Cyclone Effects on the Structure and Production of a Tropical Upland Rainforest: Implications for Life-History Tradeoffs

Abstract

Wind is known to alter the structure and functioning of forest ecosystems. Because the intensity and frequency of severe wind
events are likely to increase, it is important to understand the species- and substrate-specific effects of these disturbances.
We assessed the structure and production among 63 species of trees in an Australian tropical rainforest before and after Cyclone
Larry (March 2006). We assessed forest occurring on two different substrates: nutrient-poor schist and relatively nutrient-rich
basalt. Leaf area reduction and stem breakage were markedly variable among species, but were more evident on basalt soils
than schist soils, and were positively correlated with leaf N and P. In the 18-month period following the cyclone, litterfall,
stem biomass increment, and ANPP were 44, 20, and 27% of pre-cyclone measurements and did not differ between soils. More severe
modification of leaves, branches, and stems on basalt soils, relative to schist soils, suggests that trees/species growing
on nutrient-limited soils are less susceptible to high winds. Disturbance regime and resource availability are likely to interact,
creating potential plant strategies that increase fitness either by enhanced investments in carbon or enhanced investments
in nitrogen and phosphorus.

Full text

Cyclone Effects on the Structure
and Production of a Tropical Upland
Rainforest: Implications
for Life-History Tradeoffs
Sean M. Gleason,
1
* Laura J. Williams,
1
Jennifer Read,
1
Daniel J. Metcalfe,
2
and Patrick J. Baker
1
1
School of Biological Sciences, Monash University, Bld 18, Clayton, Victoria 3800, Australia;
2
CSIRO Sustainable Ecosystems,
Tropical Forest Research Centre, PO Box 780, Atherton, Queensland 4883, Australia
ABSTRACT
Wind is known to alter the structure and func-
tioning of forest ecosystems. Because the intensity
and frequency of severe wind events are likely to
increase, it is important to understand the species-
and substrate-specific effects of these disturbances.
We assessed the structure and production among
63 species of trees in an Australian tropical rain-
forest before and after Cyclone Larry (March 2006).
We assessed forest occurring on two different sub-
strates: nutrient-poor schist and relatively nutrient-
rich basalt. Leaf area reduction and stem breakage
were markedly variable among species, but were
more evident on basalt soils than schist soils, and
were positively correlated with leaf N and P. In the
18-month period following the cyclone, litterfall,
stem biomass increment, and ANPP were 44, 20,
and 27% of pre-cyclone measurements and did not
differ between soils. More severe modification of
leaves, branches, and stems on basalt soils, relative
to schist soils, suggests that trees/species growing
on nutrient-limited soils are less susceptible to high
winds. Disturbance regime and resource availabil-
ity are likely to interact, creating potential plant
strategies that increase fitness either by enhanced
investments in carbon or enhanced investments in
nitrogen and phosphorus.
Key words: cyclone; hurricane; nutrients; pro-
ductivity; life-history; tradeoff; soils; disturbance.
INTRODUCTION
Severe wind events, including cyclones, tornadoes,
and thunderstorm downbursts, occur in nearly all
forest systems throughout the world. Forest com-
position, structure, and functioning are affected by
the severity and frequency of these events. Al-
though much information on the effects of wind
has been gathered from Caribbean, North Ameri-
can, and South American forest systems, data re-
main limited for the Old World tropics, and in
particular, Australian rainforests. This paper de-
Received 17 January 2008; accepted 10 July 2008; published online
23 September 2008
Electronic supplementary material: The online version of this article
(doi:10.1007/s10021-008-9192-6) contains supplementary material,
which is available to authorized users.
S.M.G. contributed to the experimental design, research, data analysis,
and manuscript writing. L.J.W. contributed to the experimental design,
research, and manuscript writing. J.R. contributed to the experimental
design, data analysis, and manuscript writing. D.J.M. and P.J.B. con-
tributed to the experimental design and manuscript writing.
*Corresponding author; e-mail: sean.gleason@sci.monash.edu.au
Ecosystems (2008) 11: 1277–1290
DOI: 10.1007/s10021-008-9192-6
1277

scribes the effects of Cyclone Larry on a diverse
tropical rainforest in Queensland, Australia. We
concentrate on the species-specific and soil-specific
differences in structural modification and above-
ground production as a result of Cyclone Larry.
Considering that the intensity (Emanuel 1987) and
frequency of severe cyclones/hurricanes (Webster
and others 2005) may be increasing, species- and
soil-specific data may help improve our under-
standing and management of cyclone-affected for-
ests.
Structural modification from cyclones includes
the stripping of leaves from branches and the
breakage of branches, roots, and stems. Factors
mitigating these effects include plant attributes,
such as properties of wood (Asner and Goldstein
1997), leaves and petioles (Niklas 1996; Niklas
1999; Cordero and others 2007), roots and but-
tresses (Fraser 1962; Ennos 1993; Crook and others
1997; Stokes 1999), forest level physiognomy
(Foster 1988; Boucher and others 1990; Foster and
Boose 1992; Cooper-Ellis and others 1999), prop-
erties of the wind event itself (Canham and others
2001), as well as resource availability (Herbert and
others 1999; Beard and others 2005).
Cyclones may affect forest productivity by
inducing mortality, reducing tree vigor (Herbert
and others 1999), altering nutrient and C inputs
(Lodge and McDowell 1991; Sanford and others
1991; Scatena and others 1993; McDowell and
others 1996), or inducing salt toxicity (Blood and
others 1991). Generally, recovery of aboveground
production following wind disturbance occurs
quickly, with leaf area recovering within 2 years
or faster (Whigham and others 1991; Herbert and
others 1999), although severe or frequent wind
events may result in longer recovery periods
(Scatena and others 1996; Sherman and others
2001; Lin and others 2003; Beard and others
2005). Belowground biomass may recover more
slowly than aboveground biomass (Frangi and
Lugo 1991; Parrotta and Lodge 1991; Silver and
Vogt 1993; Herbert and others 1999; Beard and
others 2005). Quick recovery of aboveground net
primary productivity (ANPP) is generally attrib-
uted to large nutrient inputs, tight biotic control of
nutrient inputs (Cooper-Ellis and others 1999),
quick leaf area recovery (Unwin and others 1988;
Herbert and others 1999
; You and Petty 1991),
enhanced growth rates of surviving stems (Mer-
rens and Peart 1992; Scatena and others 1996),
and enhanced rates of recruitment (Frangi and
Lugo 1991; Whigham and others 1991; Guzma
´
n-
Grajales and Walker 1991; Uriarte and others
2005).
Cyclone effects on the structure and production
of forests are often species-specific (Foster 1988;
Gresham and others 1991; Asner and Goldstein
1997; Canham and others 2001; Sherman and
others 2001; Uriarte and others 2005). Addition-
ally, species may vary in their capacity to resist
wind (resistance), offset the effects of injury (tol-
erance), and recover from injury (resilience). It is
unlikely that any one taxon will display all three of
these characteristics (resistance, resilience, toler-
ance). This is because site- and habitat-specific re-
source limitations will likely force a tradeoff among
them. Wind resistance versus shade tolerance (or
regeneration requirements) is one such tradeoff
(Bellingham and others 1995; Batista and Platt
2003; Ostertag and others 2005), whereby shade
intolerant trees ‘‘trade’’ resistance for fast growth
and/or quick maturation.
Fast growth on infertile soils is ‘‘expensive’’ be-
cause nutrients are limited—fast growth is facili-
tated by readily available soil nutrients. Quick
recovery after injury (resilience) and maintaining
fitness despite injury (tolerance), for example,
increasing reproductive output, are also ‘‘expen-
sive’’ on infertile sites because these strategies re-
quire additional nutrient uptake. For this reason,
natural selection may favor resistance on infertile
sites because there are fewer direct nutrient costs to
resistance, whereas, tolerance and resilience may
be favored on fertile sites. Thus, plants will main-
tain fitness in disturbed environments either
through investments in C (resistance) on poor soils
or investments in nutrients (tolerance and resil-
ience) on fertile soils.
MATERIALS AND METHODS
Site Description and Species
Research sites were established in the World Heri-
tage rainforest of Wooroonooran National Park,
Queensland, Australia, at approximately 17°22’S
and 145°43’E. This area is approximately 32 km
from the coast and situated directly west of Mt.
Bartle Frere on the Atherton Tablelands. Elevation
ranges from 700 to 850 m throughout the study
area (Figure 1). The forest is floristically complex,
with 1,053 rainforest tree species within the re-
gional area of Northeast Queensland (Hyland and
others 2002) and approximately 80 species within a
0.25 ha area. Mean annual precipitation is
approximately 3.5 m, of which about 70% falls
between November and April.
Forest within the study area comprises ever-
green species and has been described as complex
1278 S. M. Gleason and others

