Leaf flushing during the dry season: The paradox of Asian monsoon forests

Abstract

Aim Most deciduous species of dry monsoon forests in Thailand and India form new leaves 1–2 months before the first monsoon rains, during the hottest and driest part of the year around the spring equinox. Here we identify the proximate causes of this characteristic and counterintuitive ‘spring‐flushing’ of monsoon forest trees.
Location Trees of 20 species were observed in semi‐deciduous dry monsoon forests of northern Thailand with a 5–6‐month‐long severe dry season and annual rainfall of 800–1500 mm. They were growing on dry ridges (dipterocarp–oak forest) or in moist gullies (mixed deciduous–evergreen forest) at 680–750 m altitude near Chiang Mai and in a dry lowland stand of Shorea siamensis in Uthai Thani province.
Methods Two novel methods were developed to analyse temporal and spatial variation in vegetative dry‐season phenology indicative of differences in root access to subsoil water reserves.
Results Evergreen and leaf exchanging species at cool, moist sites leafed soon after partial leaf shedding in January–February. Drought‐resistant dipterocarp species were evergreen at moist sites, deciduous at dry sites, and trees leafed soon after leaf shedding whenever subsoil water was available. Synchronous spring flushing of deciduous species around the spring equinox, as induced by increasing daylength, was common in Thailand's dipterocarp–oak forest and appears to be prevalent in Indian dry monsoon forests of the Deccan peninsula with its deep, water‐storing soils.
Main conclusions In all observed species leafing during the dry season relied on subsoil water reserves, which buffer trees against prolonged climatic drought. Implicitly, rainfall periodicity, i.e. climate, is not the principal determinant of vegetative tree phenology. The establishment of new foliage before the summer rains is likely to optimize photosynthetic gain in dry monsoon forests with a relatively short, wet growing season.

Full text

© 2006 The Authors DOI: 10.1111/j.1466-822x.2006.00213.x
Journal compilation © 2006 Blackwell Publishing Ltd www.blackwellpublishing.com/geb 1
Global Ecology and Biogeography, (Global Ecol. Biogeogr.) (2006)
RESEARCH
PAPER
Blackwell Publishing Ltd
Leaf flushing during the dry season:
the paradox of Asian monsoon forests
Stephen Elliott1, Patrick J. Baker2 and Rolf Borchert3*
ABSTRACT
Aim Most deciduous species of dry monsoon forests in Thailand and India form
new leaves 1–2 months before the first monsoon rains, during the hottest and driest
part of the year around the spring equinox. Here we identify the proximate causes of
this characteristic and counterintuitive ‘spring-flushing’ of monsoon forest trees.
Location Trees of 20 species were observed in semi-deciduous dry monsoon forests
of northern Thailand with a 5 –6-month-long severe dry season and annual rainfall
of 800–1500 mm. They were growing on dry ridges (dipterocarp–oak forest) or
in moist gullies (mixed deciduous–evergreen forest) at 680–750 m altitude near
Chiang Mai and in a dry lowland stand of Shorea siamensis in Uthai Thani province.
Methods Two novel methods were developed to analyse temporal and spatial
variation in vegetative dry-season phenology indicative of differences in root access
to subsoil water reserves.
Results Evergreen and leaf exchanging species at cool, moist sites leafed soon after
partial leaf shedding in January–February. Drought-resistant dipterocarp species
were evergreen at moist sites, deciduous at dry sites, and trees leafed soon after leaf
shedding whenever subsoil water was available. Synchronous spring flushing of
deciduous species around the spring equinox, as induced by increasing daylength,
was common in Thailand’s dipterocarp–oak forest and appears to be prevalent in
Indian dry monsoon forests of the Deccan peninsula with its deep, water-storing soils.
Main conclusions In all observed species leafing during the dry season relied on
subsoil water reserves, which buffer trees against prolonged climatic drought.
Implicitly, rainfall periodicity, i.e. climate, is not the principal determinant of vege-
tative tree phenology. The establishment of new foliage before the summer rains is
likely to optimize photosynthetic gain in dry monsoon forests with a relatively short,
wet growing season.
Keywords
Deciduousness, dipterocarp–oak forest, photoperiodic control, spring-flushing,
tropical dry forests, tropical tree phenology.
*Correspondence: Rolf Borchert, Division of
Biological Sciences, University of Kansas,
Lawrence, KS 66045–7534, USA.
E-mail: borchert@ku.edu
1Biology Department, Science Faculty, Chiang
Mai University, Chiang Mai, 502001, Thailand,
2School of Biological Sciences, Monash
University, Victoria 3800, Australia,
3Division of Biological Sciences, University of
Kansas, Lawrence, KS 66045, USA
INTRODUCTION
Climate exerts a dominant control over the distribution of major
vegetation types (Woodward, 1987). It should also determine the
characteristic vegetative phenology of major forest types often
used as indicators of climate (e.g. Koeppen’s classification of
climate). Deciduous, semi-deciduous and evergreen tropical forests
are considered to be indicators of the amount and annual dis-
tribution of rainfall (Walter, 1971) because seasonal variation in
tree water status constitutes a major determinant of tropical tree
phenology (Borchert, 1994a; Borchert et al., 2002). Severe water
stress enhances the abscission of old leaves and prevents the
expansion of new shoots and leaves. Increasing duration and
severity of the dry season should therefore result in trees being
leafless for progressively longer periods and, inversely, the dura-
tion of deciduousness among trees in a landscape should be an
indicator of the duration of severe drought. In common usage,
the term ‘deciduous’ is applied indiscriminately to tropical tree

