American Journal of Botany 86(1): 6269. 1999.

BEETLE POLLINATION OF SHOREA PARVIFOLIA

(SECTION MUTICA, DIPTEROCARPACEAE)
IN A GENERAL FLOWERING PERIOD
IN

SARAWAK, MALAYSIA1

SHOKO SAKAI,2,3 KUNIYASU MOMOSE,2,5 TAKAKAZU YUMOTO,2

MAKOTO KATO,4 AND TAMIJI INOUE2
2Center for Ecological Research, Kyoto University, Sakyo, Kyoto, 606-8502;
Biological Laboratory, Yoshida College, Kyoto University, Sakyo, Kyoto, 606-8501, Japan

Pollination ecology of an emergent tree species, Shorea (section Mutica) parvifolia (Dipterocarpaceae), was studied using
the canopy observation system in a lowland dipterocarp forest in Sarawak, Malaysia, during a general flowering period in
1996. Although the species has been reported to be pollinated by thrips in Peninsular Malaysia, our observations of flower
visitors and pollination experiments indicated that beetles (Chrysomelidae and Curculionidae, Coleoptera) contributed to
pollination of S. parvifolia in Sarawak. Beetles accounted for 74% of the flower visitors collected by net-sweeping, and
30% of the beetles carried pollen, while thrips accounted for 16% of the visitors, and 12% of the thrips carried pollen. The
apical parts of the petals and pollen served as a reward for the beetles. Thrips stayed inside the flower almost continuously
after arrival, and movements among flowers were rare. Fruit set was significantly increased by introduction of beetles to
bagged flowers, but not by introduction of thrips. Hand-pollination experiments and comparison of fruit set in untreated,
bagged, and open flowers suggested that S. parvifolia was mainly outbreeding.
Key words:

beetle; Dipterocarpaceae; general flowering; pollination; Sarawak; Shorea; thrips.

Lowland tropical rainforests in west Malesia are characterized by high diversity of tree species (Whitmore,
1984; Richards, 1997), dominance of Dipterocarpaceae
in the canopy and emergent layers (Ashton, 1982, 1988),
and general flowering. General flowering is a unique phenomenon in lowland mixed dipterocarp forests that occurs at an average interval of 5 yr but rather irregularly
(Ashton, Givnish, and Appanah, 1988; Appanah, 1993).
During a general flowering period (GFP) that usually
continues for several months, nearly all species of Dipterocarpaceae and many species of other families bloom
heavily, while many of them hardly bloom in other years
(Ashton, Givnish, and Appanah, 1988; Appanah, 1993;
Sakai et al., in press). Because such irregular and intense
general flowering can bring about immense demands for
pollinators, one of the most interesting and important
problems is what pollination systems are adopted by
these general flowering species (Ashton, 1988; Appanah,
1990).
The pollination system of Shorea section Mutica (Dipterocarpaceae) has been reported during a GFP at Pasoh

Forest Reserve, Peninsular Malaysia (Chan and Appanah,

1980; Appanah and Chan, 1981). All six species of Shorea sect. Mutica in Pasoh including S. parvifolia Dyer
were exclusively visited and pollinated by thrips (Appanah and Chan, 1981). Appanah and his colleague noted
that the short generation time and high reproductive rate
of thrips permit a quick response of thrips to an abrupt
increase of flowers at the beginning of the general flowering and that thrips provide sufficient pollination service
for multiple species of Shorea.
We began monitoring plant phenology using a canopy
observation system in August 1992 in Lambir Hills National Park, Sarawak (Inoue et al., 1995). A general flowering was observed in 1996 for the first time (S. Sakai
et al., in press). Sixty-five species of Shorea including 14
species of sect. Mutica have been recorded from the Park
(P. S. Ashton, personal communication). Among the nine
species of sect. Mutica that we monitored, four species
flowered during the period. Unexpectedly, beetles were
found to be the predominant flower visitors of all nine
species of Shorea sect. Mutica, and also other sections
under observation in emergent and canopy layers of the
forest (Momose et al., 1998).
This paper presents field observations and experiments
on the breeding system and pollination of an emergent
tree, Shorea parvifolia, a member of sect. Mutica. The
canopy observation system (Inoue et al., 1995) allowed
continuous observation of flowering phenology and pollination processes, and manipulation of experiments on
tree crowns over 60 m above the forest floor. In addition
to flower visitation frequency and the amount of body
pollen loads, the ability of flower visitors to effect fruit
set was examined by experiments in which potential pollinators were introduced to bagged flowering inflores-

1 Manuscript received 5 Feburary 1998; revision accepted 4 June

1998.
The authors thank to Dr. H. S. Lee and Mr. A. A. Hamid, Forest
Department Sarawak, and for their support and organization of our
study; R. Rapi, R. Johan, and Prof. K. Ogino, The University of Shiga
Prefecture, B. Nyambong for help in field observations; Dr. I. Kudo for
identification of thrips; Prof. P. S. Ashton, Harvard University, for reading the manuscript and making a number of helpful suggestions. This
study was partly supported by Grants-in-Aid of the Japanese Ministry
of Education, Science and Culture (Numbers 04041067, 06041013, and
09NP1501) and by JSPS Research Fellowships for Young Scientists for
S. Sakai.
3 Author for correspondence.
5 Current address: Graduate School for Asian and African Area Studies, Kyoto University, Kyoto 606-8501, Japan.