mesophyll vine forest (Tracey 1982). Using the
Holdridge classification system (Holdridge 1947),
the study area would be classified as sub-tropical
wet forest. Although logging did take place in
Wooroonooran into the late 1970s, these opera-
tions involved selective harvesting and are char-
acterized by the presence of over-grown trails and
a few large stumps. Such areas were avoided in
the study.
Wooroonooran National Park has experienced
four major cyclones in the last century: 1918, 1956
(described in Webb 1958), 1986 (Winifred), and
2006 (Larry). Cyclone Larry (personal observa-
tions) and Cyclone Winifred (Unwin and others
1988) resulted in more marked effects near lowland
coastal areas than upland rainforest farther inland.
The intensity of Cyclone Larry (category 4) and the
1918 cyclone (category 4 or 5) are likely to have
been similar, whereas Cyclone Winifred was less
severe (category 3).
We categorize the species into two main groups:
species occurring on both soil types (generalist) and
species occurring only on schist soils (schist spe-
cialists). Basalt-favoring species do exist, but are
not exclusively found on basalt, whereas schist
specialists were never found on basalt soils within
the study area. For this reason, we concentrate on
comparisons between ‘‘schist specialists’’ and
‘‘generalists’’. We use the term ‘‘schist specialist’’
only as comparative description because these
species may occur on soils other than schist outside
the area of study (Hyland and others 2002; Met-
calfe and Ford 2008).
Soil Description and P Analysis
Soils are derived from igneous basalt (Red Ferro-
sol—Maalan series) and metamorphic schist parent
materials (Red Dermosol—Galmara series) (Mal-
colm and others 1999). Schist soils (Galmara) are
characterized by low pH, high organic C, moderate
available N, moderate cation exchange capacity,
low aluminum saturation, and markedly low soil P
(Table 1). Basalt soils (Maalan) are characterized by
low pH, high organic C, high available N, high
cation exchange capacity, high aluminum satura-
tion, and high soil P (Table 1).
Phosphorus limitation of plant growth on the
schist soil has been previously demonstrated (Ker-
ridge and others 1972), as well as supported by
indirect evaluation of plant stoichiometry, nutrient
use traits, and distribution characteristics of P-
adapted species in these forests (Gleason and oth-
ers, unpublished manuscript). Phosphorus limita-
tion on schist is not surprising considering the age
of the schist soil (about 350 million years) relative
to the basalt soil (approximately 3 million years), as
well as the high P contents of basalt parent mate-
rial. Studies of P weathering in forested ecosystems
suggest that markedly weathered soils, similar to
the schist soil, are likely to exhibit P limitation
(Walker and Syers 1976; Crews and others 1995;
Vitousek 2004) even when parent materials con-
tain much P, which schist does not.
Soil P was fractionated using methods modified
from Tiessen and Moir (1993). Sodium bicarbonate
(NaHCO
3
)-extractable P represents the most labile
Figure 1. Location of large
temporary plots within
Wooroonooran National
Park. Contours show
elevation (m) above sea
level. Cyclone Larry
proceeded in an east to
west direction
approximately 17 km
south of the study area.
Small permanent plots are
located within the dashed
oval section.
Cyclone Effects on a Tropical Rainforest 1279

P fraction, whereas alkaline (NaOH) and acid (1 M
HCl) extracts represent more occluded fractions of
P associated with Al + Fe and Ca, respectively
(Table 1). Hot HCl (10 M)-extractable P represents
a more stable ‘‘residual’’ fraction (Tiessen and Moir
1993), although the nature and stability of this P
fraction is not well known (Garcia-Montiel and
others 2000). Organic P was determined in an ali-
quot of combined extracts after hot acid digestion
with potassium persulfate. Although the NaHCO
3
-
extractable P fraction (low for both soils) does
represent a relatively labile pool, it does not char-
acterize the rate of P supply to plants over time.
More stable fractions of inorganic and organic P
(including residual P) can contribute significantly
to P supply rates over time intervals from months
to years (Hedley and others 1982; Schoenau and
others 1989), particularly in the rhizosphere
(Hedley and others 1982).
Species-Specific Cyclone Effects
on Temporary Plots
Species resistance to cyclone disturbance was as-
sessed for 63 tree species in a 14 km
2
area of Wo-
oroonooran National Park. Modification to leaves,
branches, and stems (described in detail below)
were assessed for each tree larger than 10 cm
diameter at breast height (DBH) within 24 0.25 ha
temporary plots, hereafter referred to as ‘‘large
temporary plots’’ (Figure 1). Large temporary plots
were established during May and June 2005 (1–
2 months after Cyclone Larry) in an area of forest
located approximately 17 km north of Cyclone
Larry’s track. Plot locations were first randomly
located on soil maps, avoiding steep, protected, or
overly exposed locations. A two-person team then
surveyed each plot. Of the nine schist plots, three
were located on ridge sites (33%). Of the 15 basalt
plots four were located on ridge sites (27%). Plots
not located on ridges were located on somewhat
more protected hillsides, not exceeding 18° of
slope. Fewer large temporary plots were placed on
schist soils (nine) than on basalt soils (15) because
schist soils are less common in the study area. For
each individual, we recorded species, structural
change, DBH, and crown illumination index (CII).
Crown illumination index is a relative score of
canopy position/light availability and was modified
from Clark and Clark (1992). We scored tree
crowns from one to three, based on their relative
position in the canopy. ‘‘1’’ indicates a dominant
crown, exposed to direct sunlight throughout the
day. ‘‘2’’ indicates an intermediate crown, exposed
to direct sunlight during part of the day. ‘‘3’’ indi-
cates a suppressed crown, exposed to little direct
sunlight during any time of the day.
Cyclone effects on each tree were assessed by
visually estimating the percent of the crown lost to
leaf stripping (leaf separation at the petiole) and by
branch breakage. In addition, we recorded the
average size of the largest broken branch and
modification to the stem (that is, windthrown,
broken at top, broken mid-stem, broken near the
base). It was noted whether structural changes
were caused by wind or by falling branches/stems
from adjacent trees. Overall wind severity on each
plot was estimated by scoring each large temporary
Table 1. Sequential Soil P Fractions and Other Soil Variables for Basalt and Schist Soils
Soil variable Basalt Schist P
Soil P fraction (lgg
-1
)
NaHCO3 (0.5 M) 4.68 (0.56) 6.38 (1.88) 0.408
NaOH (0.1 M) 84.2 (16.3) 14.5 (2.6) 0.002
HCl (1 M) 3.99 (0.87) 1.60 (0.59) 0.046
HCl (10 M) 969 (77) 141 (69) <0.001
Total P 1515 (111) 240 (77) <0.001
Total inorganic P 1062 (91) 164 (69) <0.001
Total organic P 453 (21) 76.0 (10.1) <0.001
Other soil characteristics
pH 4.6 (0.03) 4.0 (0.1) <0.001
Organic carbon (%) 4.3 2.7 –
Total N (%) 0.32 0.22 –
ECEC (meq 100 g
-1
)41 15 –
Al saturation (%) 80 1 –
Bulk density (g cm
-3
) 1.00 (0.03) 0.86 (0.10) <0.001
All measurements reflect non-fertilized surface horizons (0–10 cm) sampled prior to Cyclone Larry. Values in parentheses represent 1 standard error of the mean (n = 12).
Variables for which no standard error is given represent data taken from Malcolm and others (1999).
1280 S. M. Gleason and others

