Thursday 24 December 2015

[Glinus.com] Thrips pollination of the dioecious ant plant Macaranga hullettii(Euphorbiaceae) in Southeast Asia 1

 

Thrips pollination of the dioecious ant plant Macaranga hullettii(Euphorbiaceae) in Southeast Asia 1

  1. Ulrich Maschwitz2

+Author Affiliations

  1. 2Department of Zoology, J.W. Goethe-University of Frankfurt, Siesmayerstr. 70, D-60054 Frankfurt, Germany;
  2. 3Department of Animal Ecology and Tropical Biology, Biozentrum, University of Würzburg, Am Hubland, D-97074 Würzburg, Germany; and
  3. 4Department of Integrative Biology, University of California at Berkeley, Berkeley, California 94720-3140 USA
  • Received for publication 12 April 2001.
  • Accepted for publication 27 July 2001.

ABSTRACT

Discussion about thrips (Thysanoptera) as main pollinators has been controversial in the past because thrips do not fit the preconception of an effective pollinator. In this study, we present evidence for thrips pollination in the dioecious pioneer tree genus Macaranga (Euphorbiaceae). Macaranga hullettii is pollinated predominantly by one thrips species, Neoheegeria sp. (Phlaeothripidae, Thysanoptera). As a reward for pollinators, the protective floral bracteoles function as breeding sites for thrips and trichomal nectaries on the adaxial surface of the floral bracteoles provide alimentation. Flowering phenology of both staminate and pistillate trees was highly synchronized within 3–4 wk periods. In contrast to pistillate trees, staminate trees start to breed the thrips inside the developing inflorescences ∼2 wk before anthesis. Breeding of Neoheegeria sp. in the laboratory indicates that the thrips development is completed within ∼17 d. Thus, staminate trees offer breeding sites for one thrips generation until the onset of pollen presentation. Intraspecific pollen transfer by thrips was proved by pollen loads of thrips taken from receptive pistillate inflorescences of M. hullettii. Bagging experiments of different mesh sizes showed that seed set reached almost the level of open-pollinated flowers when exclusively tiny insects like thrips were able to enter the net bags, but no apomictic seed set occurred when no insect access was given to the flowers.

Key words:

Macaranga Thou. (Euphorbiaceae) is the world's largest genus of pioneer trees and, though relatively uncommon 100 yr ago, its species now dominate locally disturbed areas in Southeast Asia (Whitmore, 1990). The enormous expansion in range of Macaranga species presents an interesting situation for biosystematic studies (Whitmore and Burnham, 1984). In a recent study on phenology and fecundity of Macaranga spp. in Sarawak (Borneo), Davies and Ashton (1999) hypothesized that these Macarangaspecies appear to share pollinators because the inflorescence morphologies are uniform. In the only study known to us dealing with aspects of Macarangapollination, Taylor (1982) found thrips (Thysanoptera) to be the predominant flower visitors of both staminate and pistillate trees of Macaranga spp. in Malay Pensinsula and proposed thrips to be potential pollinators. Since then, thrips have been briefly mentioned as floral dwellers exclusively of staminate inflorescences in Macaranga velutiniflora from Sarawak (Davies, 1999), but no evidence for thrips pollination has been provided. Other observations of flower visitors (trigonid bees) of Macaranga have only been anecdotal (Momose et al., 1998), and how pollination in Macaranga occurs has remained unclear.

Thrips are tiny insects with piercing/sucking mouthparts, and they are often found on flowers of various plant species, where they feed on nectar, pollen, and/or plant tissues. In the cases in which they have been seen to have a role in pollination, they are usually regarded as minor, accessory pollinators (Kirk [1997] and references therein). However, Kirk (1997) hypothesized that thrips as pollinators have been systematically overlooked because they do not fit the traditional profile of an effective pollinator. Indications for thrips as main pollinators are given only for a few plant species, e.g., in Annonaceae (Webber and Gottsberger, 1995; Momose, Nagamitsu, and Inoue, 1998; Gottsberger, 1999), Chloranthaceae (Luo and Li, 1999), Dipterocarpaceae (Appanah and Chan, 1981; but see Sakai et al., 1999), Monimiaceae (Williams, Adam, and Mound, 2001), Winteraceae (Thien, 1980), and in the cycad family Zamiaceae (Mound and Terry, 2001).

In this study, we aim to provide evidence for thrips pollination in the genusMacaranga. To attain this, one characteristic species of the sectionPachystemonMacaranga hullettii, was exemplary investigated in a detailed study. Most members of this section are myrmecophytes (ant-plants) with similar floral traits. For obligate myrmecophytes, such as M. hullettii, there is a potential conflict with regard to the pollination process. Because plant ants usually tend to attack intruders on their host plant, they may repel potential pollinators. Thus, effective barrier mechanisms against ant guards during the time of anthesis are required to protect the pollination process (Willmer and Stone, 1997). However, the role of the partner ants in the pollination of its host plant is not part of this study and will be described in future publications.

Here, we investigate the pollination biology of Macaranga hullettii (1) by monitoring the floral phenology and fecundity, (2) by observing and collecting flower visitors, (3) by describing floral morphology and rewards for pollinator, (4) by measuring pollen load of flower visitors, and (5) by conducting pollinator-exclusion experiments.

MATERIALS AND METHODS

Study species

The palaeotropical genus Macaranga Thou. (Euphorbiaceae) includes >250 dioecious shrub and tree species (Webster, 1994). Of the 52 Macaranga tree species known from the Malay Archipelago, Macaranga hullettii King ex Hook f. (section Pachystemon) is one of at least 23 species that are colonized by specific ants (Fiala et al., 1999; see references therein). Macaranga hullettii, occurring on Peninsula Malaysia, Borneo, and Sumatra, is a pioneer tree that grows in small gaps, along riverbanks, roadsides, and in logged areas (Whitmore, 1969). The small trees of M. hullettii have axillary, inconspicuous green inflorescences, which aggregate together with the leaves at the ends of the shoots. Both staminate and pistillate inflorescences are covered by caducous bracts. The staminate inflorescences are many-branched panicles that are formed by clusters of tiny apetalous unistaminate flowers. Each cluster is subtended by a green bracteole. Pistillate inflorescences differ from staminate flowers in being stouter, unbranched, and in having fewer flowers, which are also apetalous, and each flower is subtended by a bracteole. At the time of maturity, fruits are horned, and the horned carpels drop to present the showy seeds, which are covered by a fleshy, bright-pink aril.

Study site

Studies on pollination biology were conducted in 1998 (January–May and September–October) and in 1999 (March–July) mainly in the Ulu Gombak Valley (3°17′ N, 101°44′ E), ∼30–40 km northeast of Kuala Lumpur, West Malaysia. The study area is mostly covered with advanced secondary forest of the lowland dipterocarp type, at elevations of 200–400 m above sea level (asl). Additional observations were made in West Malaysia near Bentong (Pahang) and in the submontane areas of Genting Highlands and Fraser's Hill (Pahang) at elevations up to 1100 m asl, as well as in Borneo (Brunei, Temburong district).