S. Elliott et al.
© 2006 The Authors
2Global Ecology and Biogeography, Journal compilation © 2006 Blackwell Publishing Ltd
species being leafless for just a few weeks or as long as 4–
6 months. To enhance the value of deciduousness as a quantit-
ative indicator of seasonal tree water stress, the normal duration
of deciduousness in a species will be given here (e.g. 1–2-month
deciduous or 3–5-month deciduous).
The dry monsoon forests of Thailand (deciduous dipterocarp
forest, mixed deciduous-evergreen forests; Rundel & Boonpragob,
1995; Maxwell & Elliott, 2001) and the Indian semi-deciduous
forests of the Deccan peninsula (Troup, 1921; Puri, 1969) compared
here differ widely in species composition, but share two charac-
teristic features. First, both receive > 90% of annual rainfall of
800 –1500 mm during the 5-month-long monsoon season between
June and October (Fig. 1). The following 6–7-month-long dry
season is subdivided into a cool ‘winter’ season with relatively
low temperatures and a hot ‘spring’ season with rapidly rising
temperatures (Puri, 1969; Walter, 1971; Rundel & Boonpragob,
1995). Secondly, many deciduous species shed their leaves as late
as February or March and, counter-intuitively, most leaf between
March and May, during the hottest and driest period of the year,
1–2 months before the first monsoon rains (Fig. 1; Troup, 1921;
Walter, 1971; Sukwong et al., 1975; Prasad & Hegde, 1986; Bhat,
1992; Rundel & Boonpragob, 1995; Kushwaha & Singh, 2005).
Other species, including the common, wide-ranging Shorea
robusta (sal), exchange old for new leaves between January and
March (Troup, 1921; Rundel & Boonpragob, 1995; Kushwaha &
Singh, 2005). Consequently, leaf cover in Asian monsoon forests
is maintained well beyond the duration of the rainy season.
If amount and seasonal distribution of rainfall were the major
determinants of vegetative phenology, then most trees of Asian
monsoon forests should shed their leaves during the early dry
season, stand leafless for several months and leaf after the first
monsoon rains, i.e. they should be 3–5-month deciduous, as
observed in many species of Neotropical forests with a similar
climate (Frankie et al., 1974; Bullock & Solis-Magallanes, 1990;
Borchert, 1994a). The paradox leaf flushing of monsoon forest
trees during the late dry season raises several questions. (1) How
can leafless, presumably water-stressed trees rehydrate and leaf
during the dry season without rehydration of the topsoil by rain?
(2) Which environmental trigger causes bud break between
March and May, well before the arrival of the monsoon rains?
(3) Is the phenology of monsoon forest trees indeed determined
mainly by rainfall periodicity and hence predictable from cli-
matic data — as suggested by Walter (1971) and others?
The principal internal and environmental controls of major
patterns of vegetative phenology in seasonally dry Neotropical
forests have been identified in recent eco-physiological studies
(Fig. 2; Borchert, 2005). In general, increasing water stress
during the early dry season causes abscission of ageing leaves
(Borchert et al., 2002). Leaf abscission and a high water potential
of twigs are prerequisites for subsequent bud break and leaf
expansion (Borchert, 1994a). Leaf flushing at different times
during and after the dry season is caused by three different
mechanisms characteristic of different ‘functional types’ to be
described below (Fig. 2b–d). In this study, environmental con-
trol of vegetative phenology in monsoon forest species will be
inferred from the observed timing, synchrony and inter-annual
variation of leafing assessed by a novel method (Methods; Rivera
et al., 2002).
The amount and seasonality of rainfall in dry monsoon forests
near Chiang Mai in northern Thailand are similar to the rainfall
pattern of a semi-deciduous forest analysed in Costa Rica (Fig. 2a;
Borchert, 1994a, 2002). To compare the controls of vegetative
phenology in these Asian and Neotropical monsoon forests and
address the questions raised above, we analysed the phenological
records of 19 species observed near Chiang Mai, and the varia-
tion of phenology with tree size and subsoil water availability
among 123 Shorea siamensis trees in another Thai forest. We
found the characteristic spring flushing of many deciduous
monsoon forests species to be induced by increasing daylength
and dependent on root access to subsoil water reserves.