plot as a one, two, or three (from least to most
affected). Slope, aspect, and soil type (schist or
basalt) were also noted.
Permanent Plot Establishment and
Fertilization
Twenty-four 100 m
2
circular plots (hereafter re-
ferred to as ‘‘small permanent plots’’) were estab-
lished in September 2005 (6 months before the
cyclone) along two transects that crossed a basalt–
schist geologic boundary. This area is delineated
with an oval in Figure 1. These small permanent
plots provided forest production data (biomass
increment, litterfall) for both soil types. Cyclone
effects were also recorded on the small permanent
plots, using identical methods as for the large
temporary plots.
Transects were located on two ridges (approxi-
mately 300 m apart), with 12 small permanent
plots on each ridge (six plots on each soil type
within a ridge). Small permanent plots were each
placed 50 m apart. Large gaps were avoided in an
effort to homogenize basal area among plots. Small
permanent plots contained between 14 and 35
trees with a DBH greater than 5 cm. All trees with a
DBH greater than 10 cm were fitted with a stainless
steel dendrometer band (Keeland and Young
2005). Azimuth and distance from plot center,
species, DBH, and height were recorded for all
trees.
Half of all small permanent plots on each soil
type were fertilized with 100 kg of P (as super-
phosphate) and N (as urea) ha
-1
y
-1
. Fertilizer was
applied in two applications, one at the start of the
experiment (October 2005) and one midway
through the experiment (June 2006). Plots for
fertilization were chosen randomly within soil
treatment.
In addition to tree measurements within the
small permanent plots, DBH and distance to plot
center were recorded for all trees within a plot ra-
dius of 11.64 m. This was done to calculate com-
petition indexes (Daniels 1976) for every banded
stem within each small permanent plot. Although a
range of competition indexes (distance dependant
and independent models) were evaluated, the
Daniels (1976) equation best fit our production
data. This method (Daniels 1976) calculates com-
petition intensity as:
X
i
D
j
=D
i

=M
ji
where D
j
= the diameter at breast height (cm) of a
competitor tree, D
i
= the diameter (cm) of the
subject tree, and M = the distance (m) between a
competitor tree and the subject tree. The sum in-
cludes only trees that are within a fixed angle
sweep (basal area factor = 2.3 m
2
ha
-1
). Thus,
small trees and distant trees are not included in the
calculation. This measure was used as a covariate
when comparing relative growth rates (RGR) be-
tween soils and dates (see data analysis section
below).
Litter Production
Litterfall collectors were laid out in each small
permanent plot within a north–south oriented grid.
Twelve self-draining, 530-cm
2
collectors were
staked approximately 20 cm above the ground
within this grid. Litterfall was collected at two-
week intervals and was dried in a fan-forced oven
at 65°C to constant weight. On five occasions lit-
terfall was collected at monthly intervals. Litterfall
mass calculations from these longer collection
periods are likely to be underestimated due to
decomposition; however, because our main inter-
est was soil comparisons, these marginal losses are
not likely to adversely affect the conclusions of this
study. Litterfall mass was calculated for dry season,
wet season, and wet + dry season collections.
Aboveground Net Primary Productivity
(ANPP) and Relative Growth Rates (RGR)
Tree biomass was estimated using a general allo-
metric equation developed for lowland rainforest in
West Kalimantan, Indonesia (Yamakura and others
1986).
W
S
¼ 0.02903 D
2
H

0:9813
W
B
¼ 0.1192*W
1:059
S
W
L
¼ 0.09146 W
S
þ W
B
ðÞ
0:7266
where W
S
, W
B
, and W
L
represent dry mass (kg) of
trunk, branches, and leaves, respectively. D repre-
sents the diameter (cm) at breast height and H rep-
resents height (m). These equations have been used
to estimate tree biomass in a montane forest in Java
(Yamada 1997), a temperate rainforest in Japan
(Aiba and Kitayama 1999), and a montane forest in
Kinabalu Malaysia (Kitayama and Aiba 2002). We
chose this equation over similar rainforest equations
(see Brown 1997) because the Yamakura equation
has been applied to forests more similar in structure
(basal area, height), species composition, and prox-
imity to the Queensland sites.
Cyclone Effects on a Tropical Rainforest 1281

Aboveground net primary productivity was cal-
culated on each small permanent plot by summing
the dry weight production of trunk biomass + leaf
biomass + litterfall for the following time periods:
the period leading up to the cyclone (0.5 years),
the period after the cyclone (1.5 years), and the
entire study period (2.0 years). Aboveground net
primary productivity was estimated for individual
trees by summing the dry weight production of
trunk biomass + leaf biomass over the same study
periods given above (tree-specific litterfall is not
included).
Relative growth rates were calculated for all
banded stems. Relative growth rates were calcu-
lated as:
RGR¼ LN finalbiomassðÞLN initialbiomassðÞ½/time
Using RGR was considered necessary because the
cyclone created large plot-level basal area differ-
ences, making meaningful ANPP comparisons
among plots difficult.
Leaf Area
Hemispherical photographs were taken with a Ni-
kon CoolpixÒ digital camera at five fixed locations
within each plot on four different occasions during
the study period. Photographs were taken early in
the morning, before the sun rose to a position
where it became visible in the photographs. Leaf
area index (LAI) was calculated for each picture
(HemiviewÒ) and averaged to yield LAI for each
plot.
Leaf N and P, Shade Tolerance,
and Wood Density
Green and recently senesced leaves were collected
within the small permanent plots. Green leaves
were collected from fallen branches the day after
Cyclone Larry. Although every effort was made to
sample from as many trees as possible from each
soil type, for six species leaves were collected from
fewer than three trees per soil type. Senesced
leaves were collected daily from suspended shade
cloth, and the samples contained leaves from an
unknown number of individuals within (or near)
the plot. Leaves were dried at 65°C to constant
weight, ground in a ball mill, and analyzed for total
N (CHN analyzer) and total P. Total P was analyzed
using inductively coupled plasma emission spec-
troscopy following microwave-assisted nitric acid
digestion (Rechcigl and Payne 1990).
Shade tolerance scores were obtained for 26 of
our species from a previous study (Osunkoya
1996). Briefly, species were scored from one to five
based on regeneration characteristics of their
seedlings. A score of ‘‘1’’ indicates germination and
survival at the forest edge and ‘‘5’’ indicates ger-
mination and survival in understory shade (Os-
unkoya 1996). All species were independently
scored by six ecologists and averaged (Osunkoya
1996). Regression analysis was then used to eval-
uate relationships between shade tolerance (of
seedlings), wind effects, and leaf N.
Wood density values were obtained from a pre-
viously published report (Department of Primary
Industries 1998) based on an unknown number of
samples from mature trees and various locations
across northeast Queensland. From this report, we
obtained wood density values for 59 of our species.
Regression analyses were then used to evaluate
relationships between wood density and wind
effects.
Estimating Growth Rates for Species
in Large Temporary Plots
Forest growth equations developed by Vanclay
(1991) were used to calculate the species-specific
diameter increment for 54 of the species we sur-
veyed in the large temporary plots. These equations
contain species-specific and site-specific growth
parameters and are compiled from over 40 years of
permanent plot inventory from northeast Queens-
land rainforests, including our study area (Vanclay
1991). We ran the model for each species to estimate
the growth rate of a 10-cm DBH tree under condi-
tions of good (basalt) and poor (schist) site quality.
These growth rate estimates were then used in lin-
ear regression models to test the growth-rate versus
resistance tradeoff—do faster growing species
‘‘trade’’ resistance for fast growth? Additionally,
species were sorted into three equal growth rate
groups representing slow (0–0.02 cm y
-1
), medium
(0.02–0.70 cm y
-1
), or fast (>0.70 cm y
-1
) diam-
eter growth for comparison with wind effects and
leaf nutrient concentrations (Figure 7). We note
that these growth estimates are independent of our
RGR calculations (see methods above), which were
determined using dendrometer bands from the
small permanent plots.
Data Analysis
For CII comparisons, a split-plot ANOVA was used
with CII as the sub-plot term and soil as the main
plot term. Comparisons of CII measures were made
by using average values within each plot (plot-level
replication). Comparisons among species were
1282 S. M. Gleason and others

