See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237155009

Thermogenic flowering of taro (Colocasia esculenta, Araceae)

Article  in  Canadian Journal of Botany · February 2011

DOI: 10.1139/b04-118

CITATIONS READS
23 3,859

5 authors, including:

Vincent Lebot Olivier Roupsard

Cirad - La recherche agronomique pour le développement Cirad - La recherche agronomique pour le développement
153 PUBLICATIONS   3,560 CITATIONS    166 PUBLICATIONS   11,583 CITATIONS   

SEE PROFILE SEE PROFILE

José Quero Garcia Tom Okpul

French National Institute for Agriculture, Food, and Environment (INRAE) The Papua New Guinea University of Technology
80 PUBLICATIONS   1,231 CITATIONS    45 PUBLICATIONS   329 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

MACACC View project

Tàrogen View project

All content following this page was uploaded by José Quero Garcia on 09 October 2014.

The user has requested enhancement of the downloaded file.

 1557

Thermogenic flowering of taro (Colocasia

esculenta, Araceae)
Anton Ivancic, Vincent Lebot, Olivier Roupsard, José Quero Garcia, and
Tom Okpul

Abstract: Thermogenesis and its association with taro (Colocasia esculenta (L.) Schott) flowering was studied during
the warmest period of the year (December 2002 – February 2003) within a large collection of heterogeneous plant ma-
terial on Espiritu Santo, Vanuatu. On each studied inflorescence, temperatures of the three main parts of the spadix and
the ambient air were recorded during a period of 38 h. The investigation indicates that significant thermogenic activity
of taro inflorescences takes place during two successive nights: (1) during the night when an inflorescence becomes
odorous (the female phase) and (2) a night later, when microsporogenesis approaches its final phase (the male phase).
The highest average difference between mean temperatures of the ambient air and inflorescences were documented dur-
ing the female phase, at 0500 hours (the mean temperature of the sterile appendix was 29.1 ± 0.9 °C (P = 0.05) and
this was 6.8 °C above the temperature of the ambient air). Thermogenic activity is synchronized with the protogynous
nature of the species and insect pollination in the early morning hours. Its main putative functions are (1) to reduce the
deviations of ambient air temperatures during the most critical periods of flowering, and (2) to promote cross-
pollination. It stops 1–1.5 h after pollen has been released.
Key words: taro, Colocasia esculenta, thermogenesis, inflorescence development, pollination.
Résumé : La thermogénèse de la floraison du taro (Colocasia esculenta (L.) Schott) a été étudiée pendant la période la
plus chaude de l’anné (Décembre 2002 à Février 2003), au sein d’une collection importante de germoplasmes présen-
tant une forte diversité, sur l’île de Espiritu Santo, au Vanuatu. Sur chacune des inflorescences étudiées, les températu-
res des trois principales zones du spadice et de l’air ambient ont été enregistrées pendant une période de 38 h. Ce
travail indique qu’il existe une thermogénèse significative des inflorescences de taro, et qu’elle a lieu durant deux nuits
successives : (1) durant la nuit où l’inflorescence devient odorante (phase femelle); et (2) une nuit plus tard, lorsque la
microsporogénèse approche sa phase finale (phase mâle). Les plus fortes différences entre les températures moyennes et
la température de l’air ont été enregistrées durant la phase femelle, à 0500 (la températue moyenne de l’appendice sté-
rile est alors de 29,1 °C ± 0,9 °C (P = 0,05) soit 6,8 °C de plus que la température de l’air ambiant). L’activité de la
thermogénèse est synchronisée avec la protogynie de l’espèce et la pollinisation entomophile durant les premières heu-
res du jour. Ses fonctions supposées sont : (1) de réduire les écarts de température de l’air ambient durant les périodes
critiques de la pollinisation; et (2) de promouvoir les pollinisations croisées. Cette activité cesse entre 1 h et 1,5 h
après l’émission de pollen.
Mots clés : taro, Colocasia esculenta, thermogenèse, inflorescence développement, pollinisation.

Ivancic et al. 1565

Introduction monocots (Araceae, Arecaceae, Cyclanthaceae) (Knutson

1974; Meeuse 1975; Seymour and Schultze-Motel 1996,
Thermogenesis has been documented in several families 1997; Patino et al. 2000; Thien et al. 2000; Lamprecht et al.
of dicots (Annonaceae, Aristolochiaceae, Illiciaceae, Magno- 2002a, 2002b; Seymour et al. 2003). Among the Araceae,
liaceae, Nelumbonaceae, Nymphaeaceae, Rafflesiaceae) and thermogenesis appears to be common. It has been studied in

Received 26 February 2004. Published on the NRC Research Press Web site at http://canjbot.nrc.ca on 16 November 2004.
A. Ivancic.1 Faculty of Agriculture, University of Maribor, Vrbanska 30, 2000 Maribor, Slovenia.
V. Lebot. Centre de coopération internationale en recherche agronomique pour le développement (CIRAD), P.O. Box 946, Port
Vila, Vanuatu.
O. Roupsard. Centre de coopération internationale en recherche agronomique pour le développement (CIRAD) – Vanuatu
Agricultural Research and Training Center (VARTC), P.O. Box 232, Espiritu Santo, Vanuatu.
J. Quero Garcia. Centre de coopération internationale en recherche agronomique pour le développement (CIRAD), 34398
Montpellier CEDEX 01, France.
T. Okpul.2 University of Vudal, PMBS, Rabaul, East New Britain, Papua, New Guinea.
1
Corresponding author (e-mail: [email protected]).
2
Present address: Sir Alkan Research Center, National Agricultural Research Institute (NARI), P.O. Box 1639, Lae, MP411, Papua
New Guinea.