Reproductive phenology

Along a road transect of 8 km in Ulu Gombak, 48 mature trees were randomly selected. The reproductive stage of 22 pistillate and 26 staminate M. hullettiiwas monitored from beginning of March to mid-July 1999. All M. hullettii trees were scored weekly whether they had reproductive inflorescences or not. Staminate inflorescences were recorded as reproductive when they lost the outer bracts, and the inflorescences were elongated. In this stage, pollen presentation was observed. Receptivity of pistillate inflorescences was recorded as the time period between the opening of the outer bracts and the abscission of the floral bracteoles. In this stage, stigmata are elongated, and the stigmatic surface is freshly green, whereas the stigmata fade when the inflorescences lose floral bracteoles.

Floral traits and anthesis

In order to estimate number of inflorescences and flowers per tree, a total of 13 reproductive trees (8 female and 5 male) were selected in March 1999. Within the constraints imposed by accessibility, the trees chosen ranged from 3 to 6 m in height. For each tree, counts were made of the total number of inflorescences and the number of flowers per inflorescence (the latter was obtained from two staminate and five pistillate inflorescences per tree). Total flower number per tree was then estimated as the product of the mean flower number per inflorescence and the total inflorescence number.

The flowering stages of two staminate and two pistillate trees were monitored daily over a period of 30 d (March 1999). On each tree, two flowering branches with ten inflorescences each were tagged, and the flowering stage of the inflorescences was recorded daily. Additionally, the flowering stages of the remaining inflorescences of the trees was estimated. Flowering development of both staminate and pistillate inflorescences were subdivided into the following stages: (1) closed buds = buds that are still enclosed by the outer bracts without any opening; (2) open buds = developing immature inflorescences with narrow bract openings, for staminate inflorescences only; (3) mature inflorescences = pollen presentation of staminate flowers or pistillate inflorescences with bract openings; (4) post-anthesis = abortion of floral bracteoles and beginning of fruit development in pistillate inflorescences, and abscission of staminate inflorescences.

Floral odor production was assayed subjectively by the sense of smell. To detect differences of odor production between parts of inflorescences, bracteoles and flowers were transferred separately in a test tube for 5 min to let any fragrance accumulate. Samples were made of different staminate and pistillate trees, and of one pistillate individual at different daytimes (1900, 0100, 0700, 1400 and 1700) on 13–14 February 1998. To investigate for pollinator reward, histological anatomy and structure of the floral bracteoles (the feeding site of floral visitors [see below]) of both staminate and pistillate inflorescences were investigated by optical and scanning electron microscopy (SEM). The bracteole tissue was tested for lipids, oils, waxes, free fatty acids (Sudan black, Sudan IV), amino acids (ninhydrin test), and sugar (phenol + sulphuric acid). The bracteoles were stained completely, and afterwards, section preparations were inspected under the light microscope (for methods see Kearns and Inouye, 1993). Additionally, we tested for nectar secretion in the field by using glucose test strips. To avoid damaging floral tissue during handling, a droplet of water was dripped on a bracteole directly at the tree and then the solution was absorbed with glucose test paper.

Collection and observation of floral visitors

Flower-visiting insects were hidden inside the enclosed inflorescences and thus were collected by sampling inflorescences. We sampled 85 staminate and 330 pistillate inflorescences from a total of 27 staminate and 14 pistillateMacaranga hullettii trees. Sampling was stretched over the whole study area. The great majority of collections were made in the Ulu Gombak valley during January–May 1998 and March–July 1999, with one sample coming from a solely flowering staminate tree in September 1998. Additional samples were taken from West Malaysia in Fraser's Hill (two staminate individuals in 1998 and three in 1999), Bentong (one staminate and one pistillate individual), Genting Highlands (two pistillate and one staminate individual in 1998 and two staminate individuals in 1999), and Borneo (Brunei, Temburong, one staminate individual).

Usually, flower visitors were collected by first putting a plastic bag over an inflorescence and then cutting the whole inflorescence to reduce a possible loss of floral visitors during handling. If the flowering shoots were out of direct reach, the branches were cut using a pull clipper on a telescopic pole, and then the inflorescences were quickly transferred to plastic bags. After chloroform was added, all insects were picked out of the inflorescences by carefully taking apart all bracteoles under a binocular microscope. The insects were counted, classified to order, and fixed in a solution of 60% ethyl alcohol, glycerin, and glacial acetic acid. A number of Thysanoptera were mounted following the methods of Lewis (1973) for microscopic examination and determination. Voucher specimens were deposited at the Senckenberg Museum of Frankfurt (Germany) and in the collection of the first author.

The mean number of flower-visiting insects per "pollination unit" (i.e., floral bracteole, which covers one flower in pistillate trees and one flower cluster in staminate trees) was calculated for each sex by dividing the mean number of flower visitors per inflorescence (counted from 85 staminate and 330 pistillate inflorescences) by the mean number of flowers per inflorescence (counted from 12 staminate and 65 pistillate inflorescences; see above).

To test variation of flower visiting insects at different times during the day, 115 inflorescences from one pistillate tree were collected over a 24-h period on 13–14 February 1998. Every hour, five inflorescences of different shoots were sampled. On the same day, staminate inflorescences (N = 18) growing 15 m apart on one tree were also collected hourly (two per collection) until no further inflorescences were accessible (from 1900 until 0300).

Observations of flower-visiting insects were conducted during 1-h intervals (N = 20) at different day times on both male and female trees. Additional observations were made during collections. Behavior of thrips inside pistillate and staminate inflorescences was observed in the field by carefully spreading apart the floral bracteoles to extend the openings. In the laboratory, thrips adults and larvae, which were collected alive out of inflorescences, were put together with fresh floral parts between slides of glass, which allowed us to use the binocular microscope to observe thrips feeding.

Breeding experiments

Thirty thrips eggs collected from staminate and pistillate inflorescences of M. hullettii were transferred one-by-one into a glass tube (4 mm in diameter and 20 mm in length) and reared to maturity. Every day, a fresh bracteole of a female M. hullettii was added and the progress of thrips development recorded. Glass tubes were kept in the shade close to the source plants at a natural fluctuating temperature with a mean of ∼26°C. The tubes were closed with damp cotton to reduce desiccation.

Pollen transfer

In order to assess pollen transport by thrips, all individuals collected from 100 pistillate inflorescences of two M. hullettii trees (the next flowering staminate trees were ∼10 and 25 m apart) were counted and checked for pollen load under a microscope. Subsequently, all pollen-carrying thrips were processed for SEM to verify if the thrips were actually transporting intraspecific pollen. We first compiled a pollen catalog of all sympatric Macaranga species in the study area to identify pollen carried by thrips to species level. Pollen quantity carried by thrips was assessed from 25 thrips adults collected from staminate inflorescences during anthesis. Additionally, the number of pollen grains of five anthers of two staminate individuals was counted under a light microscope with a counting chamber.

Seed set and bagging experiment

Seed set of 13 pistillate trees was examined 2 mo after anthesis, ∼4 wk before fruit abortion. All infructescences per tree were harvested, and the fruits were counted. In addition, the number of ovaries of 50 fruits was counted for each tree as the sum of the developing seeds and the remaining undeveloped ovaries. Total ovary number before anthesis then was estimated as the product of mean ovary number per gynoecium, mean flower number per inflorescence, and the number of inflorescences per tree. Seed number per tree was estimated as the product of mean seed number per fruit and the fruit number per tree.