62

January 1999]

SAKAI

ET AL.BEETLE POLLINATION OF

SHOREA PARVIFOLIA

63

TABLE 1. Individuals of Shorea parvifolia observed from tree towers and walkways with developmental stage (DS). Magnitudes of flowering and
fruiting in the two flowering periods in 1996 are shown using following grades: 2, (flowers/fruits) absent; 1, few, scattered, or covering only
a small part of the crown; 1, covering less than half of the crown; 2, abundant but not over the whole crown; 3, covering the whole crown.
No flowering with a grade 1 was observed. Length of flowering period is shown in parentheses.
30 April22 June

No. of reproductive
events during
Aug 1993Dec 1996

Flowering grade
(length of flowering)

19 Sep7 Nov
Fruiting
grade

Flowering grade
(length of flowering)

Fruiting
grade

Tree

DS

233
235

III
III

0
0

2
2

2
2

2
2

2
2

228
229

IV
IV

1
1

2
3 (27 d: 11 May6 June)

2
2

3 (15 d: 8 Oct22 Oct)

2

2
2

225
231
161
230
232

V
V
V
V
V

2
2
died in Aug 1993
died in Jan 1996
died in Sep 1994

3 (42 d: 30 April10 June)

3 (33 d: 21 May22 June)
2
2
2

2
2
2
2
2

1 (19 d: 10 Oct28 Oct)

3 (49 d: 19 Sep7 Nov)
2
2
2

2
2
2
2
2

III: subcanopy trees, 12.527.5 m; IV: canopy trees, 27.542.5 m; V: emergent trees .42.5 m.

cences. This study suggests that different pollination systems work in dipterocarp forests in Peninsular Malaysia
and in Sarawak.
MATERIALS AND METHODS
Study site and plantThe study site was a primary lowland dipterocarp forest in Lambir Hills National Park, Sarawak, Malaysia (48209 N,
1138509 E, altitude 150250 m). In August 1992, a Canopy Biology
Plot (CBP) was demarcated for long-term monitoring of plant phenology and for the observation of plantanimal interactions, by the Canopy
Biology Program of Kyoto University and Sarawak Forest Department.
The CBP covered an area of 8 ha (200 3 400 m) and had a canopy
observation system that consisted of tree towers and aerial walkways
(Inoue et al., 1995).
The genus Shorea (Dipterocarpaceae) is the dominant emergent tree
genus of the lowland forest of West Malesia, with 163 species throughout Malesia (Ashton, 1982; Ashton, Givnish, and Appanah, 1988) and
65 species in Lambir (P. S. Ashton, personal communication). Shorea
parvifolia is a member of sect. Mutica with 27 species in Malesia and
14 species in Lambir and is one of the constituents of the emergent
layer. In CBP, eight trees of S. parvifolia with .40 cm dbh (diameter
at breast height) were found.
Pollination of three individuals of S. parvifolia was studied during
1428 May 1996. Two trees in CBP, trees 225 (132 cm dbh, height 60
m) and 229 (44 cm dbh, height 35 m) (Table 1), were accessed by the
canopy observation system (Fig. 1). In addition, a crown of one other
emergent tree near the headquarters of the National Park (tree 1001)
was accessed by aluminum ladders.
PhenologyReproductive activity of nine individuals of S. parvifolia
was monitored twice a month for 43 mo from June 1993 from the
canopy observation system, but three of them died before 1996. From
April to June 1996, the census was intensified to three times a month.
Among the remaining six trees, two trees were in the subcanopy layer
(12.527.5 m), two in the canopy layer (27.542.5 m), and two were
emergent (.42.5 m) (Table 1).
We recorded the magnitude of flowering and fruiting events using
the following grades: , flowers/fruits absent; 1, few, scattered, or covering only a small part of the crown; 1, covering less than half of the
crown; 2, abundant but not over the whole crown; 3, covering the whole
crown.
Collection of flower visitors and pollen on stigmaFlower visitors
were collected by net-sweeping or by flower collection. In net-sweeping, we put a branch with ;200 open flowers into an insect net and