Can. J. Bot. 82: 1557–1565 (2004) doi: 10.1139/B04-118 © 2004 NRC Canada
1558 Can. J. Bot. Vol. 82, 2004

Fig. 1. Development of a taro (Colocasia esculenta) inflorescence and fruit head over an 18-d period (the sketching took place at the
beginning of 2003, in the morning hours between 0730 and 0800, at the experimental plot of the Vanuatu Agricultural Research and
Training Center near Luganville, on the Island of Espiritu Santo, Vanuatu): 1, a 5-d-old inflorescence coming out of the membranous
flag leaf (8 d before becoming odorous); 2, 7 d old; 3, 9 d old; 4, 11 d old; 5, 12 d old (1 d before becoming fragrant); 6, 13 d old
(the release of odor; the spathe is loose, partly open, and odorous, expressing its final color, the stage when pollinating insects enter
the floral chamber); 7, 14 d old (the release of pollen, the upper part of the spathe is fully open, whereas its constricted part is tight,
preventing self-pollination); 8, 16 d old (3 d after becoming odorous); 9, 18 d old.

several genera such as Alocasia, Amorphophallus, Anubias, Amorphophallus campanulatus Bl. ex Decaisne), taro inflo-
Arum, Cercestis, Dieffenbachia, Dracunculus, Helico- rescences are small, having a relatively thick spathe. They
diceros, Homalomena, Philodendron, Symplocarpus, and appear in groups or clusters, usually from two to five, de-
Sauromatum (Nagy et al. 1972; Meeuse 1975; Skubatz et al. pending on the genetic structure, plant age, and the environ-
1991; Yafuso 1993; Bay 1995; Seymour 1999; Seymour and ment; the latter is probably the most important. The highest
Blaylock 1999; Seymour and Schultze-Motel 1999; Barabé number of inflorescences per cluster recorded is five (Young
and Gibernau 2000; Gibernau and Barabé 2000; Seymour et 1924; Ivancic and Lebot 2000).
al. 2003). Thermogenesis within the genus Colocasia has Each inflorescence consists of a spadix covered by a
been known for about a century and was mentioned for the spathe. The spadix is divided into a female part (lower part),
first time by Leick (1915). a sterile region, a male part, and a sterile tip (appendix). The
Taro (Colocasia esculenta (L.) Schott), which is one of female portion is generally shorter than the male portion.
the most important aroid root crops, has many typical aroid Female (pistillate) flowers are sessile and green, with well-
characteristics, one of them being thermogenic activity of its developed stigmas and are usually mixed with pistoids (ster-
inflorescences. Many taro botanists and breeders are familiar ile female flowers), which can be distinguished by their light
with it, although it has not yet been systematically analyzed. color. The male part of the spadix consists of sessile stami-
There are several reasons for this: flowering is relatively nate (male) flowers. According to Backer and Bakhuizen
rare, inflorescences (spadices) are small (especially the ster- van den Brink Jr. (1968) and Shaw (1975), male flowers
ile appendix), and thermogenic activity is little expressed. consist of two to six sessile anther-cells, which are connate
Taro is an allogamous, monoecious, and protogynous spe- into more or less obpyramidal synandria. The apex of the
cies. Each floral unit includes an inflorescence placed on a synandria is mostly flat and hexagonal. Thecae are dehiscent
floral stalk or peduncle (Figs. 1 and 2). Compared with other by the terminal pore. The pollen grains are spherical and
aroid species (e.g., Alocasia macrorrhizos (L.) G. Don, trinucleate.