made using a randomized complete block design
(RCB), using average plot values for each species as
replicates (plot-level replication) and blocking by
plot. Thus, for species comparisons, the maximum
number of replicates equals the total number of
large temporary plots (24). For comparisons of
schist specialists versus generalists, only schist plots
were used (maximum number of replicates is 9).
Interactions between species (or CII) and soil were
tested using the split-plot design (a = 0.01). All soil
comparisons and soil interactions were tested using
only species commonly associated with both soil
types. Cyclone severity, as described above, was
used as a covariate in all analyses, although it rarely
affected ANOVA outcomes.
Leaf area index, ANPP, and RGR were only
measured on small permanent plots. Leaf area index
was compared between soils using a split-plot AN-
OVA, using soil as the sub-plot term and month as
the mainplot term (repeated observation) and
blocking by ridge. Plot-level ANPP was compared
between soils using an RCB ANOVA, blocking by
ridge. Tree-level RGR comparisons were made using
an RCB ANOVA, blocking by ridge and using the
competition index of Daniels (1976) and original
biomass (LN-transformed) as covariates. Original
biomass (LN-transformed) was used as a covariate
because it was negatively, but linearly, related to
RGR (RGR transformation did not account for all
size-dependant growth). Tree-level RGR compari-
sons were made for each 10-cm DBH class (Clark
and Clark 1999) and for a combined analysis of all
stems. We note that all RGR and ANPP comparisons
use biomass increment data (see the ANPP and RGR
methods section) collected using dendrometer
bands on the small permanent plots.
RESULTS
Cyclone Influence on Forest Structure
Data presented below are from large temporary plots
(0.25 ha), unless otherwise indicated. Most wind
effects were confined to leaves and branches, al-
though stem breakage and mortality were common
(Figure 2). Leaf stripping was significantly
(P = 0.021) higher on basalt than schist soils, but
only when schist-specializing species (species only
found on schist soils) were included in the ANOVA,
suggesting that the ‘‘soil effect’’ is likely to be a spe-
cies effect—schist specialists had less leaf stripping
than soil generalists. However, branch breakage,
top-break, and total stem breakage were signifi-
cantly higher (P = 0.015, 0.004, 0.008, respectively)
on basalt soils even when schist specialists were re-
moved from the analysis. Wind effects did not differ
significantly (a = 0.05) among slope or aspect mea-
surements, but this likely reflects similar topography
and aspect among plots. Ridge sites were more
modified than plots in flatter areas (within soil type),
but this result was also non-significant (a = 0.05).
Leaf stripping increased as relative tree height
(CII) decreased (P = 0.009), whereas branch
breakage showed the opposite pattern (P = 0.002)
(Figure 3). Additionally, top-break (trees with the
top section of the bole broken) and total stem
breakage increased significantly (a = 0.05) as rela-
tive tree height decreased. Wind effects were sim-
ilar for both soil types except for branch breakage,
which increased slightly from CII 1 to 2, but only
on basalt (P = 0.029).
To assess the overall effect of Cyclone Larry on
forest structure, diameter distributions were
graphed for the small permanent plots (Figure 4).
Canopy loss Stem breakage
leaf stripping
b
ranch breakage
t
op-break
mid-break
mortality
% Leaf area reduction
0
10
20
30
40
% Trees with stem breakage/mortality
0
2
4
6
8
10
12
basalt
schist
base-break
windthrow
Figure 2. Canopy loss and
stem breakage on basalt
and schist soils. Data were
obtained on large (0.25 ha)
temporary plots. Error bars
represent 1 standard error
of the mean (n = 15 plots
on basalt, n = 9 plots on
schist).
Cyclone Effects on a Tropical Rainforest 1283

Diameter histograms showed an ‘‘inverse-J’’, or
negative exponential distribution, before and after
the cyclone (Figure 4).
Leaf area index was reduced by 31% on basalt
and 13% on schist following the cyclone (Fig-
ure 5). Although LAI was significantly (P = 0.003)
higher on schist 2 months after the cyclone (May
2006), LAI values were similar between soils
5 months after the cyclone and were increasing
(Figure 5). Eight months after Cyclone Larry, LAI
had returned to approximately 80 and 93% of pre-
cyclone LAI values on basalt and schist, respec-
tively (Figure 5).
Cyclone Influence on Forest Production
Results in this section, unless otherwise indicated,
are from the small permanent plots. Cyclone Larry
generated 284 and 253 g m
-2
of litter (leaves +
wood), representing approximately 45 and 41% of
a year’s undisturbed litter production on basalt and
schist soils, respectively. Four months after Cyclone
Larry litterfall values were 27 and 40% of pre-cy-
clone litterfall values on basalt and schist, respec-
Canopy loss Stem breaka
g
e
l
e
a
f
st
r
i
p
p
i
n
g
b
r
a
n
c
h
b
r
e
a
k
a
g
e
t
o
p
-
b
r
e
a
k
m
i
d
-
b
r
e
a
k
b
a
s
e
-
b
r
e
a
k
windth
row
% Leaf area reduction
0
10
20
30
40
% Trees with stem breakage/mortality
0
2
4
6
8
CII 1
CII 2
CII 3
Figure 3. Canopy loss and
stem breakage by crown
illumination index (CII)
class. Dominant,
intermediate, and
understory trees were
assigned CII classes of 1, 2,
and 3, respectively. Data
were obtained on large
(0.25 ha) temporary plots.
Error bars represent 1
standard error of the mean
(n = 15 plots on basalt,
n = 9 plots on schist).
Forests on basalt
DBH class (cm)
0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90
Trees ha
-1
200
400
600
800
1000
1200
1400
pre-cyclone
post-cyclone
Forests on schist
DBH class (cm)
0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90
200
400
600
800
1000
1200
1400
Figure 4. Diameter
distribution of trees in small
(100 m
2
) permanent plots
before and after Cyclone
Larry. Error bars represent 1
standard error of the mean
(n = 12 plots).
Jan/2006 Mar/2006 May/2006 Ju
l/2006 Sep/2006 Nov/2006
Leaf Area Index (m
2
m
-2
)
4.0
4.5
5.0
5.5
6.0
6.5
7.0
schist
basalt
Cyclone Larry
Figure 5. Leaf area index (m
2
leaf area per m
2
ground
area) within small (100 m
2
) permanent plots before and
after Cyclone Larry. Error bars represent 1 standard error
of the mean (n = 12 plots).
1284 S. M. Gleason and others