quickly shook insects into a sealable plastic bag. We repeated the procedure five times on different branches at each collection time. The
numbers of beetles and thrips per flower were calculated from eight
samples taken on 1315 May 1996 (1800 and 2200 on 13 May; 0300,
1000, 1730 and 2230 on 14 May; 0230 and 0630 on 15 May) on tree
225. Variation in flower visitors among the trees was examined using
additional samples collected at 1700 on 19 May on trees 225 and 229,
and at 1000 on 16 May and 1700 on 20 May on tree 1001. The insects
were brought to the laboratory within an hour and killed in a refrigerator. They were then pinned or fixed in alcohol and labeled.
Flower visitors that hid inside the corollas and could not be collected
by net-sweeping were collected by flower collection. For flower collection, we cut off inflorescences with 46124 flowers on tree 225 inside
a sealable plastic bag. We brought the plastic bags to the laboratory and
counted flower visitors in each of the corollas under a binocular microscope, including flower visitors fallen from the corollas in the bag. The
insects were preserved for identification in vials filled with 50% alcohol,
except for thrips, which were kept in vials with glycerin-alcohol. We
sampled two bags at each collection time at 6-h intervals on 16 May
on tree 225.
The eight net-sweeping samples on 1315 May and the four flowercollection samples on 16 May were used to examine changes in the
densities of beetles and thrips in the crown in a day. The numbers of
insects per flower during four 6-h periods were calculated by summing
the averaged number of insects per flower collected by net-sweeping
during the period and that by flower sampling. Standard errors for thrips
numbers were calculated using flower collection samples, and those for
beetle numbers were calculated using net-sweeping samples. Then we
adjusted the standard errors to the total means of both samples. The
standard error for beetles at 1200 were not computed because only one
net-sweeping sample was available.
Some of the beetles and thrips from the net-sweeping samples were
used to examine pollen loads on their bodies or stomach contents. All
collections for trees 229 and 1001, and seven collections among 11 for
tree 225 were classified to order and family for Coleoptera. All insects
collected by flower collection were classified to order and species for
Thysanoptera. In addition to the above sampling in the crown, abscised
corollas fallen on a terrace of the tower (25 m above the ground) were
collected to examine insects at 1100 on 14 May.
Finally, we collected 2025 open-pollinated flowers and fixed them
in FAA (formalin; acetic acid; alcohol) at 1800 and 2200 on 13 May,
0300, 0600, 1000, and 1500 on 14 May, and 1200 on 16 May. Pollen
grains on stigmas were counted under a microscope. Rank correlation
between sampling time and the number of pollen grains on a stigma
was examined by Spearmans rank correlation test, because intensity of
pollinator activities might change in the course of the day. Flowers

64

AMERICAN JOURNAL

OF

BOTANY

[Vol. 86

Figs. 15. Flowering and insect visitors of Shorea parvifolia. 1. Tree 225 and the tree tower. On the top terrace (height 5 60 m) of the tower,
S. Sakai (white arrow) observes immature fruits. At the base of the tower, aerial walkways run at 25 m above the ground. 2. Flowers of S. parvifolia
at anthesis viewed from the top terrace of tree 225. 3. Fruit of S. parvifolia 30 d after flowering. Bar 5 5 cm. 4. A beetle, Monolepta sp. (Coleoptera:
Chrysomelidae), feeding on a flower. A white arrow points to the tip of a petal gnawed by a beetle. Bar 5 2.5 mm. 5. Thrips hidden within a
corolla. Bar 5 1 mm.

collected on 16 May were also examined for damage due to gnawing

by beetles.
Pollination experimentsTo examine the breeding system of S. parvifolia and the contributions of insect visitors to pollination, we performed seven experiments on tree 225: (1) open: flowers on four inflorescences were left exposed permitting unhindered insect visitation; (2)
untreated, bagged: flowers on two inflorescences were bagged before
the tree started blooming; (3) open, flower-reduced: flowers on three
inflorescences were removed except for flowers that opened on 1920
May as controls of experiments 47; (4) self-pollinated (geitonogamous); (5) cross-pollinated: flowers on 11 inflorescences were handpollinated with geitonogamous pollen from different inflorescences on
the same tree for experiment 4 (self-pollinated), or with cross pollen
from trees 229 and 1001 for experiment 5 (cross-pollinated), sampled
just before anthesis on 1920 May (pollen was transferred using Chi-

nese writing brushes, and all untreated flowers were removed); (6)
thrips-introduced; and (7) beetle-introduced. Experiments 6 and 7 were
made in the following manner twice, on 22 and 27 May: we collected
flower visitors on tree 229 at 2000 by sweeping four branches (;200
open flowers on each) and separated thrips, beetles, and other insects.
The thrips and beetles were each released separately into two tetron
bags (TORAY, tetront, number 9000) enclosing the inflorescences before anthesis on tree 225, and left for 2 d. All untreated flowers on the
inflorescences were removed. This procedure resulted in introduction of
5354 beetles or 1837 thrips into each bag.
We monitored changes in the number of fruits on branches every 2
wk until fruit dispersal in experiments 1 and 2, and until 32 d after
flowering in experiments 37. In addition, we followed unbagged flowers and fruits on five inflorescences of tree 229 until seed dispersal.
To examine whether exclusion of flower visitors caused a decrease
of fruit set, we compared fruit set in open and untreated, bagged flowers

January 1999]

SAKAI

ET AL.BEETLE POLLINATION OF

TABLE 2. Orders of the insects (%) collected by net-sweeping method

on three trees of Shorea parvifolia (trees 225 and 229, 19 May;
tree 1001, 16 and 20 May). The numbers of sampling times are
shown in parentheses.