© 2004 NRC Canada

Ivancic et al. 1559

Fig. 2. A taro (Colocasia esculenta) inflorescence: 1, the struc- ers. Most of these insects will remain inside the inflores-
ture of a spadix. a, peduncle; b, female portion with pistillate cence until the next morning, when the spathe will be fully
flowers; c, constricted sterile region between female and male open and pollen will be released.
portions; d, male portion with staminate flowers; e, sterile appen- Insects are the most important agents of cross-pollination.
dix; 2, transsection of an inflorescence in the morning when it Wind pollination can be significant only for openly flower-
became odorous; 3, the same inflorescence 24 h later. a, the ing genotypes (with a fully exposed male portion) grown in
length of the lower part of the spathe before elongation; b, the dense populations. Rain mainly causes self-fertilization by
difference of the length after elongation. washing pollen grains from the male part of the spadix to the
pistillate region. Self-fertilization is possible, because the in-
compatibility system at the end of flowering becomes less
efficient, and because there is an overlap between the period
of stigma receptivity and the release of pollen. Stigmas
(in humid conditions of Morobe Province in Papua New
Guinea) can remain receptive for 10 days (Okpul and
Ivancic 1995).
Taro is a very sensitive species characterized by rare and
erratic flowering (Taro Network for Southeast Asia and
Oceania (TANSAO) 2002). Many of the genotypes never
flower. For normal flowering and seed set, taro requires an
optimal environment. In the situations of heavy rains or con-
tinuous rainy weather, or large deviations of ambient air
temperatures, plants do not flower or produce sterile flowers
(Ivancic and Lebot 2000).
The hybridization technique is not complicated, and this is
mainly because of monoecy and protogyny. The inflores-
cence has to be emasculated before the spathe becomes
attractive for insects (in most cases this is 1 d before it be-
comes odorous). The emasculation procedure is very simple.
With a small knife, one has to cut the inflorescence (the
spadix and the spathe) in the constricted region (in the
sterile zone between the male and female portions of the
spadix). The female portion remains protected with the
The spathe consists of two parts. The lower part is usually lower part of the spathe. Pollination is usually carried out
green or red, and forms the floral chamber in which female immediately after emasculation. At first, the lower part of
flowers are located. The upper part is predominantly yellow the spathe is carefully removed, then pollen is distributed on
but sometimes can be purple, red, or blotched (associated stigmas, and finally the pollinated female portion has to
with the pigmentation of the whole plant). The shape and be protected by cotton or simply by the lower part of the
size of the spathe are highly variable, depending on geno- spathe. One of the main problems associated with hybridiza-
type, plant size, and environmental conditions (Ivancic tion is the exactness of crosses, especially in areas with
1995). During anthesis, the lower part of the spathe remains abundant insect pollinators. If taro generates temperature
more or less closed and thus protecting the female portion, during anthesis, then thermogenesis should be considered as
whereas the upper part opens and curves towards the oppo- a factor that may influence hybridization results. From this
site side of the opening and consequently exposing the upper point of view the data concerning thermogenic activity can
part of the spadix (the male portion and the sterile appen- be very useful for taro breeders.
dix). The paper represents the first systematic study of thermo-
In the tropical Pacific (e.g., in Papua New Guinea, Solo- genesis and its association with flowering of the “true” taro
mon Islands, Vanuatu), the most intense flowering takes (Colocasia esculenta). The main objectives are to analyze:
place at the beginning and at the end of the rainy season. (1) the biology of inflorescences and its association with the
The beginning of taro flowering is usually associated with prevention of self-pollination; (2) the dynamics of thermo-
the emission of strong odor (mainly from the spathe). Its genic activity; (3) its role in promoting cross-fertilization;
main purpose is to attract insect pollinators. The most and (4) its relationship with other processes associated with
important of these are small flies belonging to the anthesis. The paper will also discuss the possible implica-
Drosophilidae family (Carson and Okada 1980; Matthews tions of the results on the crossing technique.
1990). Flies from this family, although different species, can
also be found in other closely related aroids such as Alo- Materials and methods
casia odora (G. Lodd.) Spach (Yafuso 1993; Miyake and
Yafuso 2003). The investigation took place from 5 December 2002 to
An odorous inflorescence is erect, and its spathe is loose 4 February 2003, at the Vanuatu Agricultural Research and
(Fig. 1, stage 6), enabling small pollinators (coming from Training Center (VARTC) near Luganville, on the Island of
a day older, pollinating inflorescences) to enter the floral Espiritu Santo, Vanuatu (15°26.7′S, 167°11.5′E). The experi-
chamber and distribute pollen on stigmas of the female flow- mental plot was located on a deep and fertile soil, covering a

