M. GIBERNAU, M. CHOUTEAU, K. LAVALLEE, D. BARABE, 2010 183 Notes on the Phenology, Morphometry and Floral Biology of Anaphyllopsis americana ()

Marc Gibernau CNRS-UMR 8172 Ecologie des Forets de Guyane BP 709, 97387 Kourou~ France Mathieu Chouteau Departement des Sciences Biologiques Universite de Montreal C.P. 6128, Succ. Center-Ville Montreal, Canada, H3C 3}7 Karlne Lavallee Institut de Recherche en Biologie Vegetale Universite de Montreal Jardin Botanique de Montreal 4101 rue Sherbrooke Est Montreal (Quebec), Canada HIX 2B2 Denis Barabe Institut de Recherche en Biologie Vegetale Universite de Montreal Jardin Botanique de Montreal 4101 rue Sherbrooke Est Montreal (Quebec), Canada HIX 2B2

ABSTRACf that the species cannot self-pollinate and thus relies entirely on external vectors for Anapbyllopsis americana (Engler) A. pollination. Despite extensive observation, Hay was studied in . It we were unable to identify day-time possesses one of the longest floral cycles pollinators. Due to the low frequency and in Araceae, about 40 d, with each sexual abundance of insects present within the phase lasting about 3 wk. In association floral chambers, the fact that these insects with such a flowering cycle, the thermo­ (ants, crickets, cockroaches, ... ) are not genic pattern is Similarly long, and consists very efficient pollinators, and the low fruit of one temperature peak of low intensity set, we suggest that the pollination system (1.5-6°C) occurring every day between in A. americana is apparently inefficient. 1000 and 1500 hours. The presence of a Finally, a long flowering cycle plus flower­ daily thermogenic pattern over a long ing periods all year long may be a flowering period of time (14-30 d) has not been strategy in A. americana or the result of an observed to date in other Aroids. inefficient pollination system achieved by The phenology and reproductive success inefficient pollinating insects in order to of the studied population appear to vary increase the probability of cross-pollina­ seasonally. In fact, the proportion of tion. flowering individuals and reproductive success are higher after the dry season INTRODUcnON when rains resume (November). Regard­ less of the season, reproductive success The Anapbyllopsis, which now remains low, with an overall mean of 17% comprises three species, was first created maturing infructescences. We have shown to accommodate ameri- 184 AROIDEANA, Vol. 33