In order to test whether thrips indeed pollinate M. hullettii, we conducted an exclusion experiment. Of three pistillate trees, three branches each were bagged with nets of two different mesh sizes ∼4 wk before anthesis. Unfortunately, the net bags of two trees were destroyed, and therefore only data of one tree remained for evaluation. Three branches of this tree, bearing nine inflorescences, were bagged with coarse net (mesh size 1 mm), and another three branches (seven inflorescences) were bagged in fine gauze (mesh size 0.2 mm). Untreated flowering branches served as controls (ten inflorescences). Prior to experimentation, we tested the permeability of different mesh sizes for insects collected from staminate inflorescences. A mesh size of 0.2 mm created an effective barrier against even tiny thrips and a mesh size of 1 mm acted as an effective barrier for all tested insects (weevils and bugs; for further explanation see below) except thrips. Four weeks after bagging, at the time of anthesis, the flowers per inflorescence were counted. Ovary number then was estimated as the product of the flower number and the mean ovary number per flower (estimated as described above). The infructescences were harvested 76 d after bagging, and the fruits and developing seeds were counted.

RESULTS

Flowering phenology

The number of the monitored trees did not differ significantly between sexes (χ2 test, P = 0.4). Flowering phenology of both male and female trees was highly synchronized (Fig. 1). From March to July 1999, three flowering events occurred lasting 3–4 wk each. While all individuals were in bloom during the first flowering period only some of these trees flowered a second or even a third time (32% of pistillate and 65% of staminate individuals flowered more than once). Within the 5-mo flowering episode, staminate trees flowered significantly more frequently than pistillate trees (G test, P = 0.03). Therefore, the sex ratio of the two flowering times following the first tended to be biased toward male trees.

Fig. 1. Reproductive phenology of M. hullettii in the Ulu Gombak valley, West Malaysia, from March to July 1999, as the proportion of flowering trees (N= 22 pistillate and N = 26 staminate individuals)

In 1998, detailed phenological studies could not be carried out, but trees of M. hullettii were also observed flowering synchronously three times in a 5-mo period from January to May 1998. Additional examinations in a 2-mo interval for flowering or fruit set from May 1998 to March 1999 and from July to December 1999 give indications that only single trees flowered within these time intervals. We found one staminate tree flowering solely in September 1998.

Floral traits and anthesis

Anthesis occurred during a period of 15–18 d in pistillate and 12–15 d in staminate individuals. Within these time intervals, all inflorescences per tree had developed mature flowers. Anthesis intervals measured per inflorescence were slightly shorter, i.e., within a single staminate inflorescence, anther dehiscence occurred in acropetal sequence within 9–13 d, whereas receptivity of flowers within a single pistillate inflorescence was estimated to extend over 12–14 d. In staminate trees, anthesis was preceded by an "open bud stage" (immature inflorescences with narrow bract openings) that lasted 11–16 d.

Mature staminate panicles measured 8–14 cm in anthesis and each panicle consisted of 72–136 partial florescences (mean = 93, N = 12), which were formed by clusters of 13–81 (mean = 48.9, N = 10) tiny apetalous flowers (∼1 mm in length). Within an inflorescence, the number of flowers per cluster decreased from the base to the apex. The floral bracteoles of staminate inflorescences measured ∼5 × 7 mm, and each bracteole tightly covered a flower cluster ("pollination unit"), even during the time of anthesis. The narrow openings (0.2–0.8 mm) along the bracteole margin allowed only for very small insects access into the pollination chambers. These openings result from "gaps" formed by the toothed margin of the bracteoles (Fig. 2). Only at the end of anthesis do some bracteoles (usually the proximal ones of an inflorescence) gradually relax their contact with the stalk, creating a wider gap of ∼2 mm.

Fig. 2. Part of a staminate inflorescence with bracteoles and "gaps" (arrows) allowing access for thrips. One bracteole is extended to show aNeoheegeria thrips inside the "bracteole chamber."

Pistillate inflorescences consist of 8–14 single flowers (mean = 10.5, N = 65). The latter contain 4–5 (rarely three or six) ovaries (mean = 4.5, N = 632). The floral bracteoles are ∼6 × 11 mm in size and, in contrast to staminate trees, each bracteole covers one flower only. The openings leading into the pollination chamber are less narrow than those of staminate trees (0.2–3 mm), and they gradually open during anthesis.

Both staminate and pistillate trees average about the same number of inflorescences (52.6 inflorescences for N = 5 staminate trees; 51.4 inflorescences for N = 8 pistillate trees). However, due to the much higher flower number per inflorescence in staminate trees, the total number of flowers per tree of staminate individuals (mean = 239 209) is much higher than for pistillate individuals (mean = 540). In terms of "pollination units" per tree, the difference between staminate and pistillate trees is less pronounced (4892 vs. 540 pollination units/tree). In staminate individuals, the anthers dehisce poricidal and pollen release occurs in clumps through the apex of the calyx tube. On average, one staminate flower produces 2209 pollen grains (±234.3 SD; N = 5). This corresponds to an estimated number of pollen grains per ovules (P/O) within a population of 2.2 × 100 000 (with ∼1 : 1 sex ratio, the P/O = [product of staminate flower number and pollen grain number]/[product of pistillate flower number and ovule number]).

Pollen of M. hullettii is tricolporate and noticeably small, measuring only 11 μm in diameter (Fig. 3). The tectum is not sculptured and is finely perforated and smooth. Sequential anthesis in staminate flowers means that each flower cluster presents pollen over several days. After dehiscence, the pollen grains form loose aggregations held together by pollenkitt, which gather around the flowers and at the base of the adjacent bracteoles. In this area, the bracteoles are densely covered with unicellular trichomes (Fig. 4). Pistillate inflorescences possess the same kind of trichomes at the base and along the median of the adaxial surface of the bracteoles. Preliminary analytical tests indicate that these trichomes mainly contain sugars (tested with sulphuric acid + phenol) and nectar secretion was proved with glucose test strips. Therefore, we assume that nectar serves as a reward for pollinators.

Figs. 3–5. Flower parts and insect visitor of M. hullettii3.Pollen grain. 4. Trichomal nectaries on the adaxial surface of a bracteole. 5. Neoheegeriathrips collected from a pistillate inflorescence. Arrows point to pollen grains adhering to thrips body

Both staminate and pistillate inflorescences release during all times of day an agreeably sweet vanilla-like scent. The bracteoles produce the strongest odor. Flowers from which the bracteoles were removed had a weak odor. We judge floral fragrance to be the primary attractant because the small, greenish flowers are not visually striking. Additionally, the bagging experiment demonstrated that thrips enter the flowers, even when they are hidden in net bags (see below). This indicates that as an attractant for thrips, visual cues play a minor role.

Floral visitors

In total we collected 2848 and 710 insects out of 85 staminate and 330 pistillate inflorescences, respectively (Table 1). Inflorescences of both sexes were predominantly inhabited by male and female adults and larvae (but no pupae) of one unidentified tubuliferan thrips species, Neoheegeria sp. (Phlaeothripinae), with 84.8% in staminate and 83.7% in pistillate inflorescences. The minor proportion of insects collected consisted of larvae and adults of terebrantian thrips, mainly of the genus Thrips (6.8% in staminate inflorescences [SI] and 4.8% in pistillate inflorescences [PI]), caterpillars of the geometrid genus Gymnoscelis (SI = 1.9%, PI = 5.2%), curculionid beetles and larvae (SI = 5.5%, PI = 2.3%), and a few heteropteran bugs and larvae of tiny dipterans (SI = 1.1%, PI = 4.1%).