© 2004 NRC Canada

1560 Can. J. Bot. Vol. 82, 2004

coral limestone plateau, located 3 km from seashore at 80 m measured in the outer zone of the spongy tissue of the fe-
altitude. It was the end of the dry and the beginning of the male and male portion (the measuring points were at one
wet season, which was characterized by relatively high tem- half of their lengths) and the sterile appendix (the measuring
peratures and very intensive flowering. The plant materials point was at one third of its length, where normally devel-
used in this investigation consisted of hybrids between wild oped appendices were on average the thickest).
and cultivated germplasm, and flowering accessions from The temperatures were measured by the electronic ther-
the national germplasm collection. These materials were mometer Ebro TFN 1093 SK (sensor NiCr-Ni (thermocouple
used as parental components for hybridization within the ex- K), 175 mm × 0.9 mm), having the resolution of 0.1 °C and
isting recurrent selection programme. The main reason for accuracy of ±0.4 °C ± 1 digit (–10 … +80 °C).
using hybrids was their regular and natural flowering, rela- Climatic parameters were measured continuously at the
tively large inflorescences, and well-developed sterile ap- height of 22 m on a tower located 750 m away from the
pendices. Without hybrids, the sample would be too small, measurement site. They were monitored on Campbell Scien-
and flowering would have to be induced artificially (by tific 10× data logger every 30 s and averaged semi-hourly.
treating plants with the gibberellic acid GA3). The artificial Global radiation (Rg), silicon cell pyranometer SKS1110
induction of flowering is probably not the best solution for (Skye Inst. Ltd., Powys, UK); air temperature and relative
the investigation of the thermogenic activity. Treatments humidity, MP100a (Campbell Scientific, Shepshed, Lei-
with gibberellic acid can cause serious deformations of in- cestershire, UK); rainfall, ARG100 (Campbell Scientific).
florescences (Ivancic 1995).
The study of the role of the spathe in the reproduction Results
system (especially in preventing self-pollination) was based
on the analysis of its growth and related changes. It included Climatic parameters
both parts of the spathe: the lower part (which enfolded the The climate is tropical oceanic, averaging 2745 mm (±SD
pistillate portion of the spadix and forms the floral chamber) 710 mm) rainfall per year (1989–2000). Rainfall peaks dur-
and the upper part (which enfolded the male portion and the ing the hot and rainy season (December-April), with an aver-
sterile appendix). Based on preliminary studies, we chose age of 335 mm@month–1. A drier season usually occurs from
(for studying the developmental changes of spathes) the pe- July to September, with an average of 117 mm@month–1. Av-
riod of 39.5 h from 1800 (approximately 6 h before releas- erage daily global radiation, maximum temperature, air hu-
ing odor) to 0930 (approximately 3 h after the first pollen midity, maximum vapor pressure deficit, and daily potential
had been released). The spathes were marked with two lines evapotranspiration (Priestley–Taylor) are, respectively,
(the first at the bottom and the second in the middle of the 20.0 MJ@m–2@day–1, 30.3 °C, 89.8%, 10.8 hPa, and 5.4 mm
constricted area), and distances (between the first and the during the rainy season, and 14.5 MJ@m–2 day–1, 27.6 °C,
second line, and between the second line and the tip) were 86.1%, 8.1 hPa, and 3.5 mm during the dry season. The pe-
recorded every 12 h. riod December 2002 – March 2003 represented the warmest
The investigation of thermogenesis involved 145 flower- period of the year. The average air temperatures (at VARTC)
ing plants, while the total number of studied inflorescences were the following: 25.3 °C in December, 25.6 °C in January,
was 240 (small and deformed inflorescences were excluded). and 26.0 °C in February and March. The lowest temperatures
It was concentrated on two developmental periods: the “fe- (in most instances, they were recorded between 0300 and
male phase” (which took place during the night when odor 0500) were as follows: 20.5 °C in December, 20.4 °C in Janu-
was released) and the “male phase” (which took place a ary, and 21.7 °C in February. December was relatively very
night later and ended after the release of pollen). On each dry (51.7 mm rain). January and February had some more
studied inflorescence, temperatures were recorded during the rain (190 and 237 mm, respectively) but still not enough. To
period of 38 h; starting at 2100 (in the night when an inflo- ensure an optimal environment, it was necessary to irrigate
rescence became odorous) and ending at 1100 (approxi- twice per week (using 12 000 L of river water@ ha–1).
mately 3–4 h after pollen had been released). The
determination of this period was based on a series of prelim- Biology of inflorescences and prevention of self-
inary recordings, which took place in January 2002 and fertilization
included the whole developmental period of taro inflorescen- The developmental period of inflorescences, recorded
ces, lasting 12–16 d. from the moment when the upper part of the spathe came
Temperatures were measured every 10 min, and later aver- out of the membranous flag leaf, to the moment when pollen
ages were computed for every 30 min. Before measuring, was released (Fig. 1, stages 1–7), varied from 14 to 17 d.
the spathe was gently loosened and partly unfolded, to deter- Anthesis lasted 2 d, and its duration was very uniform. It
mine the measuring points of the spadix. These points were was associated with several visible changes in an inflores-
marked on the outer surface of the spathe. Then the thin sen- cence. The approaching of the female phase was character-
sor of the thermometer (its end was slightly curved) was ized by the elongation of the peduncle, spathe, and spadix
carefully inserted inside. Damaged inflorescences as well as (Fig. 1), and the change of color of the upper part of the
those with very small sterile appendices were abandoned. spathe. The female phase began when the upper part of the
The following temperatures were recorded: ambient air spathe attained its final color (i.e., yellow, purple, red, or
(under the shade of a fresh leaf, at the height of the base of blotched) and started to unfold. This happened late in the af-
the studied inflorescence) and spadix. The spadix tempera- ternoon, approximately 10–12 h before the release of odor.
tures represented the compromise between the temperature The lower part remained green or nearly green. The first
of the surface and the spongy tissue in the center and were odor appeared between midnight and 0100. Its intensity was

