Is Rafflesia an endothermic flower?


  • Sandra Patiño,

    Corresponding author
    1. Institute of Ecology and Resource Management, The University of Edinburgh, Darwin Building, Mayfield Road, Edinburgh EH9 3JU, UK;
    2. Helsinki Finland;
      Author for correspondence:Sandra Patiño Fax: + 57 1 288 9564 Email:
    Search for more papers by this author
  • Tuula Aalto,

    1. Department of Physics, University of Helsinki, PO Box 9 (Siltavuorenpenger 20D), University of Helsinki, FIN−0001
    Search for more papers by this author
  • Alice A. Edwards,

    1. Chemistry Department, Universiti Brunei Darussalam, Jln. Tungku Link, Bandar Seri Begawan BE 1410, Brunei Darussalam; Present address: Alexander von Humboldt Biological Research Institute, Calle 37 #8-40 Mezanine, Bogotá DC, Colombia
    Search for more papers by this author
  • John Grace

    1. Institute of Ecology and Resource Management, The University of Edinburgh, Darwin Building, Mayfield Road, Edinburgh EH9 3JU, UK;
    Search for more papers by this author

Author for correspondence:Sandra Patiño Fax: + 57 1 288 9564 Email:


  •  The giant flowers of the parasitic Rafflesia occur in the shade of the forest understorey. They present several characteristics in common with the related species, Rhizanthes lowii, which is a strongly endothermic flower. The possible existence of endothermy in Rafflesia tuan-mudae was investigated here.
  •  The internal and surface temperature of the flowers were continuously monitored with fine thermocouples while radiation fluxes and microclimatic variables were recorded. A computational fluid dynamic model was used to predict the concentrations of CO2 inside the diaphragm of the flower.
  •  It was found that the internal parts of the flower were maintained a few degrees (1–6 K) above air temperature. It was not possible to account for this temperature rise without postulating a significant internal source of heat. It was concluded that R. tuan-mudae is an endothermic flower that generates a maximum of 50–60 W m−2 of heat in the centre of the column.
  •  The possible role of endothermy, CO2 and volatiles as elements in the mimicry of the flower to attract pollinating blowflies is discussed and compared with the related species Rhizanthes lowii.


The genus Rafflesia belongs to the wholly parasitic family Rafflesiaceae, which comprises 16 known species. It is known to be the largest flower of the plant kingdom (13–107 cm diameter) (Meijer, 1984) and also one of the rarest, being threatened with extinction as a consequence of the destruction of its habitat (Meijer, 1985; Bänziger, 1991; Nais & Wilcock, 1998). The genus is distributed from north of the Kra isthmus of Thailand through western Malaysia and the Philippines, Borneo, Sumatra and Java (Bänziger, 1991; Salleh, 1991; Meijer, 1997). Rafflesia is an understory plant, which is first evident as a small protuberance emerging from the roots or near-ground stems of a few species of the vine Tetrastigma (Vitaceae) (Fig. 1a). After 6–12 months it takes the form of a pink-brownish ‘cabbage’ (Fig. 1b), which blooms into an ephemeral flower (Fig. 1c,d) but lacks leaves or photosynthetic tissue, stems or roots, the only vegetative parts being fine filaments that penetrate the tissue of the vine host (Meijer, 1985; Ismail, 1988; Nais & Wilcock, 1998). The flowers are unisexual (Beaman et al., 1988; Bänziger, 1991) and by looking and smelling like ‘rotten flesh’ or ‘festering sore’, they attract several species of carrion flies or blowflies of the genus Lucilia and Chrysomya (Calliphoridae), which pollinate them (Bänziger, 1991; Beaman et al., 1988; Bänziger, 1996). If pollinated, after 6–9 months the structure below the column that holds the ovary of the female flower becomes the fruit, holding many thousands of miniature seeds that are likely to be dispersed by small mammals such as squirrels and treeshrews (Meijer, 1985). How the seeds germinate and penetrate the host is still unclear.

Figure 1.