Table 1. Summary of flower visitors of 27 staminate and 14 pistillate Macaranga hullettiitrees. For thrips (Neoheegeria and Terebrantia) and weevils (Curculionidae), both number of adults and larvae are given. ForNeoheegeria, the lumped percentage of adults and larvae per tree sample is added. "Others" include heteropteran bugs and dipteran larvae

The conspicuous overall dominance of Neoheegeria thrips in inflorescences of all Macaranga hullettii sampled is maintained if we look at it on the tree level (Table 1). All samples (taken from 27 staminate and 14 pistillate trees) contained adults of Neoheegeria sp., and the ratio of insect visitors was strongly Neoheegeria biased in all but six samples (of three staminate and three pistillate trees). Here, the comparatively low proportion of the tubuliferan thrips (17.9–50% of total insect number) may result from low numbers of inflorescences per tree collected. For each tree with >250 "pollination units" sampled (i.e., at least 3 inflorescences per staminate and 24 inflorescences per pistillate tree), the proportion of Neoheegeria thrips ranged in male trees from 57.9 to 96.6% (84.3 ± 11.98%, mean ± 1SD; N = 6 trees) and in female trees from 81.3 to 100% (90.2 ± 7.19%; N = 4 trees).

The abundance of adult Neoheegeria thrips per "pollination unit" was not significantly different among sex of trees (Mann-Whitney U test, Z = 0.97), year of collection (Z = 0.85), locality of collection (Kruskal-Wallis H test, H = 0.64), and stages of staminate inflorescences (Z = 1.05). The occurrence of larvae of Neoheegeria was significantly more frequent in mature staminate inflorescences than in inflorescences before anthesis (Z = 2.89, P = 0.002). The abundance of adult Neoheegeria thrips counted hourly over 24 h was not significantly different among times of collection (H = 29.06). During observations at night, the same insect visitors were seen as during daytime observations. If the hourly collected data were pooled into time-based classes (morning = 0700–1300, afternoon = 1400–1800, evening = 1900–2400, and night = 0100–0600), abundance of Neoheegeria sp. was also not significantly different among classes (H = 4.53).

The mean number of adults and larvae of Neoheegeria sp. per "pollination unit" ranged in pistillate trees from 0.08 to 0.38 (0.16 ± 0.09, mean ± 1SD, N = 14 trees) and in staminate trees from 0.02 to 2.77 (0.24 ± 0.52; N = 27 trees). The highest value (2.77) came from an asynchronously flowering staminate tree (N= 4 inflorescences with 1031 thrips sampled). By considering only the staminate trees that flowered synchronously, the number of Neoheegeria per "pollination unit" ranged from 0.02 to 0.80 (0.14 ± 0.16; N = 26 trees).

Generally, the total Neoheegeria thrips number per M. hullettii tree varies greatly, depending on total number of "pollination units" per tree and number of thrips per "pollination unit." For example, in the staminate tree (height ∼9 m), where the highest thrips abundance was found, counts of the number of inflorescences (∼750) resulted in an extrapolated number of ∼200 000 thrips for one single tree. In a female tree, ∼10 m in height, the 2090 inflorescences counted corresponded to an estimated number of ∼8000 thrips.

Behavior of floral visitors

Thrips appeared to strongly prefer the chambers formed by the bracts covering the flowers. Only in cases of thrips movement, e.g., from one chamber to another, did we observe thrips outside of these bract chambers (see below). The chambers provide an enclosed breeding site for thrips, and eggs and larvae were commonly found within these plant structures. Adults and larvae of the Neoheegeria thrips were observed feeding by piercing/sucking within a zone covered with trichomal nectaries on the adaxial base of the bracteoles (Fig. 4). Other feeding sites were never observed. The thrips fed with upwards and downwards swings of the head, and only some inflorescences showed some feeding damage caused by thrips at the end of anthesis. Then, the epidermis within the hairy tufts turned brown because of necrosis of the lower surface cells and some trichomes were found to be collapsed. Other visible signs of feeding damage to plant tissue caused by thrips was not observed. The Neoheegeria thrips showed the same feeding behavior in the laboratory as well. Here, we released thrips caught from inflorescences into an arena in which freshly harvested bracteoles were placed in distances of a few centimeters to each other. Immediately the thrips spread out, directly heading for the bracteoles. On reaching the bracteoles, they exclusively went for the trichomal adaxial surface.

In the field, thrips adults and larvae have been observed moving from one bract chamber to another. If such a chamber was already occupied by an inhabitant, the thrips got engaged in a territorial fight by flicking their abdomina at each other until one opponent was driven off. A few times we were able to observe Neoheegeria thrips taking off and settling on staminate inflorescences: settling of thrips is preceded by hovering for a few seconds in front of an inflorescence (∼15–20 cm distance), followed by a rapid approach at the landing site. After folding their wings, thrips quickly crawl into bract chambers. For a takeoff, the thrips usually ascend to an elevated point such as a tip of an inflorescence and then they fly in zigzags towards the light. Local flight movements of thrips between inflorescences (or neighboring trees) seemed to occur more frequently on sunny days.

Other insects from inflorescences were predominantly larvae of different insect orders (Table 1), which ate the trichomes of bracteoles and also destroyed parts of the flowers. Rarely, curculionid beetles were observed laying their eggs inside the flowers after they had hollowed out the flowers with their elongated mouthparts. Additional floral visitors, not collected by inflorescence sampling, were Trigona bees (Meliponinae) that have been observed at staminate inflorescences only. The bees collected pollen of inflorescences of an advanced flowering stage, i.e., some floral bracteoles had already dropped off, and some leftover pollen was deposited openly on the flower clusters, hence in reach for the bees. However, we have never observed bees visiting pistillate flowers.

Breeding of thrips

Females of Neoheegeria sp. lay their eggs always inside the bracteole chambers. We collected freshly laid eggs from inflorescences and reared them in glass tubes. In all cases the larvae hatched after 1.5–3 d. The two larval stages lasted ∼10 d (first-instar ∼1–2 d and second-instar ∼8 d). The development of the following, nonfeeding instars took place within ∼5 d (prepupae ∼1–2 d and pupae ∼3 d). Thus the life cycle is completed after 16–18 d. Moreover, our data show that M. hullettii bracteoles provide sufficient food to allow thrips to reach maturity.

Pollen transfer

Thirteen percent (12 of N = 91) of the thrips collected from pistillate inflorescences of M. hullettii carried pollen (3–71 grains/thrips). All pollen grains from thrips were identified as intraspecific. Although it was not possible to distinguish pollen of M. hullettii from one co-occurring congener,M. triloba, this Macaranga species could be excluded as a pollen donator, because no individual of M. triloba was found flowering at the time when the thrips were caught from M. hullettii.

Pollen load of adults of Neoheegeria sp. caught out of staminate inflorescences ranged from 0 to 268 grains/thrips (63.6 ± 70.3; 23 of 25 thrips carried pollen) and occurred equally on all body parts. Because of the small size of the pollen grains it is likely that the thrips (measuring up to 2 mm in length) can carry even more than 268 grains. The thrips have only a sparse hairiness, consisting of a few solitary setae and bristles. The main body surface is hairless, and pollen is sticking to the naked thrips-cuticle by pollenkitt (Fig. 5).