TREE SPECIES AND FIELD SITES
Doi-Suthep National Park (DSNP), Chiang Mai,
Thailand
The phenology of 128 trees representing 19 species (Table 1) was
monitored along a transect at 650–780 m altitude in the Doi-
Suthep National Park near Chiang Mai in northern Thailand
(19° N, 99° E; Elliott et al., 1989; 1994). Tree species will be
referred to by generic name except for the two genera with more
than one species (Dipterocarpus, Shorea). The site is heavily dis-
sected into steep gullies with mixed deciduous–evergreen forest
and narrow, well-drained ridges with deciduous dipterocarp–
oak forest (Rundel & Boonpragob, 1995; Maxwell & Elliott,
2001). Base rocks are mostly granitic and soils are generally deep
and highly weathered. The area has a typical monsoon climate
(Figs 1 and 2). Mean annual rainfall was 1120 mm during the
study period. In the observed transect, temperatures are 2–3 °C
lower and rainfall is probably 15–20% higher than at Chiang Mai
Airport (312 m altitude, 4.5 km from the study site), where
climate was recorded (Elliott, pers. obs.). For measurement of soil
water content, triplicate soil samples were collected at 0–20 cm
depth during each monthly phenology observation and later
dried for 2 days at 120 °C.
Huai Kha Khaeng Wildlife Sanctuary (HKK), Uthai
Thani Province, Thailand
The vegetative phenology of 123 Shorea siamensis Miqu.var.
siamensis trees was observed in a nearly mono-specific 50 × 50-m
plot in the Huai Kha Khaeng Wildlife Sanctuary, Uthai Thani
Province in central Thailand (15° N, 100° E). The plot slopes
slightly from 42 to 50 m elevation and an ephemeral stream bed
cuts through its upper half (see Fig. 7 in Results below). Rainfall
was measured at the Kapook Kapiang Ranger Station 4 km from
the plot.
Functional tree types (Borchert, 2005)
The following patterns of vegetative phenology (functional
types) were identified in Neotropical and Asian monsoon forests.

Dry season leaf flushing
© 2006 The Authors
Global Ecology and Biogeography, Journal compilation © 2006 Blackwell Publishing Ltd 3
Leaf-exchanging species are restricted to moist sites and remain
well hydrated during the dry season. Shedding of old leaves dur-
ing the early dry season is accompanied or immediately followed
by bud break and expansion of new leaves (Fig. 2d, January/
February, arrow; Williams et al., 1997). As in Shorea robusta,
leaf exchange generally occurs during the early dry season, but
its timing varies widely among conspecific trees with soil water
availability and between years with the time of the last major
rains of the wet season (Rivera et al., 2002; Borchert et al., 2005;
Singh & Kushwaha, 2005).
Deciduous species dehydrate strongly during the early dry sea-
son and remain leafless for 3–5 months (3–5-month deciduous).
The first wet season rainfalls of > 20–30 mm result in rapid re-
hydration and synchronous bud break of all conspecific trees at a
microsite (Fig. 2b, May/June, arrow; Borchert, 1994b). The vari-
ability and patchiness of the rainfall causes large variation in the
timing of leafing between years and among trees at different
microsites in a landscape. The opportunistic phenology of leaf-
exchanging and 3–5-month deciduous species is determined
mainly by seasonal variation in tree water status at a given site
and trees will leaf whenever they are leafless and fully hydrated.
Leafing of spring-flushing species around the spring equinox,
well before the first monsoon rains, is induced by increasing
daylength (Fig. 2c, March/April arrow; Rivera et al., 2002). It is
Table 1 Tree species of Doi Suthep National Park, Chiang Mai, Thailand, sorted by functional type (see Fig. 2; Results). Habitats: do, deciduous
dipterocarp–oak forest; md, mixed deciduous–evergreen forest (Elliott et al., 1994; Maxwell & Elliott, 2001). Leaf phenology at lower elevations:
d, deciduous; e, evergreen
Species Family
Number
of trees Habitat
Leaf
phenology Fig.
Deciduous, rain-induced
Antidesma acidum Retz. Euphorbiaceae 5 md d 3, 4a, 5
Spring-flushing
Colona flagrocarpa (Cl.) Craib Tiliaceae 5 do, md d 4b, 5
Dalbergia cultrata Grah. ex Bth. Fabaceae 6 do, md d 4d, 5
Quercus kerrii Craib var. kerrii Fagaceae 10 do d 4e, 5
Shorea obtusa Wal l. e x Bl. Dipterocarpaceae 6 do d 4f, 5
Terminalia mucronata Craib & Hutch. Combretaceae 6 md d 4c, 5
Irregularly leaf-exchanging
Craibiodendron stellatum (Pierre) W.W. Sm. Ericaceae 5 do d 4j, 5
Dipterocarpus obtusifolius Teijsm. ex Miq. Dipterocarpaceae 12 do d 4k, 5
Dipterocarpus tuberculatus Roxb. Dipterocarpaceae 6 do d 5
Leaf-exchanging
Anneslea fragrans Wall. Theaceae 7 do e 4i, 5
Aporusa villosa (Lindl.) Baill. Euphorbiaceae 7 do d 5
Eugenia albiflora Duth. ex. Kurz Myrtaceae 5 md e 4h, 5
Metadina trichotoma (Zoll. & Mor.) Bakh.f. Rubiaceae 7 md d 5
Tristaniopsis burmanica (Griff.) Wils. & Watt. Myrtaceae 10 do e 4g, 5
Evergreen
Castanopsis diversifolia King ex. Hk.f. Fagaceae 6 md e
Ilex umbellulata (Wall.) Loesn. Aquifoliaceae 5 md e
Lithocarpus sootepensis (Craib) A. Camus Fagaceae 6 md e
Rothmannia sootepensis (Craib) Brem. Rubiaceae 5 md e 5
Wendlandia tinctoria (Roxb.) D.C. Rubiaceae 9 do, md e 4l, 5
Figure 1 Rainfall periodicity and tree
phenology in Indian dry monsoon forests.