Sequence of blooming of Rafflesia tuan-mudae: (a) young buds, about 3 months and 4 months old; (b) mature bud, about 9 months old; (c) opening bud, after 9 months growth; (d) female flower almost in full bloom.

In a recent study (Patiño et al., 2000), it was reported that the related species Rhizanthes lowii is a strongly endothermic flower that exhibits weak thermoregulation. Endothermy in plants has been associated with an increase in cyanide-insensitive respiration, an alternative pathway of the respiratory electron transport system in the citric acid cycle that generates heat without producing adenosine triphosphate (ATP) (Bahr & Bonner, 1973; Meeuse, 1975; McCaig & Hill, 1977; Meeuse, 1978; McNulty & Cummins, 1987; Raskin et al., 1989). This process causes production and diffusion of copious amounts of CO2 (Buggeln et al., 1971) and volatile compounds (Meeuse, 1966; Meeuse, 1975; Meeuse, 1978) that may act as attractants of beetle pollinators. Patiño et al. (2000) suggested that CO2, when combined with other volatile compounds, increases the probability of pollination of R. lowii.

Although there are morphological and functional differences, Rafflesia species and R. lowii share many characteristics. These flowers are sympatric in many areas and occur in the dense shade of the forest understory. Both are totally parasitic on a few, but different species of Tetrastigma ssp. lianas. They both have a very specific pollination process, being pollinated only by carrion flies or blowflies, which they attract by resembling rotting flesh and/or the wounds of animals and by producing volatile substances (a form of mimicry). The aim of this study was to address the hypothesis that endothermy is part of the mimicry of Rafflesia to attract pollinating flies, as in its relative R. lowii. To test the hypothesis we studied flowers of Rafflesia tuan-mudae in natural conditions.

Materials and Methods

Plant material

Rafflesia tuan-mudae Becc. is an endemic species from Sarawak (Malaysia) confined to three isolated mountains Pueh, Gading and Rara. It usually grows in rich alluvial or limestone-derived soils and is restricted to primary and secondary forest of altitudes below approx. 2000 m. Meijer (1997) states the flower has a diameter range of between 44 cm and 56 cm, although Beccari (1868) reports an individual of 86 cm in diameter and one female flower of the present study measured 78 cm diameter. Three young buds and five flowers were located in the understory of dense canopy and studied in natural conditions in two visits in 1998 and one in 1999. One male flower and three buds were investigated from 14 to 27 July, three female flowers from 27 September to 12 October 1998 and one male flower from 13 to 17 January 1999. Four flower stages were classified as: (1) young bud, 3 months to 1 month before anthesis; (2) anthesis; (3) fresh flower, the first and the second days following anthesis; and (4) decaying flower, from the third day following anthesis. Mature buds were not used in this study to avoid any damage that could affect the final develop of the buds and the further anthesis.

Study site

This study was conducted in Taman Negara Gunung Gading, Lundu, Sarawak, Malaysia, SE Asia (1°40′ N 109°52′  E). The forest here is classified as a lowland Dipterocarp forest, 50–883 m above sea level (Meijer, 1997). The climate is aseasonal, with mean monthly rainfall exceeding 100 mm for all months.

Field measurements

The procedures described below were applied to all the flowers in this study unless otherwise specified.