Seed set under natural and experimental conditions

The fruits of M. hullettii became mature 85–90 d after floral receptivity. Seed set ranged from 0 to 72% (46.7 ± 22.53%; N = 13 trees). However, in trees with low seed set (0–44%; N = 6) the Crematogaster ant-partners had apparently destroyed at least some of the inflorescences. This ant behavior, though very interesting, is not part of this study, thus we excluded those trees on which ant damage had occurred. Then mean seed set was 64.9 ± 6.2% (range = 53–72%; N = 7).

The bagging experiment showed highly significant differences in seed set both between experimental inflorescences bagged in fine nets and the inflorescences bagged in coarse nets (G test, G = 266.49; P < 0.00001), and between the fine net group and the inflorescences used as controls (G = 418.60; P < 0.00001). When access of insects to flowers was completely excluded, none of the flowers set seeds (Table 2). This result indicates that apomixis does not occur. In contrast, when only very small insects of the size of a thrips were given access (when coarse net was used), seed set reached nearly 80% of the controls (42.2 vs. 54.1%). Nevertheless, the seed set of the bagged flowers with coarse net was significantly lower than in the control group (G = 418.60; P < 0.01). Seed set of the controls (54.1%) was slightly lower than mean seed set per tree (64.9%) but was still within the range measured (see above). Spot checks during the experiment revealed thatNeoheegeria thrips indeed had entered the inflorescences bagged in coarse nets, and we never found any insect other than Neoheegeria thrips or theCrematogaster partner ants.

Table 2. Seed set of flowers in pollination experiments. Fine net and coarse net with mesh sizes of 0.2 mm and 1.0 mm, respectively. Significance of difference from seed set in open, untreated inflorescences (controls) was examined by Gtest (after Woolf)

DISCUSSION

Floral visitors

In accordance with observations of floral visitors of Macaranga spp. in Malaya (Taylor, 1982), our field observation showed that the most abundant insects visiting M. hullettii flowers of both sexes are thrips. We found that one thrips species, Neoheegeria sp., was the predominant floral visitor, while other thrips taxa, weevils, and a few other insects were only occasionally observed.

Single observations of floral visitors on M. brevipetiolata and M. winkleri from Sarawak led to the assumption that the main pollinator of these two species are Trigona bees (Momose et al., 1998). However, we observed Trigona bees exclusively as pollen collectors at staminate individuals and never as flower visitors at pistillate trees. Although we cannot rule out the possibility that the bees might play a role in the pollination of M. brevipetiolata and M. winkleri, we suggest that bee pollination, especially for species having enclosed inflorescences and no accessible reward for bees in pistillate inflorescences, is only accidental.

Thrips as effective pollen vectors

Generally, thrips are considered to be ineffective pollinators, because they are thought to be among the weakest flying insects and little is known about the interplant flights in flower thrips (Lewis, 1973). Although interplant movements have not been observed in this study directly, we demonstrated that thrips had moved from staminate to pistillate M. hullettii trees (over a distance of at least 25 m) by detecting intraspecific pollen loads onNeoheegeria thrips collected from pistillate inflorescences. In addition, our bagging experiments strongly suggest that Neoheegeria thrips indeed act as pollinators. When pollinators were excluded by net bags, no apomictic seed set occurred. Thus, reproduction of the dioecious M. hullettii trees necessarily requires intertree pollen transfer. Seed set of inflorescences bagged in nets that only allowed tiny thrips access to the pistillate flowers reached nearly 80% the seed set observed in open-pollinated flowers. We suggest that the predominant flower visitors, Neoheegeria sp., are not only effective pollen vectors but also the main pollinators of M. hullettii.

Although thrips are weak flyers, their fringed wings enable thrips to remain airborne long enough to be dispersed widely with the wind (Lewis, 1997). For a takeoff, the Neoheegeria thrips were observed to fly upwards towards the light, thus creating the opportunity to push themselves into oncoming wind, which would give them the extra lift they need to travel over great distances (Lewis, 1997). It is widely assumed that the tiny thrips, once borne aloft the winds, have minimal control over their flight path and destination. Despite this rationale, our observations document that, at least near the host plant, thrips can settle with a degree of control sufficient to choose even among individual inflorescences.

Although the potential of active orientation in Neoheegeria thrips cannot be assessed accurately, we expect that several factors, e.g., the distance to neighboring M. hullettii trees, the depth of the vegetation, and wind speed, will cause a certain amount of pollen wastage. Intraspecific movements between staminate individuals may also contribute a waste of pollen. Pollen quantity per flower in M. hullettii is relatively low (mean = 2209), but within a population, the higher flower number in staminate trees results in a P/O of ∼220 000. Thus, the high P/O ratio may compensate pollen wastage. In addition, the pollination units are not single staminate flowers but compact, many-flowered clusters, subtended by a bracteole. Within each flower cluster, one staminate flower after the other dehisces, which limits the portion of available pollen per floral visitation. This concentration of flowers forming separate, higher level units, combined with the release of pollen in small portions, probably minimizes the wastage by the displacement of pollen-carrying thrips to sites other than M. hullettii.

Macaranga as breeding site for thrips

The Neoheegeria thrips use both staminate and pistillate inflorescences of M. hullettii as breeding sites. Thrips that develop exclusively in flowers, or flower thrips, are known from many plant taxa, where they often feed on pollen, nectar, and/or plant sap (Kirk, 1997). Some of these flower thrips have been suggested to play at least some additional role in pollination of several angiosperm taxa (Annand, 1926; Hagerup, 1950; Syed, 1979; Rust, 1980; Thien, 1980; Appanah and Chan, 1981; Norton, 1984; Pellmyr et al., 1990; Webber and Gottsberger, 1995; Saxena et al., 1996; Kirk, 1997; Momose, Nagamitsu, and Inoue, 1998; Gottsberger, 1999; Luo and Li, 1999; Sakai et al., 1999; Williams, Adam, and Mound, 2001). Our data on the life history of the Macaranga-pollinating thrips shows similarities to the life histories of other flower thrips (Lewis, 1973). When adult Neoheegeriafemales arrive at inflorescences, they feed, mate, and lay eggs on the floral tissue inside the bracteole chambers. The larvae hatch and feed at the oviposition site until the second instar is completed. We assume thatNeoheegeria pupates in the soil under the tree (as this has been recorded for other flower thrips [Lewis, 1973]) because we never found pupate stages inside the inflorescences (the nonfeeding pupate stages are known to us only from thrips reared in test tubes). The adults probably return to the flowers and start the cycle again.

Even though the time of anthesis of both staminate and pistillate M. hullettiiindividuals was highly synchronized, anthesis in staminate inflorescences is preceded by an "open bud stage" (see above), which lasts 11–16 d, and allows thrips to enter the bracteole chambers prior to the onset of anthesis. Due to the relatively short time of thrips development of ∼17 d, thrips are able to complete one generation on staminate trees until flowering time of both sexes starts. In pistillate inflorescences, thrips have only the opportunity to develop during anthesis, i.e., during an ∼16-d period from the first bract opening until the floral bracteoles drop. It appears that the preceding "open bud stage" of male flowering trees enhances the potential for rapid multiplication of the thrips population and permits a quick response to the sudden and massive increase of breeding sites at the beginning of the flowering period.