© 2004 NRC Canada

Ivancic et al. 1561

weak and gradually increased until 0730–0830 (depending approximately 2 °C higher than the temperature of the ambi-
on genotype and weather conditions). The odorous spathes ent air (the mean temperature of the ambient air was 24.2 ±
were relaxed, enabling small dipterous insects to enter the 0.3 °C). The mean temperature of the male portion was
floral chamber. The mean length of spathes (measured on 26.2 ± 0.7 °C. The second highest was the mean temperature
4 February 2003) was 28.20 ± 2.1 cm (P = 0.05). The of the sterile appendix (26.1 ± 0.9 °C). The mean tempera-
spathes continued to elongate until the next morning (when ture of the female portion was 26.1 ± 1.1 °C. The spadix tis-
pollen was ready to be released). sues were warmer mainly because they were cooling slower
The elongation of the spathe was reduced and stopped af- than the ambient air.
ter pollen had been released. The average elongation deter- The thermogenic activity increased rapidly after 0330, and
mined for the period from 0830 to 1830 (on 4 February the most active was the sterile appendix, reaching its maxi-
2003), was 1.33 ± 0.1 cm, and this was mainly due to the mum at 0500 with the mean temperature 29.1 ± 0.9 °C
growth of the lower part of the spathe. This part continued to (6.8 °C above the temperature of the ambient air). At the
grow during the following night, and its average elongation same time the mean temperature of the male portion was
from 1830 to 0530 was 0.48 ± 0.1 cm. During these night 28.8 ± 1.0 °C (6.4 °C above the temperature of the ambient
hours, the spadix stopped elongating, and consequently the air), whereas that of the female portion was 24.9 ± 0.6 °C
constricted part of the spathe “moved” upward toward the (only 2.6 °C above the temperature of the ambient air).
thick bottom of the male portion and closed the space be- The female portion generated maximum temperatures be-
tween the male and the female portions (Fig. 2, illustrations tween 0700 and 0900, reaching its peak at 0800, with the
2 and 3). At the same time the lower part of the spathe (in- average temperature 32.2 ± 0.6 °C (4.8 °C above the temper-
cluding the constriction) tightened. In this way, the female ature of the ambient air).
portion was completely isolated, and self-pollination was After 0800, the thermogenic activity ceased, and the tem-
prevented. The upper part of the spathe, however, continued peratures of the thermogenic tissues slowly approached the
to loosen and unfold, and reached its maximum opening temperatures of the ambient air. During the warmest period
later in the morning, between 0745 and 0930, depending on of the day, no thermogenic activity was recorded. The fe-
weather conditions. The first pollen was observed at 0625 male portion remained warmer than the male portion and the
(on 5 February 2003), however, most of the inflorescences sterile appendix (Fig. 3). One of the reasons could be darker
started releasing pollen 0.5 h later. On cool and wet days, pigmentation of the lower part of the spathe (which absorbed
the first pollen appeared after 0800. On rainy days, most in- more solar radiation). Another reason could be evaporation
florescences did not release pollen. of odorous substances from the upper part of the spathe,
There were obvious differences regarding the space avail- which may act as a cooling system for the male portion and
able for the movement of insect pollinators inside odorous the sterile appendix.
inflorescences, during the night and early morning hours. During the female phase, the generation of temperature
The spathe of some extreme genotypes loosened so much did not appear to depend much on the weather conditions
that even smaller honey bees could enter and move around (dry weather, rain, wind), because temperature-generating
the spadix. The other extremes were inflorescences that re- organs were still well protected by the spathe.
mained more or less completely closed with their spathes
slightly opened after pollen had been released. There were The male phase
also a few situations where the constricted part of the spathe During this phase (which started when microsporogenesis
did not close the space between the female and male por- approached its final stage and ended after the release of pol-
tions (allowing self-pollination). Differences associated with len), only the male portion of the spadix was thermo-
the behavior of the dipterous pollinating insects were less genically active. Its temperatures started to rise late in the
obvious; however, it was possible to see some of them mov- evening, between 2000 and 2200 (Fig. 3), depending on
ing among inflorescences during the entire day and most of genotype and weather conditions. At 2200, the mean temper-
the night. ature of the male portion was 27.0 ± 0.4 °C (the mean tem-
perature of the ambient air was 24.3 ± 0.4 °C). It was
Thermogenic cycles gradually increasing until 0730 (the time when most of the
Significant thermogenic activity took place from midnight pollen was released), reaching 30.6 ± 0.7 °C (3.38 °C above
to 0930, during two successive nights: (i) the night when the the temperature of the ambient air). The highest difference,
inflorescence became fragrant (the female phase), and however, was estimated much earlier, at 0400 (7.45 °C
(ii) the night when microsporogenesis approached its final above the temperature of the ambient air). This difference
phase (the male phase). The female phase ended approxi- decreased rapidly once the spathe was more or less fully
mately 12 h after the first appearance of odor. After the end open (after 0730). The highest value (36.1 °C) was recorded
of this phase, there was and interval (8–10 h long) without on January 28, at 0724 (when the ambient air temperature
thermogenic activity, which was followed by the male phase was 25.8 °C). The thermogenic activity ended between 0830
(Fig. 3). and 0930 (1–1.5 h after pollen had been released).
The mean temperatures of the sterile appendix were not
The female phase much lower when compared with the male portion (in most
During this phase (Fig. 3), the generation of temperatures situations 1–2 °C), and this was probably due to the heat
started before midnight, although temperature differences from the male portion. The male portion was also heating
between the spadix tissues and the ambient air were rela- the female portion (in most situations, the mean temperature
tively small. At 2200, the temperatures of these tissues were of the female portion was 1–5 °C lower).

© 2004 NRC Canada

1562 Can. J. Bot. Vol. 82, 2004

Fig. 3. The course of temperature differences between the three main parts of the spadix of taro (Colocasia esculenta) and the ambient
air during the period of 38 h, starting at 2100 (during the night when an inflorescence became odorous) and ending at 1100 (approxi-
mately 3–4 h after pollen had been released). The data were collected during the warmest period of the year (December 2002 – Febru-
ary 2003) on the Island of Espiritu Santo, Vanuatu.

During rainy days, the thermogenic activity of the male could have an important role in attracting insects that were
phase appeared to be less expressed. In many situations, es- not specialized for taro.
pecially after longer periods of heavy rains, the thermogenic The spathes of the pollinating inflorescences of most of
activity stopped or was reduced to a very low level (because the genotypes were widely open, so that insects could easily
the spathes were already partly open and tiny spadices were depart, carrying pollen on their bodies to the inflorescences
exposed to rain), so that there were no significant differences that were 1 d younger and intensively odorous. The female
between the temperatures of the male portion of the spadix flowers of these intensively odorous inflorescences were
and the ambient air. In such situations, most of the inflores- (because of protogyny) already fully receptive.
cences did not release pollen. Protogyny is probably not sufficient for preventing self-
pollination, because stigmas in wet conditions may remain
Discussion receptive for several (up to 10) days (Okpul and Ivancic
1995). The investigation indicated that the key role belonged
Biology inflorescences and prevention of self- to the constricted region of the spathe. Until the end of the
fertilization female phase, the growth of the spathe and the spadix were
Size, shape, color, and odor of spathes, structure of synchronized, so that the constricted part of the spathe re-
spadices, and thermogenic activity indicate that taro is a mained in the middle of the thin sterile region (between the
cross-pollinating species adapted to insect pollination. For female and the male part of the spadix), leaving enough
the evolution of this species, cross-pollination was probably space for the insects carrying pollen from other inflorescen-
crucial. In the case of self-pollination, combined with vege- ces to reach the floral chamber. As the moment of the re-
tative propagation, its evolution would be very slow. lease of pollen approached, the space between the female
The mechanism that regulated cross-pollination appeared and the male portions closed.
to be very efficient. The appearance of intensive odor and Another factor preventing self-fertilization is self-
the release of pollen (on 24-h-old inflorescences) were per- incompatibility. Although the incompatibility mechanisms of
fectly synchronized. The investigation showed that the maxi- taro have not yet been systematically studied, practical expe-
mum intensity of odor was reached between 0800 and 0830 rience indicates that they may not be as strong as those of
(0.5–1.5 h after the release of pollen on 24-h-old inflores- some other species such as sweet potato (Ivancic and Lebot
cences) and then it started to decrease. At 1900, it was al- 2000).
ready very weak and it was difficult to sense. When the
weather was cool and cloudy, it could remain present until Thermogenic cycles
the release of pollen in the following morning. The presence The thermogenic activity of taro inflorescences probably
of odor (although very weak) on pollinating inflorescences has a long and complex evolution. It consists of two non-