Temperature of buds, flowers and the microclimate

Thermocouples (0.5 mm in diameter PVC-insulated copper-constantan; Industrial Thermocouple Supplies Pty. Ltd., Thomastown, Victoria Australia) were used to measure the temperatures of buds and floral structures. To place the thermocouples in the interior (column) of the buds (Fig. 2a), they were first threaded in 7–20 cm hypodermic needles, which were fully inserted into the centre of the bud, and then carefully removed by slipping the wire through the needle, leaving the end of the thermocouple at the desired place. The temperature measurements of open flowers started on the first day of blooming. Thermocouples were placed in different parts of the flower (Fig. 2b) to measure the temperature at: the centre of the disk or column (c), anther area of male flowers (at), stigma area of female flowers (st), process (p), surface of the perigone lobe (sp), surface of the diaphragm (sd), window (w) and surface of the perigone tube (pt). An additional thermocouple was placed inside the diaphragm without touching the tissue to measure the temperature of the air inside the diaphragm (da) (Fig. 2b). Soil temperature (Tso) was also recorded with a thermocouple 5 mm below the soil surface. Air temperature (Ta), and humidity were measured with thermocouples in a custom-built forced-air radiation-shielded hygrometer located near the flower. Wind velocity, u, at the flower surface was measured with an omnidirectional hot-wire anemometer (model 8460-13E-V, TSI Inc., St Paul, MN, USA). This anemometer was used only during the day and disconnected when rain fell. A cup anemometer (Model MG2, Vector Instruments, Rhyl, Denbighshire, UK) was used night and day. This anemometer was placed at flower level. The net radiation, Rn, was measured with a Funk type net radiometer (Q*7, Campbell Scientific Ltd., Loughborough, Leicestershire, UK). The net radiometer was placed near the flower at the same level. Photosynthetic photon flux density (PPFD) was recorded with a quantum light sensor (Quantum Sensor SKP215, Sky Instruments Ltd. Llandrindod Wells, Powys, Wales, UK) to determine when the flowers were illuminated by brief periods of penumbral sunlight. The sensor was placed at the same level as the net radiometer.

Figure 2.

A schematic representation of a median longitudinal section through (a) male bud and (b) female flower of Rafflesia tuan-mudae showing the floral structures. The symbols are: sp, surface of the perigone lobe; w, windows; p, process; at, anther area of the male flowers; st, stigmatic area of the female flower; c, column; o, ovary; sd, surface of the diaphragm; da, space of the diaphragm; ra, ramenta; pt, surface of the perigone tube.

All variables for the two male flowers and the three buds were measured every 20 s and recorded as means of 15 readings once per 5 min night and day. For the three female flowers, the variables were measured every 6 s and recorded as means of 10 readings once per minute during the day, and every 20 s, and recorded as means of 15 readings once per 5 min, during the night on a data logger (Model 21X, Campbell Scientific, Ltd). They were subsequently plotted as average 10-min values.

Data analysis Differences in mean excess temperatures of the column of three female vs two male flowers and of the anther vs stigmatic area of the same flowers were analysed with one-way analysis of variance using Minitab 12.3 software (Minitab Inc., State College, PA, USA).

Heat balance models

The heat supply required to raise the tissue temperature of the central column of a female flower several degrees above ambient temperature was calculated using two different approaches. (1) The method described by van Gardingen & Grace (1991); in this method, the heat transfer is modelled as a combination of forced and free convection, assuming no transpirational cooling. (2) An exact method using computational fluid dynamics; heat and CO2 transfer in this method is solved using a commercial software package fluent (Kim et al., 1997). The solution was based on a finite difference scheme in an adaptive grid and the transfer equations were solved in a segregated mode. Flow past the flower was set to the measured value of 0.2 m s −1, and was assumed to be turbulent. Free convection caused by temperature differences was also included, since the wind velocities near the central column were low. Physical properties of the flower material (conductivity, specific heat capacity) were assumed to be the same as water.

The CO2 production rate was solved from heat production given knowledge of the heat of combustion of glucose, and assuming that six CO2 molecules are produced per one glucose oxidized (Penning de Vries et al., 1974).


Temperature of buds, flowers and microclimate

Diurnal changes in tissue temperature and microclimate variables were studied for 14 d on three young buds of undetermined sex and 5–8 d on two male and three female flowers. The buds available for study did not reach maturity because termites destroyed them at the end of the second week of measurements.

Figure 3 shows a representative 9-d course of tissue and air temperatures for the three buds. The data represent the average of interior (column) and surface tissue temperatures for the three buds.

Figure 3.

Daily pattern of air and tissue temperature for three young buds of Rafflesia tuan-mudae from 14 to 24 October 1998. Nights are indicated by grey shadows. (a) Temperature of the air (continuous bold line), interior (continuous thin line) and surface (dotted line). (b) Excess temperature for the interior (continuous line) and surface (dotted line).