The increase in the population density of thrips on staminate individuals presumably promotes the dispersal of thrips with pollen loads, leading to an increase of potential pollen vectors. In addition, we observed territorial behavior of Neoheegeria in the bracteole chambers, a behavior known from other flower thrips as well (Appanah and Chan, 1981; Terry, 1997). This behavior may also facilitate thrips movement both within and among trees. Efficiency of pollination service provided by thrips for M. hullettii may be influenced by another factor, i.e., the ratio of breeding sites in male and female trees. Generally, the mean number of pollination units (breeding sites) in male trees is about nine times higher than in female trees. This male-biased ratio is further increased by the higher number of flowering staminate trees relative to flowering pistillate trees in the second and third flowering period (Fig. 1). Both the higher number of breeding sites and the higher flowering frequency of male trees favor the mass production of potential pollen vectors.

Host specificity of Neoheegeria thrips

Breeding opportunities for pollinators are often associated with specialized pollinators (Faegri and van der Pijl, 1979). Some pollination systems, in which plants breed their pollinators, show highly specialized coevolutionary patterns like the fig/fig wasp, Yucca/yucca moth, and senita/senita moth pollination systems (e.g., Addicott, Bronstein, and Kjellberg, 1990; Fleming and Holland, 1998) or the pollination of Eupomatia (Eupomatiaceae) by weevils (Armstrong and Irvine, 1990) and Siparuna (Monimiaceae) by gallmidges (Feil, 1992). However, in most reported cases of thrips pollination, the thrips are generalists and were found also in other host plants (Appanah and Chan, 1981; Momose, Nagamitsu, and Inoue, 1998; Luo and Li, 1999; Williams, Adam, and Mound, 2001). A host-specific thrips species has recently been reported to be the sole pollinator of the cycad Macrozamia macdonnellii (Zamiaceae) (Mound and Terry, 2001). It is still unknown whether Neoheegeria sp. is host specific or not. Although this thrips species appears to be particularly attracted to inflorescences of several Macarangaspecies (U. Moog, unpublished data) nothing is known about other potential host requirements. Due to insufficient taxonomic clarity of the genusNeoheegeria, we cannot rule out the possibility that this thrips species has already been collected and described within another genus. As additional data on the ecology and distribution of Neoheegeria sp. are lacking, it remains unclear to what degree this thrips species is specialized for Macaranga.

Nutritive rewards and attractants for pollinators

No general patterns of host selection can be assigned to thrips, but they are known to use floral fragrances and other plant odors for host location, even in the absence of color (Kirk, 1985; Terry, 1997). To Neoheegeria thrips, olfactory cues seem to be a more important attractant than visual traits.

Apart from offering a protective breeding site as a reward, the inflorescences of M. hullettii provide nectar for alimentation. In previously reported cases of thrips pollinated plants, pollen and/or floral tissue (but no nectar) have been mentioned as a reward (Thien, 1980; Appanah and Chan, 1981; Webber and Gottsberger, 1995; Momose, Nagamitsu, and Inoue, 1998; Luo and Li, 1999; Mound and Terry, 2001; Williams, Adam, and Mound, 2001). In both male and female M. hullettii, thrips larvae and adults were feeding on the nectar within the hairy patches on the adaxial surface of the floral bracteoles. The tufts of trichomes resemble trichome nectaries that are described from only a few angiosperm taxa in Malvales, Dipsacales, Asclepiadaceae, Scrophulariaceae, Verbenaceae, and Cucurbitaceae (Endress, 1994; see references therein).

Trichomal patches inside the "bracteole chambers" seem to be a trait common to many other Macaranga species (U. Moog, personal observation). Those species show an enclosed inflorescence morphology similar to M. hullettii; the predominant flower visitors were thrips and for some species minute heteropteran bugs, which also fed within these trichomal nectaries (U. Moog, unpublished data). We suggest that in Macaranga, this kind of trichomal nectary might be a specialized reward for small insects like thrips and heteropteran bugs, because their elongated mouthparts enabled them to reach the small amounts of nectar within the dense tufts of trichomes.

Specific traits of thrips pollination

From the plant's point of view, the enclosed inflorescence morphology not only serves in the protection of reproductive floral parts, but the tiny openings also form an effective barrier against illegitimate visitors. Attraction of specific groups of pollinators has been found to be mainly selected for avoiding pollen wastage or stigma clogging (Faegri and van der Pijl, 1979). For tropical dioecious species, it has been suggested that they tend to have relatively small, generalized flowers that provide access to a wide variety of small insects (Bawa, 1980). However, the enclosed inflorescence morphology of several Macaranga species are particularly apt to a very restricted set of flower visitors, and we assume that several Macaranga species have a similar specialized mode of thrips pollination like M. hullettii.

A plant with an enclosed compact floral morphology that provides shelter together with narrow entrances and that has small to medium-sized pollen grains has been proposed to be traits indicative of a "thrips pollination syndrome" (Kirk, 1988; Williams and Adam, 1994). Additionally, scent is likely a primary attractant, and white to yellow floral colors are especially attractive to thrips in conjunction with scent (Kirk, 1985). Rewards in the form of pollen, nectar, and/or nutritive tissues may be present or absent. However, the definition of a "thrips pollination syndrome" remains less than satisfactory, as most of the traits are not distinctive and overlap with traits of other syndromes, including flowers pollinated by small beetles (Endress, 1994).

In addition to traits probably adaptive to various small insect pollinators (i.e., protective compact floral morphology with narrow entrances and small to medium-sized pollen grains), pollen morphology of M. hullettii shows a feature that may have evolved in adaptation to thrips pollination. The tectum of M. hullettii pollen is smooth and unsculptured (Fig. 3), which is unusual among zoophilous pollen (Faegri and van der Pijl, 1979). The smooth pollen grains adhere to the naked thrips cuticle only by pollenkitt, without the help of a sculptured tectum. In section Pachystemon sensu stricto, the same smooth pollen was also found in other Macaranga species that are predominantly visited by Neoheegeria thrips (U. Moog, unpublished data), whereas in some other Macaranga species, which have flowers that are commonly visited by small insect taxa other than thrips, pollen was found to be more sculptured. According to DNA sequence data of the nuclear ribosomal DNA internal transcribed spacer region (ITS), section Pachystemon constitutes a well-defined group in a derived position within Macaranga (Blattner et al., 2001). However, we cannot yet exclude the possibility that the observed correlation in section Pachystemon of (1) an unsculptured tectum and thrips pollination and (2) a more sculptured pollen and pollination by various small insects is purely accidental. Further research is needed to identify any underlying selective forces that may explain this pattern.

Flowering phenology

The phenological patterns found in Macaranga hullettii corresponded with patterns reported for this species in being episodic (Taylor, 1982; Mitchell, 1994; Davies and Ashton, 1999). By using a 5-wk census interval, Davies and Ashton (1999) documented a flowering period of several months for M. hullettii that was particularly concurrent among the co-occurring Macarangaspecies. The weekly census interval used in this study and the more exact classification of the flowering time allowed us to subdivide the flowering of M. hullettii into highly synchronized flowering periods lasting 3–4 wk only. Thus, the phenological patterns for M. hullettii can be categorized as a "multiple bang species" (Gentry, 1974), i.e., a species that produces many flowers suddenly and for short periods. For such mass-flowering plants, short-lived thrips may be particularly apt pollinators because they have the potential to build up large populations within a short time (Appanah and Chan, 1981).