canum Engler, the only Neotropical spe­ few of them known as efficient pollinators, cies of the Asian genus Cyrtosperma (Hay, can be found in low frequencies in the 1988; Gon~alves, 2005). Anapbyllopsis be­ . longs to the subfamily (Engler) Although some floral characters of A. which is composed of 10 genera: Dracon­ americana are well known, phenology tium 1., Dracontioides Engler, Anapbyllop­ and thermogenesis need to be document­ sis A. Hay, Pycnospatba Gagnepain, Ana­ ed. In the present paper we describe the pbyllum Schott, Cyrtosperma Griffith, reproductive and flowering cycles, and Lasimorpba Schott, N.E. Brown, analyze the thermogenic pattern in one Loureiro, Urospatba Schott (Hay, population growing in French Guiana. 1992; Mayo et al., 1997). Some floral characters of Anapbyllopsis MATERIAL AND MEmODS americana (Engler) A. Hay have been studied in different comparative works on Anapbyllopsis americana (Engler) A. the Araceae family (Chouteau et al., 2006, Hay was studied in French Guiana dur­ 2008; Barabe & Lacroix, 2008; Barabe et al., ing five different periods from May-Sep­ 2008; Gibernau et at., 2010). In summary, tember 2003, as well as in July 2005, May Anapbyllopsis americana appears to have 2006, July 2008 and November 2009. One a relatively small pollen grain which population was sampled and observed at possesses a long viability since 50% of the Kilometer Point 19 along the road leading pollen is viable after 70 hr (3 d), and 210 hr to Petit Saut dam (Kourou region). Ana­ (9 d) are necessary for a total loss of pbyllopsis individuals were growing in viability (Barabe et al., 2008). Its Pollen! seasonally inundated forest gaps near a Ovule (PO) ratio is very high (about marsh. A voucher specimen: Barabe et al. 100,000:1) suggesting a non-efficient polli­ 258 was deposited at the Marie-Victorin nation system which might be linked with a Herbarium (MT). long flowering cycle, for an Araceae, of A phenological survey in terms of about 3-4 wk. No pollinators have been number of leaves and inflorescences per documented in this species or even for the individual was performed on 88 individu­ genus (Gibernau, 2003; Chouteau et al., als. 2006, 2008). Multivariate analyzes per­ Vegetative morphometric measures Cleaf formed on Araceae in order to study the size, spathe height and spadix length) were relationships between the type of pollina­ performed on 30 inflorescences. Reproduc­ tors (bee, fly and beetles) and floral traits tive morphometric characters (height and have shown that Anapbyllopsis americana width of the spathe, spadix length, flower is distinct from the three main pollination number, number of ovules and seeds, "syndromes" (Chouteau et al., 2008; Gi­ aborted ovules and flowers) were mea­ bernau et al., 2010). The floral characters of sured on 16 inflorescences and 17 infruc­ A. americana appear to cluster with those tescences. of two other Lasiodeae species (Dracon­ Experiments on spontaneous self-polli­ tium polypbyllum 1. and Dracontioides nation, consisting of bagging inflorescenc­ desciscens (Schott) Engl.) and those of es from before opening until complete two species from the Orontioideae sub­ drying or maturation, were performed on family (Lysicbiton camtscbatcensis Schott 7 inflorescences from different individuals. and Symplocarpus renifolius Schott) to thermogenesis was record­ form a putative group of species with a ed on 11 inflorescences in 2003, 2 inflores­ generalist pollination system. Generalist cences in 2005 and 4 inflorescences in pollination systems are rare in Araceae 2006. Temperatures of the spadix and and only documented in two species, ambient air were recorded every 10 min Lysicbiton camtscbatcensis (Tanaka, 2004) using two Digi-Sense® DualLogR® thermo­ and Symplocarpus renifolius (Uemera et couple thermometers. One of the ther­ al., 1993). In these cases, various insects, a mometer's probes was inserted approxi- M. GIBERNAU, M. CHOUTEAU, K. LAVALLEE, D. BARABE, 2010 185 mately 5 mm into the middle of the spadix phase, the temperature differences be­ and the second probe was used to record tween the air and the spadix were not ambient air temperature. The temperatures constant. In some individuals, they were of the leaf base and the peduncle were also equivalent to the temperature differences recorded on 6 and 4 specimens respective­ occurring during the female phase (Fig. 1), ly. For the leaf, the probe was inserted in in others, they were lower and correspond­ the petiole 20 cm above ground level, and ed to a "non-heating" spadix (Fig. 1). for the peduncle, 5 cm below the base of However, in still other individuals, the the spathe. temperature increase during the male phase was higher than in the female phase RESULTS (Fig. 1). After that, the temperature in­ crease was weak, varying between 1 and lSc (Fig. 1). In 2005 and 2006, the Floral cycle-The floral cycle started temperature records between the spadix with the opening of the spathe and stigmas and the leaf petiole suggest that the appeared to be receptive at this time. temperature difference is high during the Interestingly, the spathe started to open female and very low, virtually inSignificant, from the apex and formed a funnel during the male phase (Fig. 2). entrance but also separated at the base leaving a direct access to the spadix (about Insect visitors-In July 2006, the floral 0.5-1 cm wide). The floral cycle lasted chambers of 16 open inflorescences were about 40 d (range: 34-48 d), starting with observed between 0900-1100 hours. Only the female phase (stigmas receptive) which 5 inflorescences had visitors, respectively: lasted between 14 and 22 d followed by the two and five ants (Hymenoptera), one male phase which lasted between 17 and cockroach (Blattodea), one woodlouse 24 d. The anthesis occurred from the top to Osopoda) and two crickets (Orthoptera). the base of the inflorescence. Our obser­ In July 2008, between 1600-1800, with 8 vations did not document the existence of a open inflorescences, only one cricket bisexual intermediate stage. (Orthoptera) was observed. In November 2009, between 0830-1030, with 11 open Thermogenesis-The long thermogenic inflorescences at female stage, only one pattern (Fig. 1) seemed to be linked with had insects: one black stingless bee (Hy­ the long flowering cycle and consisted of a menoptera) and one black grasshopper single daily temperature increase of low (Orthoptera) . intensity occurring between 1000 and 1500 hours. The temperature difference between Phenological Frequencies-The studied the spadix and ambient air varied between population contained more than one hun­ 1.5 and 6°C, but more commonly between dred individuals (Table 1). In July 2006, 2 and 3°C (Fig. 1). The spadix heating was 114 individuals were counted, and half confirmed when looking at the difference (51%) produced at least one inflorescence. between the spadix and the leaf petiole, a Among reproductive individuals, a quarter non-thermogenic tissue (Fig. 2). The spa­ (26%) produced 2 or 3 inflorescences. The dix temperature was between 3 and 5°C 58 reproductive individuals produced 76 warmer than the petiole during the day. inflorescences, 23 of them during the The temperature increase was very present flowering period. Hence, 53 inflo­ feeble before the spathe opened, less than rescences were issued from the previous 1. 5°C, certainly due to solar radiation blooming period, 46 were dried and 7 (Fig. 1). During the female phase, the maturing as infructescences. In July 2008, amplitude of the pattern was very irregular. 50 individuals were marked, and half (54%) The values of the temperature difference produced one inflorescence. Of the 28 peaks varied between 1 and 6.5°c among inflorescences observed, 12 were from the and within an individual. During the male present flowering period and 16 from the 186 AROIDEANA, Vol. 33