Information on the beginning of leaf
flushing (horizontal hatching) and leaf
abscission (diagonal hatching) was obtained
for 81 species of the Deccan peninsula from
descriptions of phenology in Troup (1921).
Rainfall (black bars) and temperature
(curve) are for Varanasi, India (Kushwaha &
Singh, 2005).

S. Elliott et al.
© 2006 The Authors
4Global Ecology and Biogeography, Journal compilation © 2006 Blackwell Publishing Ltd
highly synchronous among all conspecific trees in a landscape
and varies minimally between years.
METHODS
Analysis of phenological patterns observed at DSNP
Tree phenology in DSNP was observed monthly from December
1988 to December 1991. Each of four stages of vegetative pheno-
logy (leafless; young, light-green leaves; mature, dark-green
leaves; senescent, brown leaves) was scored on a scale of 0– 4 (0,
phenological stage not observed; 4, phenological stage at its
maximum; Elliott et al., 1994) and recorded in a spreadsheet.
Species-specific timing, synchrony and inter-annual variation
in leaf flushing among conspecific trees are essential for the iden-
tification of functional types (Rivera et al., 2002) and for inter-
specific comparisons of vegetative phenology. To identify these
variables, the recorded phenological observations were analysed
as follows. The sequence of phenophases during three observa-
tion years was obtained for each tree by sorting phenological
records by tree number and date, subdividing them into four-
column data sets and then transposing them into rows (Fig. 3a).
Missing phenophases were inferred from the timing of the next
phase (e.g. flushing of young leaves precedes the increase in
mature leaves) and added to the record. Mean monthly phenology
scores for all conspecific trees were calculated to characterize
synchrony of leafing within the same year (Fig. 3b). For the
evaluation of interannual variation, 3-year means of phenology
scores were calculated (Fig. 3c) and graphed for each species
(Fig. 4a).
Synchronous leaf exchange generally lasted less than 2–
3 months and several species were leafless or had young leaves
for only 2–3 weeks. At monthly observation frequency these
phenophases were observed only once or not at all. Phenological
patterns are therefore characterized more by the time sequence
of phenophases than by changing values of phenology scores
(Figs 3b,c and 4). Distinct minima of mature leaf scores
accompanied by peaks for leaflessness and young leaves indicate
deciduousness and strong synchrony of leaf flushing among
conspecific trees (Fig. 4a–f ). Consistently low scores for young
leaves and leaflessness may indicate very brief periods of partial
deciduousness (Fig. 4l) or large variation in deciduousness
among trees and between years (Figs 4j,k and 5, Dipterocarpus
tuberculatus; Fig. 6). Inter-specific variation in deciduousness
(Fig. 5) was quantified by counting months without leaves and
with full leaf cover (mature leaf scores of < 1 and > 3, respec-
tively) in the records of conspecific trees (Fig. 3a).
Phenology at HKK as a function of tree size and
topography
Diameter at breast height (d.b.h.), location (xt, yt) and elevation
were measured for 123 Shorea siamensis trees of a 50 × 50-m
experimental plot in HKK. During leafing in spring 1998 and
leaf shedding in early 1999, vegetative phenology was scored as
percentage of crown fullness. The effects of tree size and topo-
graphy on leaf flushing before the first rains in 1998 were quantified
by calculating distances between each tree (xt, yt) and all points
along the creek (xc, yc) as d = [(xt − xc)2 + (yt – yc)2]−2 and selecting
the smallest d for each tree. Trees were sorted into three size
classes (6–15, 15–24, 24–56 cm d.b.h.) and classified according
to their vegetative phenology as ‘flushing’ (crown cover 20–
100%) or ‘dormant’ (crown cover 0%). Trees with a crown cover
of 10% were ignored. For each size class, means and standard
errors were calculated for the fractions of flushing vs. dormant
trees, distances from the creek bed and changes in leaf cover dur-
ing flushing and leaf shedding.
RESULTS
Synchronous leaf flushing of all Antidesma trees was triggered
by the increase in soil water after the first monsoon rains, as
indicated by the 1-month difference in leafing time between 1989
and 1990 (Fig. 3b, frames; high soil water in March 1990 indicates
Figure 2 Seasonal variation in rainfall and
daylength (a) in a Neotropical semi-deciduous
forest in Guanacaste, Costa Rica (10° N) and
in a dry monsoon forest near Chiang Mai,
Thailand (19° N). Vegetative phenology
(b–d) in Neotropical trees of three functional
types. Arrows indicate the causes of
vegetative bud break during the dry season:
(b) the first heavy rains of the wet season;
(c) increasing daylength around the spring
equinox; (d) drought-induced leaf shedding.