Orthoptera
Blattaria
Hemiptera
Thysanoptera
Coleoptera
Diptera
Hymenoptera
Lepidoptera
Total %
No. of insects

229 (1)

225 (7)

1001 (2)

0.7
0.0
0.7
32.4
66.2
0.0
0.0
0.0

0.0
0.3
8.6
10.8
74.7
1.1
3.2
1.3

0.0
0.9
1.9
11.2
85.0
0.9
0.0
0.0

0.2
0.3
5.6
16.0
74.4
0.8
1.9
0.8

100.0
148

100.0
371

100.0
107

100.0
519

Coleoptera
family

Total

(experiments 1 and 2) on day 48 after flowering peak by Fishers exact

test. Then, fruit set in experiments 37 was compared with that in untreated, bagged flowers (experiment 2) on day 32 after flowering to
investigate which treatments increased fruit set. At that time, the fruits
weighed 0.116 6 0.039 g (dry mass, N 5 16), 40% of the dry mass of
mature fruits (0.271 6 0.075 g, N 5 21), and their sepals had turned
from green to red (Fig. 3).

RESULTS
Flowering phenology and floral biologySince we
began monitoring reproductive phenology in June 1993,
Shorea parvifolia flowered twice, 30 April22 June 1996
and 19 September7 November 1996. Flowering frequency was higher in larger trees: flowering occurred
twice in the two emergent trees, once in the two canopy
trees, and was lacking in the subcanopy trees (Table 1).
In the first flowering event in 1996, tree 225 flowered
from 30 April to 10 June and reached a plateau from 14
to 28 May (Fig. 2). The inflorescences are terminal or
axially panicles with 150 6 65 (N 5 8) branchlets. A
single branchlet of an inflorescence produces 4.6 6 1.7
flowers (N 5 205). Flowers are yellow, ;1.3 cm in diameter with a dry mass of 0.037 6 0.0034 g (N 5 21).
The five revolute petals form a bowl-shaped structure at
the center of the corolla, in which an ovoid ovary with
a distinct stylopodium is located, surrounded by 15 stamens. The stamens are arranged in three verticils. Each
stamen bears two-celled anthers, each with two thecae,
and a terminal awn-like appendage, which becomes reflexed at anthesis. The pollen is smooth but slightly
sticky, not dry, and not easily dislodged. The pendant
flowers open around 1800 (Fig. 2), releasing a strong,
sweet scent. The anther thecae dehisce just before anthesis. The flowers do not secrete nectar. The corollas start
to fall off in the following morning, while 68% of the
expanded corollas remain in the crown until the next
evening. They drop or are pushed off when new flowers
open. By 2300 on the next day, almost all the old flowers
are shed.
Flower visitors and pollen on stigmaIn net-sweeping, 6685% of all insects identified were small beetles
(,5 mm) and 1032% were thrips (Thysanoptera). The
composition of the net-sweeping samples collected on
different trees was almost the same at the level of orders,

65

TABLE 3. Families of flower-visiting beetles (%) collected by the netsweeping method on three trees of Shorea parvifolia (tree 225 and
299, 19 May; tree 1001, 16 and 20 May). The numbers of sampling
times are shown in parentheses.

Tree no.
Insect order

SHOREA PARVIFOLIA

Tree no.
229 (1)

225 (7)

1001 (2)

Total

Scarabaeidae
Nitidulidae
Cryptophagidae
Corylophidae
Coccinellidae
Chrysomelidae
Curculionidae
Other

1.0
7.1
4.1
0.0
0.0
57.1
29.6
1.0

1.9
12.5
0.4
0.0
1.1
31.4
49.2
3.4

3.3
0.0
0.0
24.2
0.0
60.4
8.8
3.3

2.0
8.8
1.1
4.9
0.7
42.8
36.9
2.9

Total
No. of insects

100.0
98

100.0
264

100.0
91

100.0
453

though the percentage of thrips was much higher in tree

229 (32%) than in the other trees (1011%) (Table 2).
Within Coleoptera, Chrysomelidae (42%) and Curculionidae (36%) were the most abundant (Table 3). Corylophidae was abundant (24%) on tree 1001, but were not
recorded from the other two trees. In flower-collection
samples, all of the collected insects were thrips (74%) or
beetles (26%) (Table 4; Figs. 4, 5). The number of thrips
in fallen corollas was much lower than that in the flowers
on the tree crown (0.014 thrips per corolla). Ten species
from three families of thrips were found from the flowercollection samples (Table 4). Thrips hawaiiensis accounted for 75% of all the thrips identified.
Chrysomelid beetles, especially several species of
Monolepta (Galerucinae), were very frequently observed
TABLE 4. Insects collected by the flower-collection method on three
trees of S. parvifolia. All the insects were thrips and beetles, and
only the thrips were identified to species. In addition to the adult
thrips, only one larva of Phalaeothripidae sp. was collected (four
collections of 6-h intervals, 16 May, tree 225).
Order
Suborder
Family
Species

Coleoptera

75

Thysanoptera
Terebrantia
Aeolothripidae
Desmidothrips sp. (inanditus ?)
Thripidae
Craspedothrips minor
Ernothrips sp. (lobatus ?)
Megalurothrips typicus
Thrips coloratus
Thrips hawaiiensis
Thrips sp. (aleuritis ?)
Thrips sp. (alius ?)
Tubulifera
Phlaeothripidae
Hoplandrothrips flavipes
Haplothrips sp. (tenuioennis ?)
Larva (species unknown)
Not identifieda
Total
a

3
1
1
5
25
118
2
1

1
1
1
39
273

Could not be identified because of bad specimen condition.