4 .!S Female phase Male phase ia. III 3.5 cu .c c -cu 3 cu ~p" .a - 2.5 .." .!:: cu !II 1.1- e:: C 2 f .!! cu.a ."= E!II 1,5 ..CU" C ;:, !II ~ 1 cu a. E ...cu 0,5 C cuIII :! 0 Days

4,50 Closed Female pbase Male phase 4,00 Soathe

, oE ~ 3,50 "

3,00

2,50

2,00

1.50

1,00

0.50 l 1 • . l I l.ll .1 0,00 ! • Days Fig. 1. Mean temperature differences between the spadix and ambient air recorded during about 60 d on 4 individuals (upper graph) and during 30 d on 3 individuals (lower graph). Each peak of temperature difference represents a different day (around midday). previous one, of which 14 were dried and 2 ground organ. Marking in July 200S maturing as infructescences. In November allowed us to estimate the class "0 leaf - 0 2009, we observed 71 individuals. Among inflorescence" in 2009. It is not known them only 2S% of the individuals (N = 14) whether the non-recovered individuals marked the previous year had any above- were dormant or dead. Finally, 75% of the M. GlBERl"lAU, M. CHOUTEAU, K. LAVALLEE , D. BARABE, 2010 187

Female phase Male phallic

~(------)~ ~(~------~) zoos

...... individual 1

- indi idual 2

0 <:> <:> <:> <:> <:> <:> <:> 0 <:> 0 <:> 0 <:> <:> <:> <:> 0 <:> <:> <:> <:> <:> 0 0 0 0 0 .....N 0 ,...,N 0 N,..., 0 N,..., 0 N.... 0 ....N 0 Time (Hours)

Female phase Male phase 6 ~(------~) ~(~------~> 2006

5

4 ...... individual 1 3 - individual 2

2

1

0 <:> <:> 0 0 0 <:> 0 0 <:> 0 <:> 0 <:> 0 0 <:> <:> <:> 0 0 <:> ~ <:> <:> <:> <:> <:> N 0 N 0 N 0 N 0 N 0 N 0 N 0 ~ ~ ... ~ ...... ~ Time (Hours) Fig. 2. Mean temperature differences between the spadL,( and the petiole recorded during 7 days on two individuals in 2005 (upper graph) and in 2006 (lower graph) during female and male phases. plants (N = 43) produced at least one and 5 maturing as infructescences. In inflorescence and o nl y 12% of the repro­ conclusion, in November 2009, the popu­ ductive individuals had two or three lation fl owering pattern was different from 2 inflorescences. The 43 reproducti ve indi­ July 2006 or 2008 (Chi 2 = 9.96, P = .007). viduals produced 49 inflorescences, includ­ In July (2006 and 08), half of the population ing 36 during the present fl owering peri od. was composed of reproducti ve individuals Only 13 inflorescences were issued from (no difference between 2006 and 2008: the previous blooming pe ri od, 8 were dried Chi 2] = .07, P = .79) whereas in November 188 AROIDEANA, Vol. 33

Table 1. Phenological stage of the population "PK19" in three different years and at two different seasons. The plain line in the table separates the non reproductive individual from the reproductive ones.

Frequency (observations) Number of Number of July 2006 July 2008 November 09 leaves inflorescence (N = 114) (N= 50) (N= 71) 0 0 N.A. N.A. 14 1 0 32 19 12 2 0 21 4 2 3 0 3 - - 0 1 - 3 14 1 1 29 18 18 2 1 12 4 6 3 1 2 2 - 1 2 9 - 4 2 2 3 - - 2-3 3 3 - 1 *N.A. Not Available