tively, although differences between soil types
were not significant. Eight months after Cyclone
Larry litterfall values were 91.0 and 90.1% of pre-
cyclone litterfall values on basalt and schist,
respectively.
Litterfall, stem biomass increment, and ANPP did
not differ significantly (a = 0.05) between soil
types before or after the cyclone (Figure 6), nor did
they differ between fertilized and non-fertilized
plots. Relative growth rates of banded stems were
significantly (P < 0.001) higher on basalt soils
than schist soils when corrected for competition
(Figure 6). Relative growth rates of trees on basalt
increased faster than trees on schist following the
cyclone (P = 0.047) (Figure 6), largely attributable
to the release of small understory trees. The com-
petition index (Daniels 1976) accounted for sig-
nificant amounts of variance in RGR between soils
(P = 0.006), but did not affect the outcomes of
hypothesis tests. Relative growth rates were similar
between trees in fertilized plots and non-fertilized
plots. This is not unusual in P-limited forest sys-
tems, as it is unlikely for ANPP to respond to fer-
tilization in less than 18 months (Herbert and
Fownes 1995; Harrington and others 2001). Addi-
tionally, P-sensitivity (that is, decreased production
in response to P fertilization) may also have con-
tributed to a lack of P response in some species
(Gleason and others unpublished manuscript).
Wind Effects and Biomass Production
Among Species
Partitioning variance in leaf stripping and branch
breakage within a fully nested ANOVA (Pagel and
Harvey 1988; Osunkoya 1996) suggested that spe-
cies was the best taxonomic level for comparative
analyses. Family, genus, and species explained
48.8, 4.9, and 56.3% of leaf stripping variance and
14.5, 2.8, and 82.3% of branch breakage variance.
Species-specific N and P concentrations were
positively correlated with leaf area loss (Figure 7).
Production & RGR
6-18 mo after cyclone
RGR (kg kg
-1
yr
-1
)
0.005
0.010
0.015
0.020
0.025
Production & RGR
0-6 mo after cyclone
Production & RGR
pre-cyclone
litterfall
biomass
ANPP
RGR
Production (g m
-2
yr
-1
)
500
1000
1500
2000
2500
basalt
schist
litterfall
biomass
ANPP
RGR
biomass
RGR
Figure 6. Litterfall, biomass increment, ANPP, and RGR
values on schist and basalt before and after Cyclone Larry
(March 20th). Litterfall and ANPP values are not shown
for the 6–18 month period following Cyclone Larry be-
cause litterfall was not collected for this period. Error bars
represent 1 standard error of the mean (n = 12 plots).
2
4
6
8
10
12
14
basalt
schist
schist specialists
basalt (r
2
= 0.51; P < 0.000)
schist (r
2
= 0.08; P = 0.148)
Green leaf P (mg g
-1
)
0.4 0.6 0.8 1.0 1.2 1.4
% Canopy (leaf area) loss
2
4
6
8
10
12
low growth rate
medium growth rate
high growth rate
basalt
schist
schist specialists
basalt (r
2
= 0.42; P = 0.000)
schist (r
2
= 0.15; P = 0.047)
Green leaf N (mg g
-1
)
10 15 20 25
low growth rate
medium growth rate
high growth rate
A
D
C
B
Figure 7. Upper panels (A,
B) show the correlations
between green leaf N and P
and % canopy loss. Lower
panels (C, D) show the
wide variability among
growth rate groups (high,
medium, low) within each
of these correlations. Each
point represents the
average value for one
species on one soil type.
Correlation coefficients and
P values are given for basalt
and schist soils separately.
Cyclone Effects on a Tropical Rainforest 1285

One scenario could be that overstory trees fell on
understory trees that had higher P and N. However,
the observed relationship between wind effects and
leaf nutrients was independent of soil type, canopy
position, DBH class, growth habit (understory ver-
sus overstory species), and light requirements.
Thus, this relationship does not appear to be an
artifact of growth habit or stand structure.
Leaf stripping, branch breakage (Figure 8), and
mortality varied considerably among species
(P < 0.001). Wood density was not significantly
(a = 0.05) correlated with any wind effects mea-
sure. We also note the considerable variability in
leaf stripping and branch breakage among species
(Figure 8).
Wind resistant species (10 most resistant) and
susceptible species (10 least resistant) were included
in separate analyses. A comparison of the total
number of observations across all plots showed that
resistant species were significantly more common
than susceptible species (P = 0.037). Of the 10
resistant species, five were found only on schist soils
(schist specialists). Also, schist specialists had sig-
nificantly less leaf stripping (P = 0.008) and total leaf
area loss (leaf stripping + branch breakage)
(P = 0.024) than soil generalists. Schist specialists
also had lower leaf N concentrations (P = 0.011),
lower leaf P concentrations (P = 0.008), higher leaf
C:N ratios (P = 0.008), and higher leaf C:P ratios
(P = 0.002) than generalists.
Species-specific diameter increment (as esti-
mated using Vanclay 1991) was not correlated with
leaf area loss on either soil type, although com-
bined analysis of species on both soils yielded a
weak negative correlation between branch break-
age and diameter increment (r
2
= 0.11; P = 0.009).
Shade tolerance (from Osunkoya 1996) within
large plots was positively correlated (r
2
= 0.38;
P = 0.011) with leaf stripping—shade tolerant spe-
cies experienced more leaf stripping than light-
demanding species. However, branch breakage was
not correlated (P < 0.05) with species-specific light
requirements.
DISCUSSION
Resistance—Shade Tolerance Tradeoff
Plants vary in their capacity to resist wind (resis-
tance), recover from injury (resilience), and offset
the effects of injury (tolerance). We define resistant
species as species that experience few effects from
wind disturbance. We define resilient species as
species that undergo fast rates of return to a refer-
ence state after a disturbance has occurred (modi-
fied from Grimm and others 1992). We define
tolerant species as species that experience signifi-
cant wind effects, display relatively little resilience,
but are able to maintain fitness through other
means (other than fast regrowth or resistance),
which may include (but is not limited to) enhanced
seed production (quantity, size, recalcitrance),
herbivore resistance, or improved fertilization and
propagule dispersal mechanisms.
These ‘‘strategies’’ of resistance, resilience, and
tolerance may be reflected in structural, morpho-
logical, and physiological traits. Thus, tradeoffs
between plant traits in highly disturbed environ-
ments are important and common (Read and
Stokes 2006). For example, is it better for a tree to
be fast growing and weak or slow growing and
strong? This is a commonly studied tradeoff, espe-
cially when put in a successional context: light-
loving, nutrient-demanding, fast growing, quickly
maturing species are commonly pioneers. Faster-
growing species are sometimes affected more
markedly than slow-growing species (Foster 1988;
Bellingham and others 1995; Ostertag and others
2005).
Although there are data to support the resis-
tance-shade tolerance tradeoff model (Batista and
Platt 2003; Ostertag and others 2005), our data do
not support a similar model for tropical rainforest
species of northeast Queensland. Light require-
ments and growth rates were not positively corre-
lated with wind effects. This suggests that
mechanisms other than the resistance-shade tol-
% Leaf loss (leaf strip)
0 10203040506070
% Leaf loss (branch breakage)
0
20
40
60
80
100
species on schist
species on basalt
schist specialists
Figure 8. Relationship between leaf loss via petiole fail-
ure (leaf strip) and branch failure (branch breakage).
‘‘Species on schist’’ and ‘‘species on basalt’’ represent
species averages for soil generalists when they occurred
on schist or basalt, respectively. ‘‘Schist specialists’’ rep-
resent species that occurred only on schist soils in the
area of study.
1286 S. M. Gleason and others