Air temperature fluctuated from 22.1 to 27.5°C while the internal temperature fluctuated between 23.4°C and 26.0°C. Internal temperature lagged behind air temperature, peaking in the early evening (Fig. 3a). The surface temperature closely followed the pattern of air temperature. During the warmer days (days 2, 4 and 9) the interior and surface temperatures were lower than the air with a minimum difference of −2.8 K, suggesting evaporative cooling. During days 6–9 there is evidence of metabolic heating, with excess temperatures (0–2 K) occurring most of the time (Fig. 3b). Here, the term ‘excess temperature’ means the extent to which the tissue is warmer than the air and is expressed in K. In this period, the interior temperatures usually exceed the surface temperatures (Fig. 3b). The maximum excess temperatures were remarkably consistent between buds, usually varying by ± 0.04 K. Net radiation was usually less than 15 W m−2 and PPFD was usually less than 10 µmol m−2 s−1, with a maximum of 27 µmol m−2 s−1 on day 2 as a result of a sunfleck (data not shown).

∖Figure 4a–d represents a 5-d course of temperature and microclimate for a blooming female flower in October 1998. The diameter of this flower was 49.5 cm. The patterns presented in this figure were similar for the five blooming flowers of this study regardless of the sex, but only one is shown for simplicity of presentation (some differences between individual flowers will be noted below).

Figure 4.

Temperature and microclimate of a female flower of Rafflesia tuan-mudae during a 5-d period from 3 to 7 October 1998. The flower stages are indicated by horizontal arrow-lines; from the day of anthesis (2), fresh flower (3) and decaying flower (4). Nights are indicated by grey shadows. (a) Temperature of air and tissue: lines represent column (dotted line), stigmatic area (continuous thin line), and air (thick continuous line). (b) Excess temperature of the tissue: lines are stigmatic area (continuous line), column (dotted line) and the air inside the diaphragm (continuous line and open circle). (c) Vapour pressure deficit (VPD, continuous line) and wind velocity (u, dotted line). (d) Measured net radiation (Rn, continuous line) and photosynthetic photon flux density (PPFD) (dotted line).

The internal tissue was usually higher than air temperature. In this flower, the column was the warmer part with a maxima of about 34°C on the second day in the early afternoon (Fig. 4a). There was a trend in which the excess temperature increases in the early evening reaching a maximum of 4.2 K in the stigmatic area and 5.8 K in the column (Fig. 4b). There was no significant difference in excess temperature between the column of female and male flowers (F = 11, P < 0.001) nor between the anther and the stigmatic area (F = 26.5, P < 0.0001), suggesting that the visit of the pollinating flies to female and male flowers is performed in the same way. The excess temperature of the tissue was sometimes negative (−0.5 to −1.5 K) during the day (days 4 and 5), usually when there was rain. The air temperature inside the diaphragm of the flowers sometimes exceeded that of the external air, reaching a maximum excess of 1 K (Fig. 4b). There were some differences in the maximum excess between flowers. For example, one bigger female flower (78 cm in diameter) had an excess temperature of about 2.2 K in the column (data not shown). Wind velocity at the flower level never exceeded 0.6 m s−1 and vapour pressure deficit (VPD) was less than 0.6 kPa during the whole period, so evaporation is unlikely to have contributed substantially to the heat balance of the flower(s) except at midday (Fig. 4c). Net radiation never exceeded 30 W m−2. There were no sunflecks and the maximum PPFD was about 25 µmol m−2 s−1 (Fig. 4d).

Figure 5 shows a schematic representation of tissue temperature for a female flower during the day. The day is divided into four 6-h periods. Each temperature represents the mean of 2-d fresh flower measurements for each period. The measured net radiation (Rn) is also shown. The maximum temperatures were attained in the column and stigmatic area between 15 : 00 h and 21 : 00 h (Fig. 5c) with the column reaching 31.7°C, the stigmatic area reaching 29.7°C while the air temperature was only 26.9°C and the soil temperature was 26.1°C. The coldest period inside the flower was the early morning (from 03 : 00 h to 09 : 00 h, Fig. 5a) with the column reaching 27°C and the stigmatic area 27.3°C while the mean air temperature in this period was 24.6°C. For all the flowers the column showed the most warming with maximum values normally fluctuating between 33.6°C and 27.0°C (data not shown) while the air temperature fluctuated between 24.6°C and 29.1°C.