2009, 75% of the population was reproduc­ relationship between spathe height and the tive. Considering only the previous flower­ number of flowers . The same held true ing period, the percentages of developing between spadix height and the number of infructescences in the population were the flowers (Y = 26.306. X + 32.409; R2 = .76; same (13%) inJuly 2006 and 2008 (Chi21 = F1•14 = 43.97 ; P < 10-3). .05, P = .94) but higher (38%) in November 2009 (Chi21 = 4.47, P = .034). Reproductive success--Reproductive suc­ cess was low. The rate of non-maturing Morphometry-The leaf (N = 30) was inflorescences was high, estimated at be­ 58.9 ± 14.6 cm long (range: 28-90) and tween 60 and 900Al (N = 82); the number of 42.9 ± 10.7 cm wide (range: 16--65). There maturing seeds per infructescence was low, was a significant positive linear relationship since only 52.3 ± 31.7 berries (N = 25, range: 8-103) matured per infructescence, repre­ between leaf length and spadix height (Y = senting only one third of the total number of .285. X + 7.725) (R2 = .71; F1•32 = 79.1 ; P < 10-3). The spathe was on average 24.4 ± flowers per inflorescence. Thus, on average, two thirds of flowers were not pollinated or 4.3 cm long (N = 58, range: 12-34) and the spadix 4.12 ± .88 cm high (N = 42, range: did not develop fruit. Moreover, there was no linear relationship between the size of the 2.5-6.7). The positive relationship between inflorescence (in term of flower number) and spathe and spadix heights (Y = .188. X - the number of maturing seeds (R2 = .007) . .606) was significant (R2 = .61; F132 = -3 . Finally, none of the 7 bagged inflorescences 50.40; P < 10 ). set fruits, indicating that spontaneous self­ The inflorescence bears bisexual flow­ pollination might be rare. ers; each composed of 4 stamens surround­ ing the unilocular ovary containing one DISCUSSION ovule. The average number of flowers per inflorescence was 149 ± 33 (N = 32, range: The lasioid species, Anapbyllopsis amer­ 79-244). However, we found a positive icana (Engler) A. Hay, possesses one of the (Y = 5.636 • X + 5.721) and significant longest floral cycles in Araceae, about 40 d. (R2 = .47; F1.14 = 12.63 ; P = .003) linear With each sexual phase lasting about 3 wk, M. GIBERNAU, M. CHOUTEAU, K. LAVALLEE, D. BARABE, 2010 189 such results are comparable to some long­ of the population was reproductive, versus flowering Anthurium species (Croat, 75% in November, at the beginning of the 1980). Contrary to many Anthurium spe­ rainy season. Similarly, reproductive suc­ cies (Croat, 1980), there is no efficient cess was three times higher in November bisexual intermediate stage in A. america­ than in July. However, further studies are na since none of the bagged inflorescences needed to address this question in more fructified. detail. The thermogenic pattern is also long Whatever the season, reproductive suc­ (Fig. 1) and seems to be linked with the cess remained low with an overall mean of unusually long flowering cycle. It consists 17% maturing infructescences. Since the of one temperature peak of low intensity species cannot self-pollinate, it relies en­ occurring every day between 1000 and tirely on' external vectors for pollination, 1500 hours. Although the thermogenic and apparently on insects. Based on pattern appears to be periodic, the ampli­ herbarium specimens, the species appears tude of the daily thermogenic peak varies to flower all year long (Hoff et al., 2007). considerably between and within an indi­ This may be a flowering strategy associated vidual. Hence the pattern of thermogenesis with a non-efficient pollination system. is not well defined as is the case in other Multiplying flowering events at different Aroids, as for example in Monstera (Chou­ periods throughout the year may increase teau et al., 2009), Syngonium (Chouteau et the probability of cross-pollination by al., 2007) or Philodendron (Barabe et al., inefficient pollinating insects. 2002). This might be due to extrinsic Although we have made extensive ob­ (weather) or intrinsic (physiology, mor­ servations, we have no precise data regard­ phology) factors or both. Even if the spadix ing pollinators. From the low frequency is protected at the bottom of an above­ and abundance of insects within the floral open floral chamber, its small size (about chambers, the fact that these insects (ants, 4 cm high and less than 1 cm width) may crickets, cockroaches, ... ) are not very prevent it from being highly thermogenic, efficient pollinators, and the low fruit set, because of the high surface - volume ratio. we suggest that the pollination system in A. Applying heat transfer models (Gibernau et americana appears to be rather inefficient. al., 2005), temperature differences in the One could simply hypothesize that the long small spadix of A. americana should range flowering cycle increases the probability of between 1.8 and 6°C, which is in accor­ visits to an inflorescence by inefficient dance with our measures. However, we do pollinating insects. Such pollination sys­ not possess enough data to clarify this tems have been already studied in other question. The presence of a daily thermo­ plants, and cockroaches in particular have genic pattern over a long period of time been reported as pollinators in the Asian (14-30 d) has not been observed to date in Annonaceae, Ulvaria elmeri (Nagamitsu & other Aroids (Gibernau et al., 2005). Inoue, 1997) and in one French-Guianan However, we do not know whether this Clusia (Vlasakova et al., 2008). On one particular thermogenic pattern plays a role hand, observations made in these previous in the emission of odors or in pollination studies can explain some of our observa­ mechanisms. tions on A. americana. The flower visits The phenology and reproductive success occurred at night. Flowers were visited by a of the studied population appeared to vary variety of insects (ants, crickets, bugs). The seasonally. One can expect a weather cockroaches fed on stigmatic exudates and effect, particularly with regard to the pollen, and the visit frequency was low impact of rain on pollination success. In (less than 2 per hour). On the other hand, fact, the proportion of flowering individu­ these previous studies present some differ­ als and reproductive success were higher ences with regard to our observations. In after the dry season when rains resumed. In Clusia, cockroaches were efficient pollina­ July, at the beginning of the dry season, half tors leading to a fruit production of 42% 190 AROIDEANA, Vol. 33