66

AMERICAN JOURNAL

Fig. 6. Means and standard errors of the numbers of thrips (open

circle) and beetles (open square) per flower in the crown estimated
based on two sampling methods, net-sweeping and flower collection.
The standard error for beetles at 1200 is not shown because only one
net-sweeping sample was available.

walking around and flying among inflorescences during

the night. They were observed feeding on flower petals
and pollen on the inner surface of the corolla (Fig. 4). A
considerable amount of pollen was found in their stomachs. Petals of S. parvifolia have a thin, frilled edge toward the apical half, and we found that many beetles feed
on this part of the petal (Fig. 4). Seventy-five percent of
the 63 flowers collected at 1200 had damaged apical petal
fringes and 29% had damaged stigmas, but we observed
no damage to the ovaries. Mating behavior of the beetles
was not observed. Thrips were observed walking on the
inflorescences and creeping into the corolla between the
petals as soon as the petals unfurled at anthesis. However,
after that time they rarely emerged from the corollas (Fig.
5). Thirty percent (N 5 30) of beetles and 12% (N 5 97)
of thrips carried pollen on their bodies.
The estimated number of thrips per flower (0.31) was
1.7 times larger than beetles (0.18). However, the beetles
were three times more numerous at night than in the daytime, and their number was almost equal to the number
of thrips at 0000 (Fig. 6). Changes in the numbers of
thrips samples were smaller than for beetles. Higher density of beetles at night than in the daytime and smaller
fluctuation in thrips density than that of beetles were
demonstrated by both sampling methods, net-sweeping
and flower collection.
The number of pollen grains on the stigmas of openpollinated flowers gradually increased following anthesis,
particularly between 2200 and 0300, and rank correlation
between sampling time and the number of pollen grains
was significant (Spearmans coefficients of rank correlation: rs 5 0.24, P 5 0.003, Fig. 7). The number of pollen
grains per stigma showed great variation within each
sample, and 47% (N 5 17) of the flowers collected at
1730, just before anthesis, already had some self pollen
on the stigma. The number slightly decreased between
0300 and 0600.
Changes in fruit numbersImmature fruits on open
inflorescences aborted continuously until day 48 after the

OF

BOTANY

[Vol. 86

Fig. 7. Changes in the number of pollen grains on a stigma after

flower opening. Maximum and minimum in each sample are shown by
the upper and lower ends of vertical bar, 75 and 25% points are given
by the upper and lower ends of box, and mode is shown by horizontal
bar in the box. Considerable increase of pollen grains is indicated by
an asterisk.

flowering peak (Fig. 8). Only 0.7% of the original ovaries

remained by that time. The first viable fruits were collected 93 d after flowering (23 August). If all fruits fallen
before 23 August were immature, only 0.17% of the
flowers gave rise to mature fruits. The number of fruits

Fig. 8. Changes in percentage of the fruits remaining on untreated,

bagged, and open inflorescences. Arcsine transformation is used in the
axis of percentage. Time point for comparison of fruit set is indicated
by a dotted line.

January 1999]

SAKAI

ET AL.BEETLE POLLINATION OF

SHOREA PARVIFOLIA

67

TABLE 5. Fruit set of the flowers in pollination experiments 37. Percentages of the total flowers (N) remaining on the inflorescences
on day 32 after flowering are shown. Significance of difference
from fruit set in untreated, bagged flowers (experiment 2) was examined by Fishers exact test (one-sided).

Experiment

3
4
5
6
7

Self-pollinated (geitonogamous)
Cross-pollinated
Open, flower reduced
Beetle-introduced
Thrips-introduced

% fruit
remaining
on day 32

Significance
of
deviation
(P)

176
345
323
529
368

0.00
1.74
2.17
1.13
0.54

0.636
0.003
0.000
0.019
0.327

than the natural condition observed by net-sweeping and

flower collection on tree 225 (0.18 beetles and 0.31
thrips). The numbers of fruits fertilized per insect were
0.014 for beetles and 0.009 for thrips.
DISCUSSION

remaining on tree 229 decreased more quickly. However,

mature fruits remained ;10 d longer on tree 229 than on
tree 225 (Fig. 8).
Abortion in untreated, bagged inflorescences was much
higher than in open inflorescences (Fig. 8). The percentages of fruits per bagged inflorescence at day 48 after
flowering peak (0.05%) was significantly lower than that
in open inflorescences (0.67%) (Fishers exact test; P ,
0.001). Only one fruit matured in bagged inflorescences.

Breeding system of S. parvifoliaThe results of the

pollination experiments show that the study tree received
enough pollination service for outcrossing. Very low fruit
set was characteristic of all treatments and has generally
been observed by us and others (e.g., Chan, 1981). Fruit
set was significantly increased by application of cross
pollen, but fruit set did not change by application of geitonogamous pollen. The results demonstrate that S. parvifolia is strongly self-incompatible, as are other species
of Shorea (Chan, 1981) and related genera (Momose, Nagamitsu, and Inoue, 1996). This is further supported by
the fact that only a small proportion of fruits from untreated, bagged flowers remained until just before mature
fruit dispersal, although nearly half of the flowers sampled just before anthesis had pollen grains on the stigma.
In spite of the strong self-incompatibility, fruit set was
not different between open- and cross-pollinated flowers,
probably because all functional pollination in the open
treatment was cross pollination.