Figure 5.

Schematic representation of tissue temperature of a female flower during the day. The day is divided into four periods of 6 h each. Each temperature value represents the mean temperature of 2-d measurements for the fresh flower (stage 3). (a) Period from 03 : 00 h to 09 : 00 h; (b) from 09 : 00 h to 15 : 00 h; (c) from 15 : 00 h to 21 : 00 h; (d) from 21 : 00 h to 03 : 00 h. The numbers inside ovals are the mean values of temperature in °C. Net radiation (Rn) is given in W m−2.

Heat balance models

The possibility that the energy for heating the tissue by up to 6 K is not derived from metabolism but from radiation was explored (Fig. 6). According to both the models the heat supply necessary to produce an excess temperature of 3–6 K in a dry flower at an air flow of 2.4 × 10−4 m s−1 inside the diaphragm would be about 50–60 W m−2. Given some evaporative cooling, the heat supply would need to be more than this. As the net radiation was usually less than 15 W m−2, we conclude that metabolic heat must have been an important part of the heat supply.

Figure 6.

Calculation of the heat supply needed to raise the temperature of the flowers to the range observed for different wind speeds (u, m s−1) inside the diaphragm. The symbols indicate the calculated heat supply made by method one using the observed temperature of the flower surface in the understory of the forest. The calculated heat supply by method two using fluent software is represented by diamonds and u. The air flow inside the diaphragm was calculated with fluent. An air flow of 2.4 ×  10−4 m s−1 inside the diaphragm corresponds to a wind speed of 0.2 m s−2 in the understory and an air flow of 1.7 m s−2 corresponds to a wind speed of 1 m s−1 in the understory.

The concentration of CO2 obtained with fluent in the stigmatic area is about four to 17 times higher than near the surface of the diaphragm, 76 times higher than at the surface of the flower and approximately 76 times compared with the understory concentration (Fig. 7). For example, the CO2 concentration in the understory (1 m above the ground) in the Amazonian rain forest at the Reserva Jaru, Brazil varied from 360 ppm in the evening to 570 ppm early in the morning (Kruijt et al., 1996)

Figure 7.

Map of CO2 concentration inside the flower. The CO2 concentration is estimated by fluent software, on the basis of the observed rise in temperature. Respiratory activity was assumed to be confined to the central column where the cavity with the reproductive parts is located.


In this study we conclude that R. tuan-mudae is an endothermic flower, showing a maximum of 6 K of excess tissue temperature in one flower. Endothermy in Rafflesia has been suggested previously (Meeuse, 1978; Beaman et al., 1988) and efforts to detect heating were made (Nais, 1997). Nais’s studies on R. keithii in Sabah, Malaysia, did not reveal any significant excess tissue temperature. There were important differences in the methodology of the two studies. In the present study, we inserted thermocouples inside the tissue and measured every 6 or 20 s and recorded as means of 10 or 15 readings once per 1 or 5 min during the whole period of blooming (5–8 d). We detected 1.5–6.0 K excess temperatures in the internal parts of the flower. In Nais’s study, only the surface temperatures were measured, as spot readings, with a probe that was held against the surface.

During the day, the buds were cooler than the air (Fig. 3b), suggesting that evaporative heat transfer may be significant, and may affect the thermal balance of the buds cooling down the surface when the radiation load from the surroundings is higher and VPD increases. In the present study it was not possible to measure the stomatal conductance or transpiration rate of the floral surface, but stomata are present in the adaxial surface of the perigone lobe (Cammerloher, 1920). Consequently, we have not been able to construct the complete energy budget of the flower.

Further evidence that radiation is not the most important component of the heat supply comes from comparison of the time courses of the flux of radiant energy and (calculated) convection. When net radiation was maximum, convective heat loss was at a minimum (Fig. 7), suggesting that the energy is produced internally and is not derived from the net radiation in the forest understory.