Dry season leaf flushing
© 2006 The Authors
Global Ecology and Biogeography, Journal compilation © 2006 Blackwell Publishing Ltd 5
rainfall not recorded at the Chiang Mai airport). Antidesma is the
only deciduous species with rain-induced leafing and a leafless
period of 3–5 months among the 19 tree species monitored at
DSNP (Figs 3–5).
At the altitude of the observed transect, deciduousness varies
widely among the five spring-flushing species (Figs 4b–f and 5),
which at lower elevations are 1–3-month deciduous. These spe-
cies leafed each year synchronously in March/April, well before
the first monsoon rains (Fig. 4b–f ). In the five leaf-exchanging
species, moderately synchronous leaf flushing in January/Febru-
ary was preceded and hence probably caused by the shedding of
many old leaves (Figs 4g–i and 5; Williams et al., 1997; Borchert,
2000). As in many Neotropical leaf-exchanging species, early
monsoon rains in May/June triggered a minor second leaf flush
(Fig. 4g,i, filled circle in May/June). In the five evergreen species
(Table 1), there was almost no discernable seasonality of vegeta-
tive development (Fig. 4l), i.e. old leaves were apparently
replaced throughout the year.
In D. obtusifolius, D. tuberculatus and Craibiodendron, 3-
year mean phenology scores show low levels of mature leaves,
deciduousness and new leaves throughout the year (Figs 4j,k
and 5). Phenological records of individual trees in these species
indicate that throughout the year some branches were leafing
while others shed old leaves or were leafless (Fig. 6). Thus, in
contrast to the other deciduous species, seasonal development
at higher elevations was asynchronous within the crowns of
these species. Most leaf flushes were preceded by a distinct
decline in the fraction of mature leaves, i.e. irregular leaf
shedding caused leaf flushing unrelated to climatic seasonality
(Fig. 6).
To assess the crucial role of subsoil water reserves in dry-
season flushing, we monitored the vegetative phenology of 123
Shorea siamensis trees of different size classes growing at HKK on
a gentle slope transected by an ephemeral stream bed, the probable
location of the largest subsoil water reserves (Fig. 7). In mid-
March 1998, 2 weeks before the first monsoon rains, the frac-
tions of flushing (20–100% crown cover) and dormant trees
were, respectively, correlated directly and inversely with tree size
(Fig. 8a, symbols). Small trees had formed new leaves only when
growing within 2 m of the creek bed (Fig. 7). The mean distance
of small flushing trees from the creek bed was therefore one-third
that of dormant trees, but the difference between distances of
large flushing and dormant trees was relatively small (Fig. 8a,
bars). Two weeks before the first rains, the largest trees
(> 24 cm d.b.h.) had expanded most of their new leaves, but
in medium-sized trees leaf cover was only approximately
20% (Fig. 8b left, circles, squares); leaf buds had just begun to
expand in about half of the many small trees with d.b.h. < 15 cm
(Fig. 7), in which rapid leaf expansion started immediately
after the first rains (Fig. 8b, triangles). During the following
dry season, trees of all size classes shed leaves at the same rate
(Fig. 8b right).
DISCUSSION
Control of vegetative phenology in Thailand and India
At the altitude of the observed transect in DSNP (650–780 m)
the transition from deciduous to evergreen phenology is more
gradual than at lower elevations (Fig. 5), but the environmental
and endogenous controls of vegetative phenology (Fig. 2) can be
deduced reliably from timing and synchrony of the phenophases
established by the novel analysis of phenological records (Figs 3
and 4; Methods). Distinct minima in mature leaf scores followed
by maxima for young leaves in the 3-year mean phenological
scores indicate synchronous leaf shedding and flushing of
Figure 3 Identification of synchronous
bud break among conspecific trees of the
deciduous species Antidesma acidum in
a mixed deciduous–evergreen forest
near Chiang Mai, Thailand. (a) Seasonal
development of two individual trees over 2
consecutive years. Framed scores — new
leaf flush. (b) Mean phenology scores for
five trees over 2 years. (c) Mean phenology
scores for five trees over 3 years (December
1988–December 1991). For explanations
see Methods. Phenological stages: M, mature
leaves; S, senescent leaves; B, bare, leafless;
Y, young light-green leaves. Grey bars
indicate the time of spring-flushing induced
by increasing daylength.

S. Elliott et al.
© 2006 The Authors
6Global Ecology and Biogeography, Journal compilation © 2006 Blackwell Publishing Ltd
conspecific trees at species-specific times with low interannual
variation (Fig. 4a–h, filled circles). Such regular patterns of
phenology are controlled by seasonal variation in environmental
factors (Fig. 2). In contrast, the absence of distinct maxima or
minima for these phenophases indicates that the replacement of
old by young leaves is determined mainly by endogenous factors
such as leaf ageing (Fig. 4i–l).
Antidesma is the only deciduous species with rain-induced
leafing, because leafing time varied with the first monsoon rains
(Fig. 3b). Highly synchronous leafing of ‘spring-flushing’ species
at DSNP and in the majority of deciduous tree species in Indian
monsoon forests (Fig. 1; Kushwaha & Singh, 2005) occurs every
year around the spring equinox, well before the first monsoon
rains (Fig. 4b–f); it is therefore induced by increasing daylength
(Rivera et al., 2002).