Comparison of fruit setThe number of fruits decreased considerably up to days 2124 after flowering in
experiments 37, but the numbers more or less stabilized
thereafter (Fig. 9). Observed fruit set was highest in open,
flower-reduced inflorescences (2.17%, N 5 323), followed by that in cross-pollinated ones (1.74%, N 5 345),
and fruit set in both experiments was significantly higher
than in untreated, bagged inflorescences (0.27%) (P ,
0.001 for open, flower-reduced; P 5 0.003 for cross-pollinated; Table 5). However, fruit set in self-pollinated
(geitonogamous) inflorescences did not differ from that
in bagged inflorescences.
Fruit set of both beetle-introduced (1.13%) and thripsintroduced flowers (0.54%) was between that of open
flowers (1.44%) and untreated, bagged flowers (0.27%).
Flowers in beetle-introduced bags had significantly higher fruit set than did untreated, bagged flowers (P 5
0.019), but flowers in thrips-introduced bags did not (Table 5). Densities of introduced insects per flower per day
were 0.81 for beetles and 0.60 for thrips. These are higher

The pollinators of S. parvifoliaSmall beetles, particularly Chrysomelidae and Curculionidae, were the major flower visitors of Shorea parvifolia collected by the
net-sweeping method on the three study trees, and the
spectra of beetle families were not different among trees
except for the occurrence of Corylophidae on one tree.
On the other hand, thrips were the most abundant insects
collected by flower collection. The density of thrips in
the crown was greater than that of beetles if averaged
throughout a day, but beetles and thrips were equally
abundant at night. Two other lines of evidence, higher
mean pollen loads on beetles than on thrips and more
frequent movements of beetles among flowers than thrips,
suggest higher potential contribution of beetles to crosspollination than thrips. However, the contribution of
flower visitors to pollination cannot be measured only by
their visitation frequency (Schemske and Horvitz, 1984,
and references therein), and the amount of body pollen
is not always a good index of their ability to effect fruit
set (Inouye et al., 1994). To evaluate this parameter, we

Fig. 9. Changes in percentage of the fruits remaining on open, flower-reduced, self- (geitonogamous) and cross-pollinated, and beetle- and
thrips-introduced inflorescences resulting from pollination experiments
37.

68

AMERICAN JOURNAL

performed introduction experiments of the two major

flower visitors.
Bagged flowers onto which beetles had been introduced showed significantly higher fruit set than untreated,
bagged flowers, while bagged flowers onto which thrips
had been introduced did not. This demonstrates that the
introduced beetles were more successful than thrips at
depositing pollen on stigmas and fertilizing the flowers.
The number of flowers fertilized per beetle was 1.5 times
higher than that per thrips.
What was not incorporated into the experiments was
proportions of self- and cross-pollen that insects carry
under natural conditions, namely, the proportion of beetles or thrips individuals that move between conspecific
trees each day because the beetles and thrips used in the
experiments were artificially transferred between trees.
Though movements of the beetles and thrips among trees
were not observed, beetles seem more likely to move
between conspecific trees than thrips. The number of beetles increased threefold at night compared to daytime, in
spite of persistence of many corollas for .24 h in the
crowns. Some thrips remained in fallen corollas.
The other problem is that net-sweeping may introduce
pollen load artifacts for flower visitors. It may affect both
the numbers of individuals with pollen load, and the result of experiment 7, in which flower visitors collected
by net-sweeping were introduced into bagged inflorescences. However, even if the sampling caused pollen load
artifacts, we believe that the difference between beetles
and thrips in a pollen load is still meaningful because
beetles and thrips were collected in the same method.
Nearly half of the flowers had pollen on their stigma
before anthesis, and a significant increase of pollen grains
was observed following anthesis. The slight decrease at
1200 may be caused by earlier drops of fertilized flowers
than unfertilized ones. The activity of thrips, which had
a density of only 0.31 per flower, could not be responsible
for all the pollen on the stigmas. Dehiscence of the anthers before anthesis probably permits pollen to be shed
on the stigma as well as inner surface of the corolla,
though this is paradoxical in an obligate outcrosser. Scattered pollen on the inner surface of the corolla may be
more easily attached to flower visitors than pollen remaining in the anthers. The increase of pollen grains on
the stigma following anthesis may be brought about
mainly by the beetles. Thrips were most active at the
onset of anthesis, and between-flower movement of thrips
was not observed thereafter, while the beetles were observed flying around the flowers throughout the night.
The contribution of beetles as pollinators at our study
site, Lambir, Sarawak, was confirmed by the beetle-introduction experiment. The situation in Lambir seems to be
different from that at Pasoh, Peninsular Malaysia, from
where Appanah and Chan (1981) reported thrips pollination in six species of Shorea sect. Mutica. Beetles accounted for one-third of flower visitors at Lambir, and
the density of thrips (0.31 thrips per flower) was much
lower than that at Pasoh (2.4) where most flower visitors
were thrips. However, we cannot eliminate the possibility
that thrips contributed to pollination.
Thrips often occur as general pollen feeders on various
plant species (Kevan and Baker, 1983; Kirk, 1984).
Thrips hawaiiensis, the most abundant thrips species on