The role of endothermy in flowers of R. tuan-mudae and R. lowii of the parasitic family Rafflesiaceae

The question posed was: is endothermy part of the mimicry of the flower to attract the pollinating flies? Endothermy was detected in young buds and mature buds of R. lowii and remained during blooming. This flower showed signs of thermoregulation by maintaining fairly constant temperature during and after anthesis (Patiño et al., 2000). Similarly, young buds of R. tuan-mudae showed some endothermy (mature buds were not studied for the reasons stated previously), but the flowers were shown to have only a weak pattern of endothermy and there was no sign of thermoregulation. Therefore, the mechanism for endothermy in these two flowers is present at the early stages of floral development. The photosynthetic products that provide the respiratory substrate for endothermy are available ‘free of charge’ from the vine host. There would, of course, be a certain limit imposed by the productivity of the host and the translocation rates: the vine might not have unlimited photosynthates for its parasites.

It has been demonstrated that many flowers increase their temperature above ambient as a consequence of endothermy, the heat being produced by the cyanide-insensitive respiration (Meeuse & Raskin, 1988; Skubatz et al., 1990). It is possible that endothermy is a facultative characteristic in most plants and perhaps all flowers. This assumption is supported by the fact that most plant mitochondria contain a cyanide and antimycin-insensitive alternative terminal oxidase (Lambers, 1980; Moreau & Romani, 1982). The alternative respiratory pathway has been detected in different tissues of many plants belonging to different taxa, for example ripening fruits (Cruzhernandez & Gomezlim, 1995), roots and cotyledons of soybean, potato, sweet potato and cassava (Day et al., 1994; Millar et al., 1994; Ribascarbo et al., 1995), sugar beet callus (Shugaev et al., 1998), Acer pseudoplatanus (Aubert et al., 1997), rootstocks of pears (Wagner et al., 1992; Tamura et al., 1996), nongreen tissues of Petunia hybrida (Wagner & Wagner, 1995), water-stressed plants of sorghum (Kumar & Sinha, 1994), shoot tips of Douglas fir (Fielder & Owens, 1992), roots of white spruce (Weger & Guy, 1991), Convolvulus (Van der Plas et al., 1977), wheat (Lundegårdh, 2001) and beans (Rychter et al., 1992).

As the tropical environment is rarely cold, it seems likely that endothermy in R. tuan-mudae and R. lowii is present as the result of alternative respiration that is not coupled to energy conservation. High rates of respiration are assumed to occur owing to the nature of the flowers: nonphotosynthetic and parasitic on the roots of the host (where they can easily can obtain the substrate of respiration). Therefore, endothermy may have evolved not merely to maintain the tissue at a high temperature, but to ensure pollination. The suggestion from this present work is that the production of CO2, and possibly other volatiles is important. The flowers in this case are conspicuous by virtue of the CO2 and/or volatiles they produce. The poorly mixed air at the forest floor may contain the olfactory signal to pollinators, but an important question is whether or not CO2 plays a role in the pollination of R. lowii and R. tuan-mudae. Further studies are now needed to prove the relationship between the high respiration rate of parasitic plants and attraction of pollinators.

In support of the role of CO2 as an insect attractant, it has been shown, for example, that CO2 is the only attractant volatile of the larvae of western corn rootworm (Diabrotica virgifera virgifera) to corn roots (Bernklau & Bjostad, 1998). The detection of CO2 by identified peripheral sensory organs of some terrestrial arthropods (e.g. nematodes, larva and adult beetles, centipedes, ants, termites, fig pollinators, honey bees, mosquitoes, flies, bugs, ticks and moths) is now well established, and the resulting coordinated behavioural responses at concentrations that occur naturally in the habitats of these organism has been well documented (Stange & Wong, 1993; Stange, 1996). Furthermore, there is evidence that the blowfly has CO2-specific sensory receptors (Stange, 1975). It has also been demonstrated that CO2 has an anaesthetic effect in blowflies and that they are considerably more sensitive to CO2 than to other anaesthetics (Diesendorf, 1975). It is therefore possible that the CO2 produced by the flowers of R. tuan-mudae and R. lowii plays a role on the pollination by the blowflies.