erance tradeoff are responsible for much of the
variance in leaf stripping and branch breakage
among our species.
Nutrient availability adds complexity to resis-
tance-growth rate tradeoffs. Herbert and others
(1999) found that trees with access to P fertilizer
were less resistant to wind, but recovered more
quickly after injury (they were more resilient).
Thus, wind resistance/resilience may not be an
adaptive response, but simply a consequence of
within-plant nutrient economics, at least in some
cases.
Canopy loss was positively correlated with leaf N
and P for species on both soil types. Leaf area loss in
a Hawaiian rainforest was also positively associated
with nutrients (Herbert and others 1999). Why
should nutrient concentrations and canopy loss be
positively correlated? Major N-containing com-
pounds in plant leaves include amino acids,
nucleotides, enzymes, alkaloids, and pigments.
These compounds are not major components of
structural tissues such as lignin, cellulose, and
hemicellulose. Thus, high leaf nutrient contents
may be associated with reduced proportions of
structural tissues in plants.
Total C, lignin, and tannin concentrations are
often negatively correlated with rates of leaf decay
(Constantinides and Fownes 1994; Hobbie and
others 2006; Parton and others 2007) and con-
centrations of these compounds, relative to con-
centrations of N and P, are often negatively
correlated with soil N and P availability (Ares and
Gleason 2007). Therefore, the same properties that
make plants frugal with nutrients when nutrients
are scarce might also result in leaves and wood that
are resistant to wind and decomposition. Con-
versely, induced resistance (by exposing plants to
chronic wind) in Acer saccharum leaves and petioles
was related to leaf and petiole shape (Niklas 1996),
which are less likely to be affected by soil nutrient
supply. Thus, we note that leaf and petiole resis-
tance reflects not only structural strength, but also
petiole and leaf shape, length, flexibility, and
movement in high winds (Vogel 1989).
Scaling up from Species to Forests
Average traits among species do not necessarily
reflect forest-level traits. Schist specialists are few
compared to the overall species richness on schist
soils. Within the small permanent plots there are 61
species on schist soil, but only six schist specialists
(see supplementary material for species-level
information), yet these six species contribute
nearly 37% of the total basal area. Thus, the
properties of these few species, in large part, dom-
inate forest functioning on schist soils.
Scale is also important when interpreting
growth-rate versus resistance tradeoffs. Even
though there is no positive correlation between
growth rate and resistance among species, there is
at least one species (Acronychia acidula Mueller) that
does have growth and wind effects that are not in
conflict with growth-rate versus resistance theory.
Because this species contributes only 0.02% to total
stand basal area, it has little affect on stand-level
dynamics. It is this decoupling between species
traits and forest traits that emphasizes the impor-
tance of scale (species-level versus forest-level)
when evaluating the interaction between resource
availability and disturbance.
Other Strategies for Success
in Wind-Prone Tropical Forests
Variance among canopy loss measures was best
explained by substrate type and leaf N and P con-
centrations, suggesting that resource availability
influences wind resistance. These results support
the findings of Herbert and others (1999). This is
not to say that resistance-growth rate tradeoffs are
irrelevant, but that resource availability interacts
with disturbance regime so that other plant traits,
such as seed morphology and physiology, dispersal
mechanisms, and herbivore resistance, may also be
‘‘traded’’ for wind resistance. We suggest that
plants will ‘‘trade’’ resistance for proficiency in
these traits, depending on the nutrient ‘‘cost’’—the
demand relative to the availability of resources. For
example, a species that experiences structural
modification to leaves, branches, stems, or roots
may compensate by producing larger or more
plentiful seeds, leaves with higher carboxylation
efficiency, or simply complete their lifecycle in the
relative safety of the understory. All of these
strategies require greater nutrient uptake. Seed
strategies in particular may play an important role
in determining the fitness of shade tolerators in
wind-prone environments, as seed morphology
varies widely in these species (Grubb and Metcalfe
1996; Metcalfe and Grubb 1997; Metcalfe and
Turner 1998). When nutrients are limited, as on
the schist sites, strategies requiring high nutrient
uptake are costly and strategies requiring low
nutrient uptake (for example, resistance) are more
likely to enhance fitness.
A level of complexity not addressed by our con-
ceptual model of resource-driven wind resistance is
that wind intensity itself may affect species differ-
ently. During Cyclone Larry, many species of trees
Cyclone Effects on a Tropical Rainforest 1287