OF

BOTANY

[Vol. 86

S. parvifolia at Lambir, has been collected on oil palm

flowers (Elaeis guineensis, Palmae) in a plantation in
Peninsular Malaysia (Syed, 1979) and on Lantana camara (Verbenaceae) in India (Mathur and Mohan Ram,
1978). They usually play a minor roll as pollinators, although predominant thrips pollination has been reported
for Calluna (Ericaceae) in the Faeroes, where weather
condition and lack of larger insects prohibit pollination
by other insects (Hagerup, 1950), and for Annonaceae
(Webber and Gottsberger, 1995; Momose, Nagamitsu and
Inoue, in press), Araceae (Rust, 1980), Lauraceae (Norton, 1984), and Winteraceae (Thien, 1980; Pellmyr et al.,
1990) in other regions with rich insect fauna. For a thripspollinated tree species in the understory of Lambir forest,
Popowia pisocarpa (Annonaceae), Momose, Nagamitsu,
and Inoue (in press) suggested limited pollen dispersal by
thrips based on low fruit set of isolated trees. Thrips, with
their oar-like wings, may be dispersed by local air drafts,
including convectional movements, which are frequent in
tropical forest canopies, especially during the day, but
seldom penetrate the understory (Richards, 1997).
S. parvifolia and beetle pollinatorsFlowers of S.
parvifolia have a character adapted to beetle pollination.
The apical edge of their five petals is thin and soft, and
damage by beetles is concentrated in that part of the petals. The apical region of the petals may be more attractive
than stigmas or ovaries for the beetles. Moving from one
petal to another of a flower or feeding on pollen on the
inner surface of the corolla, beetles acquire pollen loads,
and then deposit pollen on the stigma. These edible rewards for the beetles may promote pollination by both
rewarding beetle visits and by reducing potential damage
to stigmas and ovaries.
Beetles pollinate a wide range of plant species with
various reproductive characters, and the specificity of the
plant-pollinator interaction also varies (Endress, 1994).
Throughout the Lambir forest, many plant species are
reported to be pollinated by beetles, especially in the Annonaceae (Momose, Nagamitsu, and Inoue, in press).
These beetle-pollinated Annonaceae have more or less
specialized relationships with their beetle pollinators and
offer stigmatic secretions and/or mating sites to their pollinators (e.g., Gottsberger, 1989, 1990; Momose, Nagamitsu, and Inoue, in press). In contrast, many beetle species
were collected on the flowers of S. parvifolia, though a
few species accounted for most visits. The flowers are
exposed in the emergent layer and never act as a refuge
for the beetles. Mating behavior of the beetles was not
observed.
The life histories of Monolepta species, the beetle genus that was most abundant on the flowers of S. parvifolia, remain unknown. One possibility is that they feed
on young leaves of Shorea and other dipterocarps, which
are available all year-round even in seasons out of general
flowering periods. These beetles were collected outside
the general flowering period by beating on Shorea leaves
(M. Kato, unpublished data).
Plantpollinator interactions in Shorea at Lambir appear to be different from that observed in lowland dipterocarp forests in Peninsular Malaysia: there, Shorea
species in the same section are pollinated by common
pollinators, and the species in each section flower se-

January 1999]

SAKAI

ET AL.BEETLE POLLINATION OF

quentially. In the case of Shorea sect. Mutica, sequential

flowering brings about drastic increase of their pollinators, thrips, with a extraordinary short generation time
(;8 d). Thrips populations large enough for their pollination are established in a few weeks (Ashton, Givinish,
and Appanah, 1988; Appanah, 1990, 1993). At Lambir,
in contrast, beetles were collected on the flowers of, and
may pollinate nine Shorea species including species of
sect. Mutica and other sections (Momose et al., 1998).
Thrips densities on their flowers were much lower than
that observed in Shorea at Pasoh (Appanah and Chan,
1981; Sakai, unpublished data). How can beetles with a
longer generation time than thrips offer enough pollination service to these Shorea species, which flower only
in general flowering periods? Monolepta species collected on S. parvifolia flowers were found rather constantly
in monthly light trap samples (T. Itioka, Nagoya University, unpublished data). Instead of such a drastic multiplication as thrips show, the beetles probably respond to
an abrupt increase of floral resource in a general flowering by changing foods from dipterocarp leaves to flowers. For irregular, infrequent reproductions, Shorea species may rear the pollinators during nongeneral flowering
periods by providing their leaves to the pollinators.
LITERATURE CITED
APPANAH, S. 1990. Plantpollinator interactions in Malaysian rain forests. In K. S. Bawa and M. Hadley [eds.], Reproductive ecology
of tropical forest plants, 85102. UNESCO, Paris and the Parthenon Publishing Group, Lancs.
. 1993. Mass flowering of dipterocarp forests in the aseasonal
tropics. Journal of Bioscience 18: 457474.
, AND H. T. CHAN. 1981. Thrips: the pollinators of some dipterocarps. Malaysian Forester 44: 234252.
ASHTON, P. S. 1982. Dipterocarpaceae. Flora Malasiana, Series 1, 9:
237552.
. 1988. Dipterocarp biology as a window to the understanding
of tropical forest structure. Annual Review of Ecology and Systematics 19: 347370.
, T. J. GIVNISH, AND S. APPANAH. 1988. Staggered flowering in
the Dipterocarpaceae: new insights into floral induction and the
evolution of mast fruiting in the aseasonal tropics. American Naturalist 132: 4466.
CHAN, H. T. 1981. Reproductive biology of some Malaysian dipterocarps. III. Breeding systems. Malaysian Forester 38: 160170.
, AND S. APPANAH. 1980. Reproductive biology of some Malaysian dipterocarps. Malaysian Forester 44: 2836.
ENDRESS, P. K. 1994. Diversity and evolutionary biology of tropical
flowers. Cambridge University press, Cambridge.
GOTTSBERGER, G. 1989. Beetle pollination and flowering rhythm of
Annona species (Annonaceae) in Brazil. Pant Sytematics and Evolution 167: 165187.