The signals emitted by R. tuan-mudae and R. lowii may be different. Despite Rhizanthes and Rafflesia attracting blowflies of the same genera Lucilia, Chrysomya and Hypopygiopsis (Beaman et al., 1988; Bänziger, 1991, 1996; Hidayati et al., 2000), it has been observed that R. lowii stimulated oviposition in the flies, suggesting that R. lowii is releasing specific volatiles that trick female flies. Oviposition by the blowflies on R. tuan-mudae was not observed in this study and has not been observed in other Rafflesia species (Bänziger, 1991; Beaman et al., 1988; Bänziger, 1996).

Further work is required to determine whether the olfactory signal is more complex than simply CO2. The volatiles identified in the headspace of R. lowii– 3-hydroxy-2-butanone, 2-ethyl-1-hexanol and N,N-diethyl-3-methyl-benzamide (S. Patiño, A. A. Edwards and J. Grace, unpublished) – have been found in other flowers (Knudsen et al., 1993). The volatiles identified in R. tuan-mudae– dimethyl disulphide and dimethyl trisulphide (S. Patiño, A. A. Edwards and J. Grace, unpublished) – have been found in various flowers from the family Araceae such as Amorphophallus, Pseudodracontium (Stransky & Valterova, 1999) and Hydrosme (Kite & Hetterschieid, 1997), in one Hydnoraceae (Hydnora) (Burger & Munro, 1988) and in bat-pollinated flowers (Knudsen & Tollsten, 1995), with dimethyl disulphide being the component preferred by the bats (von Helversen et al., 2000). Dimethyl disulphide and dimethyl trisulphide are compounds commonly related to bacterial growth on meat (Senter et al., 2000) and dimethyl disulphide is the compound that gives the characteristic taste to Camembert cheese (Demarigny et al., 2000). It has been found that the antennae of the female blowfly have a specific receptor neurone tuned to dimethyl disulphide (Park & Cork, 1999) and it has been suggested that dimethyl trisulphide may be one of the major cues for host location by calliphorid flies (Nilssen et al., 1996). Location of the flower by the flies may follow the classical downwind model (Stange, 1996) and once the flies have landed on the flower, mechanical and contact chemical inputs may guide them into the diaphragm and to the gynoecium. Although, Rafflesia is not a ‘trap’ flower, as defined by Dafni (1984), it has been observed that a few flies remain inside the R. tuan-mudae diaphragm for some hours (pers. obs.), perhaps because of the CO2 anaesthetic effect, as suggested by Dafni (1984).

Further work is now needed on the behaviour of the pollinating flies and identification of the species visiting the flowers.

A further step in elucidating the role of carbon dioxide in the pollination syndrome of Rafflesia tuan-mudae and Rhizanthes lowii would be to perform direct measurements of gas exchange, and careful identification of the volatiles that compose the odour, with bioassays of the substances using the pollinating flies and other species of Rafflesia and Rhizanthes. Experiments with model flowers, in which the odours are experimentally emitted, would enable identification of the role of specific gases as well as the importance of size, shape and colour of the flower.


Sandra Patiño was funded from Colombia by a Colciencias scholarship and supported by the Instituto de Investigación de Recursos Biológicos Alexander von Humboldt. We acknowledge financial support from the Davies Expedition Fund and Development Trust of the University of Edinburgh. We thank Ernest Chai and the staff at Taman Negara Gunung Gading in Sarawak and Sem Pasam, Jamel. We thank Batien and John Reky for locating the sites where R. tuan-mudaei grows and Dr Jamili Nais in Sabah for assistance and providing access to his thesis and bibliography. We also thank Laure Grison and Emmanuelle Jousselin for help in the field during the last trip to Sarawak. We thank Dr Timo Vesala from the Physics Department at the University of Helsinki for his kind collaboration and the Finnish Centre of Scientific Computing for providing computational resources. We especially thank Dr Hans Bänziger and Dr E. Allen Herre for useful comments and criticism that improved substantially the final version of this manuscript.