of habitat association and edaphic
conditions on tree mortality during El Ni~
no-induced drought in a Bornean Dip-
Preston KA, Cornwell WK, Denoyer JL (2006) Wood density and vessel traits as distinct correlates of ecological strategy in 51 California coast range angiosperms.
New Phytologist, 170, 807818.
Rosenzweig C, Casassa G, Karoly DJ et al. (2007) Assessment of observed changes
and responses in natural and managed systems. In: Climate Change 2007: Impacts,
Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assess-
Lang ARG, Xiang YQ (1986) Estimation of leaf-area index from transmission of direct
sunlight in discontinuous canopies. Agricultural and Forest Meteorology, 37, 229
243.
Latif M, Anderson D, Barnett T et al. (1998) A review of the predictability and prediction of ENSO. Journal of Geophysical Research-Oceans, 103, 1437514393.
Lemmens RHMJ, Soerianegara I, Wong WC (1995) Timber Trees: Minor Commercial
Timbers. Leiden, Backhuys Publishers, Plant resources of South-East Asia.
ment Report of the Intergovernmental Panel on Climate Change. (eds. Parry ML, Canziani OF, Palutikof JP, van der Linden PJ, Hanson CE) pp. 79131. Cambridge
University Press, Cambridge.
Schuldt B, Leuschner C, Horna V, Moser G, K
ohler M, van Straaten O, Barus H (2011)
Change in hydraulic properties and leaf traits in a tall rainforest tree species subjected to long-term throughfall exclusion in the perhumid tropics. Biogeosciences, 8,
21792194.
depletion under simulated drought in the Amazon: impacts on deep root uptake.
New Phytologist, 187, 592607.
McCulloh KA, Johnson DM, Meinzer FC, Voelker SL, Lachenbruch B, Domec J-C (2012)
Hydraulic architecture of two species differing in wood density: opposing strategies
in co-occurring tropical pioneer trees. Plant, Cell & Environment, 35, 116125.
McDowell N, Pockman WT, Allen CD et al. (2008) Mechanisms of plant survival and
Slik JWF (2006) Estimating species-specific wood density from the genus average in
Indonesian trees. Journal of Tropical Ecology, 22, 481482.
Smith MD, Knapp AK, Collins SL (2009) A framework for assessing ecosystem
dynamics in response to chronic resource alterations induced by global change.
Ecology, 90, 32793289.
Soepadmo E (1972) Fagaceae. Flora Malesiana, 1, 265403.
mortality during drought: why do some plants survive while others succumb to
drought? New Phytologist, 178, 719739.
Meir P, Woodward FI (2010) Amazonian rain forests and drought: response and vulnerability. New Phytologist, 187, 553.
Soerianegara I, Lemmens RHMJ (1993) Timber Trees: Major Commercial Timbers. Leiden, Backhuys Publishers, Plant Resources of South-East Asia.
Sosef MSM, Hong LT, Prawirohatmodjo S (1998) Timber Trees: Lesser-Known Timbers.
Leiden, Backhuys Publishers, Plant Resources of South-East Asia.
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
I N D O N E S I A N R A I N F O R E S T D R O U G H T E X P E R I M E N T 1497
van Straaten O (2010) Drought effects on soil carbon dioxide efflux in two ecosystems
in Central Sulawesi. PhD thesis, University of G
ottingen, G
ottingen. Available at:
http://webdoc.sub.gwdg.de/diss/2010/straaten/straaten.pdf.
van Straaten O, Veldkamp E, Corre MD (2011) Simulated drought reduces soil CO2
efflux and production in a tropical forest in Sulawesi, Indonesia. Ecosphere, 2, art119.
Thornley JHM (1972a) A balanced quantitative model for root:shoot ratios in vegetative plants. Annals of Botany, 36, 431441.
Thornley JHM (1972b) A model to describe the partioning of photosynthate during
vegetative plant growth. Annals of Botany, 36, 419430.
2013 John Wiley & Sons Ltd, Global Change Biology, 20, 14811497
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Figure S1 Maps of the mean volumetric soil water content
in the six plots (uppermost 30 cm) according to weekly measurements with a mobile TDR system (each 100 measurements per plot and occasion). Given are the averages of the
experiments second phase (January 1, 2008 to May 7, 2009).