At lower elevations of DSNP, spring-flushing species of
the dipterocarp–oak forest are leafless between late January
and March (2–3 months deciduous; Elliott et al., 1994; Maxwell
& Elliott, 2001), i.e. deciduousness is similar to that observed
in Indian dry monsoon forests of the Deccan Peninsula (Fig. 1;
Troup, 1921; Kushwaha & Singh, 2005). Increasing deciduous-
ness with decreasing altitude has been also described for the
Neotropical Erythrina poeppigiana and Tabebuia rosea (Borchert,
1991).
Differences in phenology between Asian and
Neotropical monsoon forests
The amount and seasonal distributions of rainfall are similar in
the monsoon forests of northern Thailand and Costa Rica
(Fig. 2a), yet the fractions of evergreen vs. deciduous species vary
widely between these forests (Table 2; Rundel & Boonpragob,
1995). The low degree of deciduousness in DSNP indicates that
water stress during the dry season is distinctly lower than in
Figure 4 Three-year mean scores of
vegetative phenology (see Fig. 3c) for 12 tree
species of five functional types observed in
Doi Suthep National Park near Chiang Mai,
Thailand. Filled circles indicate the first
distinct increase in the formation of young
leaves, i.e. the start of leafing. (a) Deciduous
species: flushing induced by the first heavy
monsoon rains after the spring equinox.
(b–f) Spring-flushing species: synchronous
flushing induced by increasing daylength at
the same time each year around the spring
equinox (March/April, dotted vertical lines).
(g–i) Leaf-exchanging species: flushing
induced during the early dry season
(January/February) by partial leaf-shedding.
Early monsoon rains (May/June) may induce
a second flush (filled circles in May/June).
(j–k) Irregularly leaf-exchanging species:
leaf-shedding and flushing vary widely
among trees and between years (compare
with Fig. 6). (l) Evergreen species.

Dry season leaf flushing
© 2006 The Authors
Global Ecology and Biogeography, Journal compilation © 2006 Blackwell Publishing Ltd 7
Costa Rica. This probably has two main causes. First, because of
higher latitude (19 vs. 10° N) and altitude (600,800 m vs. 40 m),
temperatures and hence evaporative water loss are substantially
lower at DSNP than in lowland Costa Rica during the early dry
season (22–24 °C vs. 28 °C). Secondly, subsoil water storage is
likely to be larger in the deeply weathered soils at DSNP than at
the prevailing dry upland sites in Costa Rica.
Tree phenology at lower elevations of DSNP and in Indian
monsoon forests is intermediate between that described here
(Fig. 4) and observed in Costa Rica (Table 2; Fig. 2). The major-
ity of Indian tree species leaf during the late dry season before the
arrival of the monsoon rains (Table 2; Fig. 1), indicating that
substantial subsoil water storage in old, deeply weathered soils
permits spring-flushing during the late dry season.
Soil water reserves and spring flushing
The dramatic variation in the time of leafing with tree size and
topography among Shorea siamensis trees at HKK (Figs 7 and 8)
constitutes the first quantitative assessment of the crucial role of
subsoil water reserves for leaf expansion during the dry season.
Different rates of leafing indicate that near the dry creek bed sub-
soil water reserves were accessible by trees of all sizes, but at
uphill sites they were within the reach of large, deep-rooted trees
only (Figs 7 and 8a). The rapid increase in leaf cover of small
trees within a week after the first rains (Fig. 8b, week 13) indic-
ates that growth of non-dormant buds had been inhibited by
water stress and resumed immediately after rehydration. Leaf
abscission was independent of subsoil water reserves (Fig. 8b,
1999) and was probably caused by increasing leaf water stress
in ageing leaves (Borchert et al., 2002).
The effect of tree size, topography and soil quality on dry-season
phenology depending on subsoil water reserves has been
Table 2 Fraction of species of three major functional types in a Neotropical semi-deciduous forest (Costa Rica: Frankie et al., 1974) and in three
Asian monsoon forests (India: *Troup, 1921; †Kushwaha & Singh, 2005; Thailand: this study). At all sites > 90% of annual rainfall of 800–
1600 mm is received during the wet season between May/June and October/November
Functional type
Costa Rica India* India† Thailand
species % species % species % species %
3–6 month deciduous 38 41 23 23 2 22 1 5
Spring-flushing 32 34 62 63 6 67 8 42
Leaf-exchanging: evergreen 23 25 13 13 1 11 10 53
Species total 93 98 9 19
Figure 5 Var iations in deciduousness (period w ith mature leaf score
< 1) and full crown cover (period with mature leaf score > 3) among
16 species of five functional types in Doi Suthep National Park near
Chiang Mai, Thailand.
Figure 6 Seasonal variation of vegetative phenology in three
Dipterocarpus obtusifolius trees in Doi Suthep National Park near
Chiang Mai, Thailand.