experienced more branch breakage relative to leaf
stripping, whereas others displayed mainly leaf
stripping (Figure 8). This may reflect an ‘‘all-or-
nothing’’ strategy whereby trees with resistant
leaves and/or petioles avoid injury during wind
events unless winds are strong enough to break
their branches, in which case they experience sig-
nificant canopy loss (that is, branches break with
leaves attached). Conversely, species with suscep-
tible petioles (petioles fail before branches) may
avoid branch breakage during severe winds, but
may lose canopy dominance to species with more
resistant leaves/petioles in less severe winds. It is
not clear whether or not these structural differ-
ences among species are adaptive, but they were
evident at our sites. We also note that if winds
exceed threshold values for all species, leaf area will
be lost regardless of branch or petiole traits.
CONCLUSION
Patterns of structural modification among species
and between soil types suggest nutrient availability
and use affect wind resistance. Disturbance regime
and resource availability are likely to interact,
engendering plant strategies that increase fitness
either through investments in C (resistance or C-
based herbivore resistance) or investments in soil
nutrients (reproductive, N-based herbivore de-
fenses, pioneer or understory specialization). Resis-
tance-shade tolerance tradeoffs were not evident,
although resistance was clearly related to soil type
and leaf nutrient concentrations. Although schist
specialists had higher C:N and C:P ratios and expe-
rienced fewer wind effects, this likely reflects strong
selection pressure toward efficient nutrient eco-
nomics, rather than adaptation toward wind resis-
tance. Whether wind resistance traits are adapted or
not, we might expect a change in species composi-
tion with increased cyclone frequency. If meaning-
ful species composition shifts are likely to take place,
as a result of predicted wind events (Emanuel 1987;
Webster and others 2005), we might be able to
predict these outcomes in advance by gathering site-
specific information on resource availability and
evaluating species-specific traits. Such knowledge
could benefit forestry, agroforestry, ecosystem
rehabilitation, and C accounting efforts in the near
future, particularly in wind-prone areas.
ACKNOWLEDGEMENTS
We would like to thank Kumi Gleason for her help
in the field and Andrew Ford for his help with tree
identification, leaf collecting, and rewarding dis-
cussions. We are grateful to Adrian Ares for his
advice on forest growth measurements, Dennis
O’Dowd for his critique of the overall study plan,
and Aiden Sudbury for statistical advice. We would
also like to thank three anonymous reviewers for
their helpful comments on an earlier version of this
manuscript. Two grants provided funding for this
research: The Holsworth Wildlife Research
Endowment, ANZ Charitable Trust, Australia, and
the Monash Small Grant Scheme, Monash Uni-
versity, Australia. This study was completed under
Queensland EPA permit no. WITK03219805.
REFERENCES
Aiba S, Kitayama K. 1999. Structure, composition and species
diversity in an altitude-substrate matrix of rain forest tree
communities on Mount Kinabalu, Borneo. Plant Ecol
140:139–157.
Ares A, Gleason SM. 2007. Foliar nutrient resorption in tree
species. In: Skaggs AK, Ed. New research on forest ecology.
New York, New York: Nova Science Publishers. p 1–32.
Asner GP, Goldstein G. 1997. Correlating stem biomechanical
properties of Hawaiian canopy trees with hurricane wind
damage. Biotropica 29:145–50.
Batista WB, Platt WJ. 2003. Tree population responses to hur-
ricane disturbance: syndromes in a south-eastern USA old-
growth forest. J Ecol 91:197–212.
Beard KH, Vogt KA, Vogt DJ, Scatena FN, Covich AP, Sigur-
dardottir R, Siccama TG, Crowl TA. 2005. Structural and
functional responses of a subtropical forest to 10 years of
hurricanes and droughts. Ecol Monogr 75:345–61.
Bellingham WB, Tanner EVJ, Healey JR. 1995. Damage and
responsiveness of Jamaican montane tree species after dis-
turbance by a hurricane. Ecology 76:2562–80.
Blood ER, Anderson P, Smith PA, Nybro C, Ginsberg KA. 1991.
Effects of Hurricane Hugo on coastal soil solution chemistry in
South Carolina. Biotropica 23:348–55.
Boucher DH, Vandermeer JH, Yih K, Zamora N. 1990. Con-
trasting hurricane damage in tropical rain forest and pine
forest. Ecology 71:2022–4.
Brown S. 1997. Estimating biomass and biomass change of
tropical forests: a primer. Rome: FAO Forestry Paper 134, p 55.
Canham CD, Papaik MJ, Latty EF. 2001. Interspecific variation
in susceptibility to windthrow as a function of tree size and
storm severity for northern temperate tree species. Can J For
Res 31:1–10.
Clark DA, Clark DB. 1992. Life history diversity of canopy and
emergent trees in a neotropical rainforest. Ecol Monogr
62:315–44.
Clark DA, Clark DB. 1999. Assessing the growth of tropical rain
forest trees: issues for forest modeling and management. Ecol
Appl 9:981–97.
Constantinides M, Fownes JH. 1994. Nitrogen mineralization
from leaves and litter of tropical plants; relationship to nitro-
gen, lignin and soluble polyphenol concentrations. Soil Biol
Biochem 26:49–55.
Cooper-Ellis S, Foster DR, Carlton G, Lezberg A. 1999. Forest
response to catastrophic wind: results from an experimental
hurricane. Ecology 80:2683–96.
1288 S. M. Gleason and others

Cordero RA, Fetcher N, Voltzow J. 2007. Effects of wind on the
allometry of two species of plants in an elfin cloud forest.
Biotropica 39:177–85.
Crews TE, Kitayama K, Fownes JH, Riley RH, Herbert DA,
Mueller-Dombois D, Vitousek PM. 1995. Changes in soil
phosphorus fractions and ecosystem dynamics across a long
chronosequence in Hawaii. Ecology 76:1407–24.
Crook MJ, Ennos AR, Banks JR. 1997. The function of buttress
roots: a comparative study of the anchorage systems of but-
tressed (Aglaia and Nephelium ramboutan species) and not-
buttressed (Mallotus wrayi) tropical trees. J Exp Bot 48:1703–
16.
Daniels RF. 1976. Simple competition indices and their corre-
lation with annual loblolly pine tree growth. For Sci 22:454–6.
Department of Primary Industries. 1998. Queensland Timbers:
their nomenclature, density and lyctid susceptibility. Depart-
ment of Primary Industries and Fisheries, Catalogue No.
2770000000215. p 126.
Emanuel KA. 1987. The dependence of hurricane intensity on
climate. Nature 326:483–5.
Ennos AR. 1993. The function and formation of buttresses.
Trends Ecol Evol 8:350–1.
Foster DR. 1988. Species and stand response to catastrophic
wind in central New England, USA. J Ecol 76:135–51.
Foster DR, Boose ER. 1992. Patterns of forest damage resulting
from catastrophic wind in central New England, USA. J Ecol
80:79–98.
Frangi JL, Lugo AE. 1991. Hurricane damage to a flood plain
forest in the Luquillo Mountains of Puerto Rico. Biotropica
23:324–35.
Fraser AI. 1962. The soil and roots as factors in tree stability.
Forestry 35:117–27.
Garcia-Montiel DC, Neill C, Melillo J, Thomas S, Steudler PA,
Cerri CC. 2000. Soil phosphorus transformations following
forest clearing for pasture in the Brazilian Amazon. Soil Sci
Soc Am J 64:1792–807.
Gresham CA, Williams TM, Lipscomb DJ. 1991. Hurricane Hugo
wind damage to southeastern U.S. coastal forest tree species.
Biotropica 23:420–6.
Grimm V, Schmidt E, Wissel C. 1992. On the application of
stability concepts in ecology. Ecol Modell 63:143–61.
Grubb PJ, Metcalfe DJ. 1996. Adaptation and inertia in the
Australian tropical lowland rain-forest flora: contradictory
trends in intergeneric and intrageneric comparisons of seed
size in relation to light demand. Funct Ecol 10:512–20.
Guzma
´
n-Grajales SM, Walker LR. 1991. Differential seedling
responses to litter after Hurricane Hugo in the Luquillo
Experimental Forest, Puerto Rico. Biotropica 23:407–13.
Harrington RA, Fownes JH, Vitousek PM. 2001. Production and
resource use efficiencies in N- and P-limited tropical forests: a
comparison of responses to long-term fertilization. Ecosystems
4:646–57.
Hedley MJ, White RE, Nye PH. 1982. Plant-induced changes in
the rhizosphere of rape (Brassica napus var. emerald) seedlings,
III. changes in L value, soil phosphate fractions and phos-
phatase activity. New Phytol 91:45–56.
Herbert DA, Fownes JH. 1995. Phosphorus limitation of forest
leaf area and net primary production on a highly weathered
soil. Biogeochemistry 29:223–35.
Herbert DA, Fownes JH, Vitousek PM. 1999. Hurricane damage
to a Hawaiian forest: nutrient supply rate affects resistance
and resilience. Ecology 80:908–20.
Hobbie SE, Reich PB, Oleksyn J, Ogdahl M, Zytkowiak R, Hale C,
Karolewski P. 2006. Tree species effects on decomposition and
forest floor dynamics in a common garden.Ecology 87:2288–97.
Holdridge LR. 1947. Determination of world plant formations
from simple climatic data. Science 105:367–8.
Hyland BPM, Whiffin T, Christophel DC, Gray B, Elick RW.
2002. Australian tropical rain forest plants: trees, shrubs and
vines. Interactive CD identification tool. Australia: CSIRO
Publication.
Keeland BD, Young PJ. 2005. Construction and installation of
dendrometer bands for periodic tree-growth measurements.
United States Geological Service webpage: http://www.nwrc.
usgs.gov/Dendrometer/index.htm.
Kerridge PC, Andrew CS, Murtha GG. 1972. Plant nutrient sta-
tus of soils of the Atherton Tableland, North Queensland. Aust
J Exp Agric Anim Husb 12:618–27.
Kitayama K, Aiba S. 2002. Ecosystem structure and productivity
of tropical rain forests along altitudinal gradients with con-
trasting soil phosphorus pools on Mount Kinabalu, Borneo. J
Ecol 90:37–51.
Lin KC, Hamburg SP, Tang SL, Hsia YJ, Lin TC. 2003. Typhoon
effects on litterfall in a subtropical forest. Can J For Res
33:2184–92.
Lodge DJ, McDowell WH. 1991. Summary of ecosystem-level
effects of Caribbean hurricanes. Biotropica 23:373–8.
Malcolm DT, Nagel BKA, Sinclair I, Heiner IJ. 1999. Soils and
agricultural land suitability of the Atherton Tablelands, North
Queensland. Queensland: Land Resources Bulletin, Depart-
ment of Natural Resources, p 114–8.
McDowell WH, McSwiney CP, Bowden WB. 1996. Effects of
hurricane disturbance on ground water chemistry and ripar-
ian function in a tropical rain forest. Biotropica 28:577–84.
Merrens EJ, Peart DR. 1992. Effects of hurricane damage on
individual growth and stand structure in a hardwood forest in
New Hampshire, USA. J Ecol 80:787–95.
Metcalfe DJ, Ford AJ. 2008. Floristic biodiversity in the wet
tropics. In: Stork N, Turton S, Eds. Living in a dynamic tropical
forest landscape. Oxford, UK: Blackwell. p 123–32.
Metcalfe DJ, Grubb PJ. 1997. The responses to shade of seedlings
of very small-seeded tree and shrub species from tropical rain
forest in Singapore. Funct Ecol 11:215–21.
Metcalfe DJ, Turner IM. 1998. Soil seed bank from lowland rain
forest in Singapore: canopy-gap and litter-gap demanders. J
Trop Ecol 14:103–8.
Niklas KJ. 1996. Differences between Acer saccharum leaves from
open and wind-protected sites. Ann Bot 78:61–6.
Niklas KJ. 1999. A mechanical perspective on foliage leaf form
and function. New Phytol 143:19–31.
Ostertag R, Silver WL, Lugo AE. 2005. Factors affecting mortality
and resistance to damage following hurricanes in a rehabili-
tated subtropical moist forest. Biotropica 37:16–24.
Osunkoya OO. 1996. Light requirements for regeneration in
tropical forest plants: taxon-level and ecological attribute ef-
fects. Aust J Ecol 21:429–41.
Pagel MD, Harvey PH. 1988. Recent development in the analysis
of comparative data. Q Rev Biol 63:413–40.
Parrotta JA, Lodge DJ. 1991. Fine root dynamics in a subtropical
wet forest following hurricane disturbance in Puerto Rico.
Biotropica 23:343–7.
Parton W, Silver WL, Burke IC, Grassens L, Harmon ME, Currie
WS, King JY, Adair C, Brandt LA, Hart SC, Fasth B. 2007.
Cyclone Effects on a Tropical Rainforest 1289