© 2004 NRC Canada

Ivancic et al. 1563

overlapping cycles and represents an important factor that atively low and changeable, and thermogenesis was proba-
helps to ensure successful fertilization and seed set in a wide bly crucial to assure optimal and stable temperatures. With-
range of environments. It might have two main functions: out it, microsporogenesis would probably last longer.
(i) to reduce deviations from optimal ambient air tempera-
tures during the most critical periods of flowering, such as
final stages of gametogenesis and early stages of embryo Pollination
development, and (ii) to promote cross-pollination. The Inflorescences in the female phase attracted pollinators in
maximum increase of the inflorescence temperature due to the early morning hours by odor and color of spathes, and
thermogenic heating was approximately 7 °C above the tem- perhaps infrared radiation. Odor was obviously the most
perature of the ambient air and, on tropical Espiritu Santo, important attractant. It has been demonstrated in aroids
this was probably sufficient to maintain an optimal environ- (Meeuse 1975) that odor is produced as a result of volatiliza-
ment during the most critical stages of the floral develop- tion of various compounds caused by heating. Its intensity
ment. The adaptation to a cooler climate would probably was gradually increasing with a current increase of tempera-
require the generation of higher temperatures. The examples ture. Such a phenomenon is relatively common among
of such an adaptation are inflorescences of the dead horse thermogenic plants; e.g., Alocasia odora (Miyake and
arum (Helicodiceros moscivorus), which can generate tem- Yafuso 2003), Amorphophallus paeoniifolius (Dennst.)
peratures up to 15 °C above those of the air (Seymour et al. Nicolson, Sauromatum guttatum (Wall.) Schott, and Victoria
2003), and in that way maintain a more or less stable floral cruziana Orb. (Lamprecht et al. 2002a). Infrared radiation
temperature at about 24 °C. was probably less important, because the majority of taro
The thermogenic activity, which took place during two pollinators on Espiritu Santo started moving to odorous
successive nights (i.e., during the female and the male inflorescences when darkness was over (in January, it was
phase), appeared to be well synchronized with the period of after 0510) and the air temperature rose above 24 °C.
maximal receptivity of stigmas, the emission of intense odor, The spathe started to loosen late in the afternoon, between
the release of pollen, and activities and movements of polli- 1600 and 2000, depending on genotype and environmental
nating insects. Such a synchronization is very common in conditions (e.g., weather conditions, presence of shade). The
the Araceae family (Bay 1995; Seymour and Blaylock 1999; environmental conditions appeared to be much more impor-
Gibernau and Barabé 2002; Miyake and Yafuso 2003). tant.
Compared with some thermogenic species such as Ne- After 0500, the upper part of the spathe was already partly
lumbo nucifera Gaertn. (Seymour et al. 1998) and Alocasia unrolled (although its edges remain overlapped), enabling
odora (Yafuso 1993; Miyake and Yafuso 2003), the duration pollinating insects to enter and reach the male portion and
of thermogenic activity of taro inflorescences was found to the sterile appendix. Inflorescences with a sufficiently large
be relatively short. However, this period is not shorter than opening on the lower part of the spathe (to enable insects to
that of Arum italicum Mill. (Barabé et al. 2003), Helico- reach the pistillate flowers directly) were extremely rare.
dicerus muscivorus Schott ex K. Koch (Seymour et al. This characteristic, which is more frequent in Papua New
2003), Montrichardia arborescens L. (Gibernau et al. 2003), Guinea (the example is the cv. Kowune from the Eastern
and Philodendron species (Gibernau and Barabé 2000). Highlands), appears to be genetically controlled. In the
During the female phase, most of the heating was pro- morning, when odor is released, the lower part of the spathe
duced by the sterile appendix and the male portion. Their (which protects the pistillate flowers) opens, whereas the up-
temperature peaks were reached very early in the morning per part (which protects the staminate flowers and the sterile
(at 0500). Thermogenesis of these two portions probably in- appendix) slightly unrolls itself. The lower part of the spathe
teracted; however, it was difficult to demonstrate. The spe- then closes, whereas the upper part opens.
cific structure of taro inflorescences and the narrow space Relatively high temperatures inside the upper part of the
inside the spathe suggested that the male portion was heating floral chamber increased the insects’ activity and move-
the sterile appendix and vice versa. The exceptions were ments, so that they reached also the female portion and dis-
inflorescences with a very small appendix (smaller than tributed pollen. It was also possible that the temperature in
6 mm × 2 mm), which probably had no significant influence the upper part of the spadix was too high to be comfortable,
on the temperature of the much larger male portion (having and therefore insects moved to a bit cooler female portion.
dimensions 45–48 mm × 4.2–4.5 mm). The temperature Their movement was slow and most of them were crawling
course of the female portion differed significantly from that on the rough spadix surface, because the space inside the
of the male portion and the sterile appendix. The peak was floral chamber was narrow and the spathe wall was slippery;
reached relatively late (at 0800), and this suggests that the in many ways similar to the situations described by Bay
female portion received at least some (if not all) heating (1995) in the genus Arum.
from the upper portions of the spadix, which were therm- Most of the insects remained inside the inflorescence, be-
ogenically much more active (such heating was possible, be- ing protected from direct insulation and extreme tempera-
cause the constricted region between the female and the tures during the day and the following night. In most
male portion was not yet closed). situations they were not trapped inside, because the spathe
The course of the thermogenic activity of the male phase remained “relaxed” and (or) continued to unroll. The excep-
was synchronized with the final stages of microsporogenesis tions were spathes that gradually closed, as the air tempera-
and the release of pollen. The final stages of the micro- ture approached 30 °C, and started to reopen late in the
sporogenesis took place when the air temperatures were rel- afternoon. Among the exceptions were also a few genotypes