S. Elliott et al.
© 2006 The Authors
8Global Ecology and Biogeography, Journal compilation © 2006 Blackwell Publishing Ltd
observed elsewhere (Sayer & Newbery, 2003; Borchert et al.,
2004). It is particularly well documented for the life history of
Shorea robusta. During the monsoon season Shorea seedlings
grow well in a variety of soils; during the dry season those on
sandy riverbanks usually die, whereas those on sandy loam sur-
vive, but die back regularly until they have developed deep roots
(Troup, 1921). The vegetative phenology of mature trees varies
with topography from evergreen to 1-month deciduous (Singh &
Kushwaha, 2005). During consecutive years of severe drought,
mortality is low at moist sites with deep sandy loams and high at
relatively dry sites (Troup, 1921; Seth et al., 1960; Borchert, 1998).
Deep-rooted trees of evergreen forests in eastern Amazonia
maintain full leaf cover during exceptional drought by extracting
> 500 mm water from a potential reservoir of > 800 mm of
plant-available water (Nepstad et al., 1994).
The opportunistic phenology of other drought-resistant
dipterocarp species also appears to be determined by tree
water balance during the dry season. Trees exchanged leaves
irregularly at higher elevations with moderate water loss (Fig. 6),
but were 1–3-month deciduous at lower elevations, where
Figure 7 Distribution, size and leaf cover
of 123 Shorea siamensis trees observed in
mid-March 1998, 2 weeks before the first
rainfall of the wet season. The 50 × 50-m
experimental plot at the Huai Kha Khaeng
Wildlife Sanctuary in central Thailand is
slightly sloped and crossed by a dry creek
bed (thick grey line with squares to calculate
distances of trees from the creek bed).
Numbers on straight grey contour lines
give elevation in m above the lowest point
at right. Diameters and tree numbers (in
parentheses) in the four size classes shown
are: 6–13 cm (32); < 17 cm (29); < 22 cm
(34); 23–56 cm (28).
Figure 8 Var iation of vegetative phenology in 123 Shorea siamea trees with tree size and topography at the Huai Kha Khaeng Wildlife Sanctuary
in central Thailand (see Fig. 7). (a) Size-dependent fraction of flushing (squares) and dormant trees (circles) and their mean distance from the
creek bed (flushing trees, hatched bars; dormant trees, grey bars). (b) Leaf flushing before (weeks 10–12) and after the first rainfall (bars) in week
12 of 1998 (left) and leaf shedding during the mid-dry season in 1999 (right). Mean d.b.h. and fraction of all trees for the three experimental
groups are 25.5 cm (14%, circles), 19 cm (15%, squares) and 16.7 cm (71%, triangles). Standard errors of all means for crown cover, d.b.h. and
distances from creek are < 0.1 (not shown).

Dry season leaf flushing
© 2006 The Authors
Global Ecology and Biogeography, Journal compilation © 2006 Blackwell Publishing Ltd 9
they leafed whenever their water balance became positive
(Fig. 8b, left).
Adaptive significance of spring flushing
In India and Thailand, increasing daylength after the spring
equinox signals the approach of the monsoon season. Induction
of leaf flushing by increasing daylength assures that a full com-
plement of young, photosynthetically competent leaves is in
place when the monsoon rains begin, yet precludes prolonged
exposure of young leaves to severe drought. The predominance
of spring-flushing species among Indian monsoon forest trees
(Table 2; Fig. 1; Kushwaha & Singh, 2005) suggests that this pheno-
logical strategy is particularly advantageous, probably because
it optimizes use of large subsoil water reserves for photosynthetic
activity during seasonal drought and thus extends the relatively
short, wet growing season. The characteristic phenology of
spring-flushing species is controlled mainly by two non-climatic
environmental variables, water storage in deep soils and photo-
periodic induction of leafing, which also determine tree pheno-
logy in other tropical dry forest species around the globe
(Borchert, 1994a; Rivera et al., 2002).
REFERENCES
Bhat, D.M. (1992) Phenology of tree species of tropical moist
forest of Uttara Kannada district, Karnataka, India. Journal of
Bioscience, 17, 325 –352.
Borchert, R. (1991) Growth periodicity and dormancy. Physio-
logy of trees (ed. by A.S. Raghavendra), pp. 221–245. John
Wiley, New York.
Borchert, R. (1994a) Water storage in soil or tree stems deter-
mines phenology and distribution of tropical dry forest trees.
Ecology, 75, 1437–1449.
Borchert, R. (1994b) Induction of rehydration and bud break by
irrigation or rain in deciduous trees of a tropical dry forest in
Costa Rica. Trees, 8, 198–204.
Borchert, R. (1998) Responses of tropical trees to rainfall
seasonality and its long-term changes. Climatic Change, 39,
381–393.
Borchert, R. (2000) Organismic and environmental controls of
bud growth in tropical trees. Dormancy in plants: from whole
plant behavior to cellular control (ed. by J.D. Viemont and
J. Crabbé), pp. 87–107. CAB International, Wallingford.
Borchert, R. (2005) The phenology of tropical trees and its
environmental control. UTP: http://www.biology.ku.edu/
tropical_tree_phenology/index.html.
Borchert, R., Meyer, S.A., Felger, R.S. & Porter-Bolland, L.
(2004) Environmental control of flowering periodicity in
Costa Rican and Mexican tropical dry forests. Global Ecology
and Biogeography, 13, 409 – 425.