In contrast to flowering patterns found in Macaranga spp. (including M. hullettii) in Sarawak (Davies and Ashton, 1999), none of the M. hullettii trees studied failed to reproduce during a 5-mo period, but staminate individuals flowered significantly more frequently than pistillate individuals. In terms of resource allocation, staminate flowers are relatively less "expensive" to produce (Cipollini and Whigham, 1994) and consequently staminate trees may not only bear a higher number of flowers (Allen and Antos, 1988; Bawa, 1980), but also were found to flower more frequently than pistillate individuals (Armstrong and Irvine, 1989). Both patterns appear to hold for M. hullettii as well. It is suggested that reproduction in Macaranga is resource limited, as species have greater reproductive frequency and flower density in higher light availability (Davies and Ashton, 1999).

Generally, congeners that occur in the same community tend to have staggered flowering if they are pollinated by the same vectors (Stiles, 1975; Chan and Appanah, 1980; Neill, 1987; Ashton, Givnish, and Appanah, 1988; LaFrankie and Chan, 1991). Macaranga hullettii is found to be predominantly pollinated by one thrips species. The same thrips species was found in a similar abundance in inflorescences of co-occurring Macarangaspp. from West Malaysia and Borneo (U. Moog, unpublished data). Field observations suggested that these episodic Macaranga species flower mainly within the nonflowering sequences of M. hullettii. In contrast, thrips taxa different from Neoheegeria and mirid bugs (Heteroptera) were the predominant flower visitors of coflowering species (U. Moog, unpublished data). Although these preliminary data need to be supported in future studies, the reproductive isolating mechanisms in Macaranga appear to be divers, as (1) temporal patterns are likely to play a role for some species that share pollinators, but (2) in species with phenological synchrony, reproductive isolation may occur by using different pollen vectors. Detailed studies are required on the community level to assess the influence of both phenological patterns and pollinator requirement on Macaranga reproduction.

Conclusions

Analyses of pollination biology indicate that Macaranga hullettii is predominantly pollinated by one thrips species, Neoheegeria sp. The thrips use the enclosed chambers formed by floral bracteoles of staminate and pistillate inflorescences both as feeding and breeding sites. In staminate individuals of M. hullettii, the floral bracteoles are colonized by thrips ∼11–16 d prior to the onset of anthesis ("open bud stage"), thus allowing thrips to complete one generation until flowering time of both sexes starts. We assume that this preceding "open bud stage" of staminate flowering trees rapidly increases the population density of the thrips and permits a quick response to the massive increase of breeding sites at the beginning of the flowering period. However, host specificity of Neoheegeria thrips is not known. Although we collected this thrips species only from Macaranga, the insufficient taxonomic clarity of the genus Neoheegeria prevented us from assessing its host specificity by a survey of host records in the literature. We propose that thrips pollination of M. hullettii is not an isolated case within the genus Macaranga, at least in section Pachystemon sensu stricto, because we collected thrips from several other members of the genus that share the floral traits observed in M. hullettii: small, inconspicuous inflorescences with enclosed bracteole chambers and narrow entrances, small, smooth pollen grains, and trichomal "feeding zones" on the adaxial surface of the floral bracteoles. Macaranga, the world's largest genus of pioneer trees, may be an excellent model to explore the phenomenon of thrips pollination in greater detail.

Footnotes

  • 1 The authors thank Dr. Azarae Hj Idris and Dr. Rosli Hashim (University Malaya, Kuala Lumpur) for logistical support; Dr. R. zur Strassen (Senckenberg Museum Frankfurt) for his help in identifying thrips specimens; M. Ruppel for assistance in the preparation of SEM micrographs; and J. Moog and Dr. F. R. Blattner for comments on the manuscript. We are grateful for financial support from the German Research Council (DFG). U. Moog was also supported by grants from the German Academic Exchange Service (DAAD).

  • 5 Author for reprint requests (Ute.Moog@zoology.uni-frankfurt.de ).

LITERATURE CITED

  1.  
    Addicott J. F. J. L. Bronstein F. Kjellberg 1990. Evolution of mutualistic life-cycles: yucca moths and fig wasps. In F. Gilbert [ed.], Insect life cycles, 143–163. Springer, New York, New York, USA.
  2.  
    Allen G. A. J. A. Antos 1988. Relative reproductive effort in males and females in the dioecious shrub Oemleria cerasiformisOecologia 76111-118.
     
  3.  
    Annand P. N. 1926. Thysanoptera and the pollination of flowers. American Naturalist 60177-182.
     
  4.  
    Appanah S. H. T. Chan 1981. Thrips: the pollinators of some dipterocarps.Malaysian Forester 44234-252.
  5.  
    Armstrong J. E. H. K. Irvine 1989. Flowering, sex ratios, pollen-ovule ratios, fruit set, and rain forest communities. American Journal of Botany 76:74-85.
     
  6.  
    Armstrong J. E. H. K. Irvine 1990. Functions in staminodia in the beetle-pollinated flowers of Eupomatia laurinaBiotropica 22429-431.
     
  7.  
    Ashton P. S. T. J. Givnish 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 13244-66.
  8.  
    Bawa K. S. 1980. Evolution of dioecy in flowering plants. Annual Review of Ecology and Systematics 1115-39.
     
  9.  
    Blattner F. R. K. Weising G. Bänfer U. Maschwitz B. Fiala 2001. Molecular analysis of phylogenetic relationships among myrmecophytic Macarangaspecies (Euphorbiaceae). Molecular Phylogenetics and Evolution 19331-344.
  10.  
    Chan H. T. S. Appanah 1980. Reproductive biology of some Malaysian dipterocarps. I. Flowering biology. Malaysian Forester 43132-143.
  11.  
    Cipollini M. L. D. F. Whigham 1994. Sexual dimorphism and costs of reproduction in the dioecious shrub Lindera benzoin (Lauraceae). American Journal of Botany 8165-75.
     
  12.  
    Davies S. J. 1999. A new myrmecophytic, thrips-pollinated species ofMacaranga from the highlands of Sarawak. Harvard Papers in Botany 4433-436.
  13.  
    Davies S. J. P. S. Ashton 1999. Phenology and fecundity in 11 sympatric pioneer species of Macaranga (Euphorbiaceae) in Borneo. American Journal of Botany 861786-1795.
     
  14.  
    Endress P. K. 1994. Diversity and evolutionary biology of tropical flowers. Cambridge University Press, New York, New York, USA.
  15.  
    Faegri K. L. van der Pijl 1979. The principles of pollination ecology, 3rd ed. Pergamon Press, Oxford, UK.
  16.  
    Feil J. P. 1992. Reproductive ecology of dioecious Siparuna (Monimiaceae) in Ecuador: a case of gall midge pollination. Botanical Journal of the Linnean Society 90129-144.
  17.  
    Fiala B. A. Jakob U. Maschwitz K. E. Linsenmair 1999. Diversity, evolutionary specialization and geographic distribution of a mutualistic ant–plant complex: Macaranga and Crematogaster in South East Asia. Biological Journal of the Linnean Society 66305-331.
     