Global-scale similarities in nitrogen release patterns during
long-term decomposition. Science 315:361–4.
Read J, Stokes A. 2006. The ecology of plant biomechanics, from
individual to ecosystem. Am J Bot 93:1546–65.
Rechcigl JE, Payne JE. 1990. Comparison of a microwave
digestion system to other digestion methods for plant tissue
analysis. Commun Soil Sci Plant Anal 21:2209–18.
Sanford RL, Parton WJ, Ojima DS, Lodge DJ. 1991. Hurricane
effects on soil organic matter dynamics and forest production
in the Luquillo Experimental Forest, Puerto Rico: results of
simulation modelling. Biotropica 23:364–72.
Scatena FN, Silver W, Siccama T, Johnson A, Sa
´
nchez MJ. 1993.
Biomass and nutrient content of the Bisley Experimental
Watersheds, Luquillo Experimental Forest, Puerto Rico, be-
fore and after Hurricane Hugo, 1989. Biotropica 25:15–27.
Scatena FN, Moya S, Estrada C, Chinea JD. 1996. The fist five
years in the reorganisation of aboveground biomass and
nutrient use following Hurricane Hugo in the Bisley Experi-
mental Watersheds, Luquillo Experimental Forest, Puerto
Rico. Biotropica 28:242–440.
Schoenau JJ, Stewart JWB, Bettany JR. 1989. Forms and cycling
of phosphorus in prairie and boreal forest soils. Biogeochem-
istry 8:223–37.
Sherman RE, Fahey TJ, Martinez P. 2001. Hurricane impacts on
a mangrove forest in the Dominican Republic: damage pat-
terns and early recovery. Biotropica 33:393–408.
Silver WL, Vogt KA. 1993. Fine root dynamics following single
and multiple disturbances in a subtropical wet forest ecosys-
tem. J Ecol 81:729–38.
Stokes A. 1999. Strain distribution during anchorage failure of
Pinus pinaster Ait. at different ages and tree growth response to
wind-induced root movement. Plant Soil 217:17–27.
Tiessen H, Moir JO. 1993. Characterization of available P by
sequential extraction. In: Carter MR, Ed. Soil sampling and
methods of analysis. Boca Raton: Lewis Publishers. pp 75–86.
Tracey JG. 1982. The vegetation of the humid tropical region of
North Queensland. Melbourne: CSIRO Publication, p 124.
Unwin GL, Applegate GB, Stocker GC, Nicholson DI. 1988. Ini-
tial effects of Tropical Cyclone Winifred on forests in north
Queensland. Proc Ecol Soc Aust 15:283–96.
Uriarte M, Canham CD, Thompson J, Zimmerman JK, Brokaw
N. 2005. Seedling recruitment in a hurricane-driven tropical
forest: light limitation, density-dependence and the spatial
distribution of parent trees. J Ecol 93:291–304.
Vanclay JK. 1991. Aggregation tree species to develop diameter
increment equations for tropical rainforests. For Ecol Manage
42:143–68.
Vitousek P. 2004. Nutrient cycling and limitation, Hawaii as a
model system. Princeton, USA: Princeton University Press, p
223.
Vogel S. 1989. Drag and reconfiguration of broad leaves in high
winds. J Exp Bot 40:941–8.
Walker TW, Syers JK. 1976. The fate of phosphorus during
pedogenesis. Geoderma 15:1–19.
Webb LJ. 1958. Cyclones as an ecological factor in tropical
lowland rainforest, north Queensland. Aust J Bot 6:220–8.
Webster PJ, Holland GJ, Curry JA, Chang H-R. 2005. Changes in
tropical cyclone number, duration, and intensity in a warming
environment. Science 309:1844–6.
Whigham DF, Olmsted I, Cano EC, Harmon ME. 1991. The
impact of Hurricane Gilbert on trees, litterfall, and woody
debris in a dry tropical forest in the northeastern Yucatan
Peninsula. Biotropica 23:434–41.
Yamada I. 1997. Tropical rain forests of Southeast Asia. Hono-
lulu: University of Hawaii Press, p 438.
Yamakura T, Hagihara A, Sukardjo S, Ogawa H. 1986. Above-
ground biomass of tropical rain forest stands in Indonesian
Borneo. Vegetatio 68:71–82.
You C, Petty WH. 1991. Effects of Hurricane Hugo on Manilkara
bidentata, a primary tree species in the Luquillo Experimental
Forest of Puerto Rico. Biotropica 23:400–6.
1290 S. M. Gleason and others