Borchert, R., Rivera, G. & Hagnauer, W. (2002) Modification of
vegetative phenology in a tropical semideciduous forest by
abnormal drought and rain. Biotropica, 34, 27–39.
Borchert, R., Robertson, K., Schwartz, M. & Williams-Linera, G.
(2005) Phenology of temperate trees in tropical climates. Inter-
national Journal of Biometeorology, 50, 57–65.
Elliott, S., Maxwell, J.F. & Beaver, O.P. (1989) A transect survey of
monsoon forest in Doi-Suthep National Park. Natural History
Bulletin of the Siam Society, 37, 137–141.
Elliott, S., Promkutkaew, S. & Maxwell, J.F. (1994) Flowering and
seed production phenology of dry tropical forest trees in
Northern Thailand. Proceedings: International symposium
on genetic conservation and production of tropical forest tree
seed (ed. by R.M. Drysdale, S.E.T. John and A.C. Yapa), pp. 52–
61. ASEAN — Canada Forest Tree Seed Centre, Saraburi,
Thailand.
Frankie, G.W., Baker, H.G. & Opler, P.A. (1974) Comparative
phenological studies of trees in tropical wet and dry forests in
the lowlands of Costa Rica. Journal of Ecology, 62, 881–919.
Kushwaha, C.P. & Singh, K.P. (2005) Diversity of leaf phenology
in a tropical deciduous forest in India. Journal of Tropical Eco-
logy, 21, 47–56.
Maxwell, J.F. & Elliott, S. (2001) Vegetation and vascular flora of
Doi Sutep–Pui National Park, Chiang Mai Province, Thailand.
Thai Studies in Biodiversity 5. Biodiversity Research and Train-
ing Programme, Bangkok.
Nepstad, D.C., de Carvalho, C.R., Davidson, E.A., Jipp, P.H.,
Lefebvre, P.A., Negreiros, G.H., Trumbore, S.E. & Vieira, S.
(1994) The role of deep roots in the hydrological and carbon
cycles of Amazonian forests and pastures. Nature, 372, 666–
669.
Prasad, S.N. & Hegde, M. (1986) Phenology and seasonality
in the tropical deciduous forest of Bandipur, South India.
Proceedings of the Indian Academy of Sciences (Plant Sciences),
96, 121–133.
Puri, G.S. (1969) Indian forest ecology. Oxford Book and Station-
ery Co., New Delhi.
Rivera, G., Elliott, S., Caldas, L.S., Nicolossi, G., Coradin, V.T.R.
& Borchert, R. (2002) Increasing day-length induces spring
flushing of tropical dry forest trees in the absence of rain. Trees,
16, 445 – 456.
Rundel, P.W. & Boonpragob, K. (1995) Dry forest ecosystems of
Thailand. Seasonally dry tropical forests (ed. by S.H. Bullock,
H.A. Mooney and E. Medina), pp. 93–123. Cambridge Uni-
versity Press, Cambridge.
Sayer, E.J. & Newbery, D.M. (2003) The role of tree size in the
leafing phenology of a seasonally dry tropical forest in Belize,
Central America. Journal of Tropical Ecology, 19, 539 –548.
Seth, S.K., Khan, M.A.W. & Yadav, J.S.P. (1960) Sal mortality in
Bihar. Indian Forester, 86, 645 – 679.
Singh, K.P. & Kushwaha, C.P. (2005) Paradox of leaf phenology:
Shorea robusta is a semi-evergreen species in tropical dry
deciduous forests in India. Current Science, 88, 1820 –1824.
Sukwong, S., Dhamanitayakul, P. & Pongumphai, S. (1975)
Phenology and seasonal growth of dry dipterocarp forest tree
species. Kasetsart Journal, 9, 105 –114.
Troup, R.S. (1921) The silviculture of Indian trees, Vol. I. Claren-
don Press, Oxford.
Walter, H. (1971) Ecology of tropical and subtropical vegetation.
Oliver & Boyd, Edinburgh.

S. Elliott et al.
© 2006 The Authors
10 Global Ecology and Biogeography, Journal compilation © 2006 Blackwell Publishing Ltd
Williams, R.J., Myers, B.A., Muller, W.J., Duff, G.A. & Eamus, D.
(1997) Leaf phenology of woody species in a Northern
Australian tropical savanna. Ecology, 78, 2542–2558.
Woodward, F.I. (1987) Climate and plant distribution. Cam-
bridge University Press, London.
BIOSKETCHES
Stephen Elliott is Co-Director of the Forest Restoration
Research Unit (FORRU) in the Department of Biology at
Chiang Mai University in Chiang Mai, Thailand. He is
Lecturer in Ecology and Wildlife and has carried out
research on the restoration of tropical forest ecosystems for
18 years.
Patrick Baker is Lecturer of Vegetation Ecology at Monash
University, Melbourne, Australia. He studies the historical
stand dynamics and disturbance histories of tropical forests.
Rolf Borchert is Professor Emeritus of Ecophysiology in
the Division of Biological Sciences at the University of
Kansas, Lawrence, KS, USA. He has analysed the
environmental control of the phenology of tropical forest
trees for more than 25 years.