  18.  
    Fleming T. H. J. N. Holland 1998. The evolution of obligate pollination mutualisms: senita cactus and senita moth. Oecologia 114368-375.
  19.  
    Gentry A. H. 1974. Flowering phenology and diversity in tropical Bignoniaceae. Biotropica 664-68.
     
  20.  
    Gottsberger G. 1999. Pollination and evolution in Neotropical Annonaceae.Plant Species Biology 14143-152.
     
  21.  
    Hagerup O. 1950. Thrips pollination in CallunaBiologiske Meddelelser 18:1-16.
  22.  
    Kearns C. A. D. W. Inouye 1993. Techniques for pollination biologists. University Press of Colorado, Niwot, Colorado, USA.
  23.  
    Kirk W. D. J. 1985. Effects of some floral scents on host finding by thrips (Insecta: Thysanoptera). Journal of Chemical Ecology 1135-43.
     
  24.  
    Kirk W. D. J. 1988. Thrips and pollination biology. In T. N. Ananthankrishnan and A. Raman [eds.], Dynamics of insect–plant interactions, 129–135. Oxford and IBH, New Delhi, India.
  25.  
    Kirk W. D. J. 1997. Feeding. In T. Lewis [ed.], Thrips as crop pests, 65–118. CAB International, Oxon, UK.
  26.  
    La Frankie J. V., Jr. H. T. Chan 1991. Confirmation of sequential flowering in Shorea (Dipterocarpaceae). Biotropica 23200-203.
     
  27.  
    Lewis T. 1973. Thrips: their biology, ecology and economic importance. Academic Press, London, UK.
  28.  
    Lewis T. 1997. Flight and dispersal. In T. Lewis [ed.], Thrips as crop pests, 175–196. CAB International, Oxon, UK.
  29.  
    Luo Y.-B. Z.-Y. Li 1999. Pollination ecology of Chloranthus serratus(Thunb.) Roem. et Schult. and Ch. fortunei (A. Gray) Solms-Laub. (Chloranthaceae). Annals of Botany 83489-499.
     
  30.  
    Mitchell T. C. 1994. The ecology of Macaranga (Euphorbiaceae) trees in primary lowland mixed dipterocarp forest, Brunei. Ph.D. dissertation, Cambridge University, Cambridge, UK.
  31.  
    Momose K. T. Nagamitsu T. Inoue 1998. Thrips cross-pollination ofPopowia pisocarpa (Annonaceae) in a lowland dipterocarp forest. Biotropica30444-448.
     
  32.  
    Momose K. T. Yumoto T. Nagamitsu M. Kato H. Nagamasu S. Sakai R. D. Harrison T. Itioka A. A. Hamid 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 851477-1501.
     
  33.  
    Mound L. A. I. Terry 2001. Thrips pollination of the Central Australien cycad, Macrozamia macdonnellii (Cycadales). International Journal of Plant Sciences 162147-154.
     
  34.  
    Neill D. A. 1987. Trapliners in the trees: hummingbird pollination ofErythrina sect. Erythrina (Leguminosae: Papilionoideae). Annals of the Missouri Botanical Garden 7427-41.
     
  35.  
    Norton S. A. 1984. Thrips pollination in the lowland forest of New Zealand.New Zealand Journal of Ecology 7157-164.
  36.  
    Pellmyr O. L. B. Thien G. Bergström I. Groth 1990. Pollination of New Caledonian Winteraceae: opportunistic shifts or parallel radiation with their pollinators?. Plant Systematics and Evolution 173143-157.
     
  37.  
    Rust R. W. 1980. Pollen movement and reproduction in Arisaema triphyllumBulletin of the Torrey Botanical Club 107539-542.
     
  38.  
    Sakai S. K. Momose T. Yumoto M. Kato T. Inoue 1999. Beetle pollination ofShorea parvifolia (Section Mutica, Dipterocarpaceae) in a general flowering period in Sarawak, Malaysia. American Journal of Botany 8662-69.
  39.  
    Saxena P. M. R. Vijayaraghavan R. K. Sarbohoy U. Raizada 1996. Pollination and gene flow in chilies with Scirothrips dorsalis as pollen vectors.Phytomorphology 46317-327.
  40.  
    Stiles F. G. 1975. Ecology, flowering phenology, and hummingbird pollination of some Costa Rican Heliconia species. Ecology 56285-301.
  41.  
    Syed R. A. 1979. Studies on oil palm pollination by insects. Bulletin of Entomological Research 69213-224.
     
  42.  
    Taylor C. E. 1982. Reproductive biology and ecology of some tropical pioneer trees. Ph.D. dissertation, University of Aberdeen, Aberdeen, UK.
  43.  
    Terry L. I. 1997. Host selection, communication and reproductive behaviour. In T. Lewis [ed.], Thrips as crop pests, 65–118. CAB International, Oxon, UK.
  44.  
    Thien L. B. 1980. Patterns of pollination in the primitive angiosperms.Biotropica 121-13.
  45.  
    Webber A. C. G. Gottsberger 1995. Floral biology and pollination ofBocageopsis multiflora and Oxandra euneura in Central Amazonia with remarks on the evolution of stamens in Annonaceae. Feddes Repertorium106515-524.
  46.  
    Webster G. L. 1994. Synopsis of the genera and suprageneric taxa of Euphorbiaceae. Annals of the Missouri Botanical Garden 8133-144.
  47.  
    Whitmore T. C. 1969. First thoughts on species evolution in MalayanMacaranga. (Studies in Macaranga III). Biological Journal of the Linnean Society London 1223-231.
  48.  
    Whitmore T. C. 1990. An introduction to tropical rain forests. Oxford University, Oxford, UK.
  49.  
    Whitmore T. C. C. P. Burnham 1984. Tropical forest of the Far East. Clarendon Press, Oxford, UK.
  50.  
    Williams G. P. Adam 1994. A review of rainforest pollination and plant–pollinator interactions with particular reference to Australian subtropical rainforests. Australian Zoologist 29177-212.
  51.  
    Williams G. P. Adam L. A. Mound 2001. Thrips (Thysanoptera) pollination in Australian subtropical rainforests, with particular reference to pollination ofWilkiea huegeliana (Monimiaceae). Journal of Natural History 351-21.

__._,_.___

Posted by: Green Convert Network <greenconvertnetwork@gmail.com>
Reply via web post Reply to sender Reply to group Start a New Topic Messages in this topic (1)
http://www.glinus.com/

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=-=-=-=-=-=-=-=-=-=-=-=-=-=-
.To join : glinus-subscribe@yahoogroups.com
.To post a message : glinus@yahoogroups.com
.To unsubscribe: glinus-unsubscribe@yahoogroups.com
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

.

__,_._,___

READ MORE RECENT NEWS AND OPINIONS

WASH news Africa

News from Friends of the Earth

IUCN - News

Institute of Development Studies News

Human Nature - Conservation International Blog

Energia: News

Traditional Knowledge Bulletin

Water Conserve: Water Conservation RSS News Feed

Water Supply and Sanitation News

FAO/Forestry/headlines

InforMEA

Sustainable Development Policy & Practice - Daily RSS Feed

IISD - Latest Additions

IISD Linkages

Climate Change Headlines

DESERTIFICATION

Human Nature - Conservation International Blog