© 2004 NRC Canada

1564 Can. J. Bot. Vol. 82, 2004

with inflorescences that never opened (most of them were References

local cultivars with very small inflorescences).
During the night, before the release of pollen, the heating Backer, C.A., and Bakhuizen van den Brink, R.C., Jr., 1968. Flora
was generated only by the male portion. The consequence of Java. Vol. 3. Wolters–Nordhoof, Groningen, the Netherlands.
was the movement of insects to the upper (warmer) part of Barabé, D., and Gibernau, M. 2000. Étude comparative de la pro-
the spadix. The lower part of the spathe elongated and be- duction de chaleur chez quelques Araceae. Adansonia, 22: 253–
263.
came tight, and consequently closed the space between the
male and female portions (to prevent self-pollination). In Barabé, D., Lacroix, D., and Gibernau, M. 2003. Development of
the flower and inflorescence of Arum italicum (Araceae). Can. J.
other related aroids such as Alocasia odora (Yafuso 1993),
Bot. 81: 622–632.
the situation appears to be very similar, although the role of
Bay, D.C. 1995. Thermogenesis in the Aroids. Aroideana, 18: 32–39.
the elongation of the lower part of the spathe has not been
Carson, H.L., and Okada, T. 1980. Drosophilidae associated with
studied. In some aroids, e.g., Arum italicum (Barabé et al.
flowers in Papua New Guinea; I. Colocasia esculenta. Kontyu,
2003), the female and male portions separate because of the
Tokyo, 48: 15–29.
elongation of the inflorescence axis.
Gibernau, M., and Barabé, D. 2000. Thermogenesis in three Philo-
After 0400, the upper part of the spathe started to unfold; dendron species (Araceae) of French Guiana. Can. J. Bot. 78:
at first slowly and later (after 0600) faster. When pollen was 685–689.
released (between 0730 and 0800), it was already open, ex- Gibernau, M., and Barabé, D. 2002. Pollination ecology of Philo-
posing the upper part of the spadix (the male portion and the dendron squamiferum (Araceae). Can. J. Bot. 80: 316–320.
sterile appendix). The pollinating insects lost their shelter Gibernau, M., Barabé, D., Labat, D., Cerdan, P., and Dejean, L.
and were exposed to lower atmospheric temperature. Dusted 2003. Reproductive biology of Montrichardia arborescens
with pollen, they had to search for new shelters, and these (Araceae) in French Guiana. J. Trop. Ecol. 19: 103–107.
were a day younger (odorous) inflorescences. Ivancic, A. 1995. Abnormal and unusual inflorescences of taro,
The results of this investigation indicate that therm- Colocasia esculenta (Araceae). Aust. J. Bot. 43: 475–489.
ogenesis is one of the key factors promoting cross- Ivancic, A., and Lebot, V. 1999. Botany and genetics of New Cal-
pollination. Although its influence on artificial hybridization edonian wild taro, Colocasia esculenta. Pac. Sci. 53: 273–285.
has not yet been studied, the breeders should take it into Ivancic, A., and Lebot, V. 2000. The genetics and breeding of taro.
consideration when planning and carrying out crosses and Séries repères; Centre de coopération internationale en recher-
self-pollinations. Thermogenically active (and intensively che agronomique pour le développement (CIRAD), Montpellier.
odorous) inflorescences, which are not protected from polli- Knutson, R.M. 1974. Heat production and temperature regulation
nating insects, should be avoided, because the insects with in eastern skunk cabbage. Science (Washington, D.C.), 186:
pollen from other inflorescences may already be inside the 746–747.
floral chamber. The most suitable time for hybridization Lamprecht, I., Schmolz, E., Blanco, L., and Romero, C.M. 2002a.
(emasculation and pollination of inflorescences used as fe- Flower ovens: thermal investigations on heat producing plants.
male components) is 2 d (two mornings) before pollen shed Thermochim. Acta, 391(1,2): 107–118.
(23–30 h before they would have become thermogenically Lamprecht, I., Schmolz, E., Hilsberg, E., and Schlegel, S. 2002b. A
the most active and intensively odorous). This stage, which tropical water lily with strong thermogenic behaviour – thermo-
can be recognized by the change of the color of the upper metric and thermographic investigations on Victoria cruziana.
part of the spathe, is also the most suitable time for the pro- Thermochim. Acta, 391(1,2): 199–210.
tection of the male components. The hybridization in such Leick, E. 1915. Die Erwärmungstypen der Araceen und ihre
an early stage has been practiced by most of the breeders blütenbiologische Deutung. Ber. Dtsch. Bot. Ges. 33: 518–536.
(Ivancic and Lebot 2000), although no systematic data about Matthews, P.J. 1990. The origins, dispersal and domestication of
thermogenesis have been available. taro. Ph.D. thesis, Australian National University, Canberra.
Meeuse, B.J.D. 1975. Thermogenic respiration in aroids. Annu.
In further studies, it would be useful to determine the rela- Rev. Plant Physiol. 26: 117–126.
tionship between the expression of the thermogenic activity Miyake, T., and Yafuso, M. 2003. Floral scents affect reproductive
and spadix dimensions. It may also be useful to identify var- success in Fly-pollinated Alocasia odora (Araceae). Am. J. Bot.
ious volatile compounds responsible for odor. As the maxi- 90: 370–376.
mum temperature differences between an inflorescence and Nagy, K.A., Odell, D.K., and Seymour, R.S. 1972. Temperature
the ambient air probably increase with the spadix length regulation by the inflorescence of Philodendron. Science (Wash-
(Gibernau and Barabé 2002), it would be good to include ington, D.C.), 178: 1195–1197.
genotypes of which spadices are more than 20 cm long. So Okpul, T., and Ivancic, A. 1995. Study of stigma receptivity in
far, the longest spadices were recorded within natural popu- intraspecific hybridization of taro, Colocasia esculenta. Paper
lations of the “green morphotype” in the Nimbaye Valley (in presented at the Taro Seminar sponsored by the South Pacific
the Northeast Coast of the Grande Terre, New Caledonia) Community, Department of Agriculture and Livestock (Papua
and described by Ivancic and Lebot (1999). Their maximum New Guinea Government) and University of Technology (Lae,
lengths were as follows: 44.20 cm (spathe), 28.50 cm Papua New Guinea), 26–30 June 1995, Lae, Papua New Guinea.
(spadix), 6.10 cm (female portion), 8.30 cm (male portion), Patino, S., Grace, J., and Banziger, H. 2000. Endothermy by flow-
and 11.30 cm (sterile appendix). It would also be useful to ers of Rhizanthes lowii (Rafflesiaceae). Oecologia, 124: 149–
evaluate the influence of the climate (to evaluate the therm- 155.
ogenic activity during the cool period of the year and (or) Seymour, R.S. 1997. Plants that warm themselves. Sci. Am. 276:
outside the tropics). 90–95.