SHOREA PARVIFOLIA

69

. 1990. Flowers and beetles in the South American tropics. Botanical Acta 103: 360365.
HAGERUP, O. 1950. Thrips pollination in Calluna. Biologiske Meddelelser Kongelige Danske Videnskabernas 18: 116.
INOUYE, D. W., D. E. GILLE, M. R. DUDASH, AND C. B. FENSTER. 1994.
A model and lexicon for pollen fate. American Journal of Botany
81: 15171530.
INOUE, T., T. YUMOTO, A. A. HAMID, H. S. LEE, AND K. OGINO. 1995.
Construction of a canopy observation system in a tropical rainforest
of Sarawak. Selbyana 16: 100111.
KEVAN, P. G., AND H. G. BAKER. 1983. Insect as flower visitors and
pollinators. Annual Review of Entomology 28: 40753.
KIRK, W. D. J. 1984. Pollen feeding in thrips (Insecta: Thysanoptera).
Journal of Zoology 204: 10717.
MATHUR, G., AND H. Y. MOHAN RAM. 1978. Significance of petal colour
in thrips-pollinated Lantana camera L. Annals of Botany 42: 1473
1476.
MOMOSE, K., T. NAGAMITSU, AND T. INOUE. 1996. The reproductive
ecology of an emergent dipterocarp in a lowland rain forest in
Sarawak. Plant Species Biology 11: 189198.
, , AND . In press. Thrips cross pollination of Popowia pisocarpa (Annonaceae) in a lowland dipterocarp forest in
Sarawak. Biotropica.
, T. YUMOTO, T. NAGAMITSU, M. KATO, H. NAGAMASU, S. SAKAI,
R. D. HARRISON, T. ITIOKA, A. A. HAMID, AND T. INOUE. 1998.
Pollination biology in a lowland dipterocarp forest in Sarawak, Malaysia. I. Characteristics of the plant-pollinator community in a lowland dipterocarp forest. American Journal of Botany 85: 1477
1501.
NORTON, S. A. 1984. Thrips pollination in the lowland forest of New
Zealand. New Zealand Journal of Ecology 7: 157164.
PELLMYR, O., L. B. THIEN, G. BERGSTORM, AND I. GROTH. 1990. Pollination of New Caledonian Winteraceae: opportunistic shifts or
parallel radiation with their pollinators? Plant Systematics and Evolution 173: 143157.
RICHARDS, P. W. 1997. The tropical rain forest, 2d ed. Cambridge University Press, Cambridge.
RUST, R. W. 1980. Pollen movement and reproduction in Arisaema
triphyllum. Bulletin of the Torrey Botanical Club 107: 539542.
SAKAI, S., K. MOMOSE, T. YUMOTO, T. NAGAMITSU, H. NAGAMASU, A.
A. HAMID, T. NAKASHIZUKA, AND T. INOUE. In press. Plant reproductive phenology over four years including an episode of general
flowering in a lowland dipterocarp forest, Sarawak, Malaysia.
American Journal of Botany 86.
SCHEMSKE, D. W., AND C. C. HORVITZ. 1984. Variation among floral
visitors in pollination ability: a precondition for mutualism specialization. Science 225: 519521.
SYED, R. A. 1979. Studies on oil palm pollination by insects. Bulletin
of Entomological Research 69: 213224.
THIEN, L. B. 1980. Patterns of pollination in the primitive angiosperms.
Biotropica 12: 113.
WEBBER, A. C., AND G. GOTTSBERGER. 1995. Floral biology and pollination of Bocageopsis multiflora and Oxandra euneura in Central
Amazonia, with remarks on the evolution of stamens in Annonaceae. Feddes Repertorium 106: 515524.
WHITMORE, T. C. 1984. Tropical rainforests of the Far East, 2d ed.
Clanton Press, Oxford.