© 2004 NRC Canada

Ivancic et al. 1565

Seymour, R. 1999. Pattern of respiration by intact inflorescences of Shaw, D.E. 1975. Illustrated notes on flowering, flowers, seed and
the thermogenic arum lily Philodendron selloum. J. Exp. Bot. germination in taro Colocasia esculenta. Papua New Guinea
50: 845–852. Dep. Agric. Stock Fish. Res. Bull. 13: 39–59.
Seymour, R.S., and Blaylock, A.J. 1999. Switching off the heater: Skubatz, H., Nelson, T.A., Meeuse, B.J.D., and Bendich, A.J.
influence of ambient temperature on thermoregulation by east- 1991. Heat production in the voodoo lily (Sauromatum
ern skunk cabbage Symplocarpus foetidus. J. Exp. Bot. 50: guttatum) as monitored by infrared thermography. Plant Physiol.
1525–1532. 95: 1084–1088.
Seymour, R.S., and Schultze-Motel, P. 1996. Thermoregulating lo- Taro Network for Southeast Asia and Oceania (TANSAO). 2002.
tus flowers. Nature (London), 383: 305. Final report (covering period from January 1998 to December
Seymour, R.S., and Schultze-Motel, P. 1997. Heat-producing flow- 2001). Centre de coopération internationale en recherche agro-
ers. Endeavour, 21: 125–129. nomique pour le développement (CIRAD), Montpellier, France.
Seymour, R.S., and Schultze-Motel, P. 1999. Respiration, tempera- Thien, L.B., Azuma, H., and Kawano, S. 2000. New perspectives
ture regulation and energetics of thermogenic inflorescences of on the pollination biology of basal angiosperms. Int. J. Plant.
the dragon liliy Dracunculus vulgaris (Araceae). Proc. R. Soc. Sci. 161(Suppl. 6): S225-S235.
London B, Biol. Sci. 266/1432: 1975–1983. Yafuso, M. 1993. Thermogenesis of Alocasia odora (Araceae) and
Seymour, R.S., Schultze-Motel, P., and Lamprecht, I. 1998. Heat the role of Colocasiomyia Flies (Diptera: Drosophilidae) as
production by sacred lotus flowers depends on ambient tempera- cross-pollinators. Pop. Ecol. 22/3: 601–606.
ture, not light cycle. J. Exp. Bot. 49: 1213–1217. Young, R.A. 1924. Taros and yautias. Promising new food plants
Seymour, R.S., Gibernau, M., and Ito, K. 2003. Thermogenesis and for the South. US Dept. Agric. Bull. 1247.
respiration of the dead horse arum Helicodiceros muscivorus, a
pseudo-thermoregulatory aroid associated with fly pollination.
Funct. Ecol. 17: 886–894.

© 2004 NRC Canada

View publication stats