(Vlasakova et al., 2008), whereas in Uvaria, --- & c. Lacroix. 2008. Developmental cockroaches had a low pollen load and morphology of the flower of Anaphyl­ poor pollen efficiency leading to a fruit set lopsis americana and its relevance to of 2%, (Nagamitsu & Inoue, 1997). By our understanding of basal Araceae. contrast with A. americana, the flowering Botany 86: 1467-1473. cycle of Uvaria is short since a flower ---, K. Lavallee & M. Gibernau. 2008. bloomed for 50 hr (Nagamitsu & Inoue, Pollen viability and germination in 1997). Finally, the flowers of Clusia ap­ some Neotropical Aroids. Botany 86: peared to be highly specialized to cock­ 98-102. roach visits with secretory organs produc­ Boos, J. O. 1997. Observations on New ing an attractive non-sugary liquid World Araceae - Lasieae. Aroideana secretion and a buttery floral odor rich in 20: 13-26. acetoin (Vlasakova et al., 2008). Hence Chouteau, M., D. Barabe & M. Gibernau. night pollination should be studied more 2006. A comparative study of inflores­ carefully, since some of our observations cence characters and Pollen-ovule are concordant with such a hypothetical ratios among the genera Philoden­ pollinator. dron and Anthurium (Araceae). Int.]. The flowering biology of the other Sci. 167(4): 817-829. lasioid genera is not known (Gibernau, ---, --- & ---. 2007. Thermo­ 2003). Only a few mentions of visiting genesis in Syngonium (Araceae). Can. insects are available. In and ]. Bot. 85: 184-190. Dracontioides, the inflorescences are visit­ ---, --- & ---. 2009. Flowering ed by flies (Boos, 1997; Zhu & Croat, 2004; and thermogenetic cycles in two Gons;alves, 2005; Grayum, pers. observ.). species of Monstera (Araceae). Bull. Nitidulidae (Coleoptera) are reported to Soc. Hist. Nat. Toulouse (In press). visit inflorescences of the African Cyrto­ ---, M. Gibernau & D. Barabe. 2008. sperma or the Neotropical Relationships between floral charac­ (Knecht, 1983; Boos, pers. observ.). Inter­ ters, pollination mechanisms, life estingly, Urospatha has a habit and a floral forms and habitats in Araceae. Bot. j. morphology very close to Anaphyllopsis, Linn. Soc. 156: 29-42. but unfortunately no more precise infor­ Croat, T. B. 1980. Flowering behavior of mation is available. the neotropical genus Anthurium (Araceae). Amer. ]. Bot. 67(6): 888- ACKNOWLEDGMENTS 904. Gibernau, M. 2003. Pollinators and visitors We are grateful to the Staff of Labor­ of Aroid inflorescences. Aroideana 26: atoire Environnemental de Petit Saut (Hy­ 66-83. dreco), Kourou, French Guiana, for logis­ ---, D. Barabe, M. Moisson & A. tical help. This research was supported in Trombe. 2005. Physical constraints part by an individual operating grant from on temperature difference in some the Natural Sciences and Engineering Re­ thermogenic aroid inflorescences. search Council of Canada to D.B and by Ann. Bot. 96: 117-125. the "Project Amazonia" from the CNRS­ ---, M. Chartier & D. Barabe. 2010. Guyane. Recen! advances towards an evolu­ tionary comprehension of Araceae LITERATURE CITED pollination. Fourth International Con­ Barabe, D., M. Gibernau & F. Forest. 2002. ference on the Comparative Biology of Zonal thermogenetic dynamics of two the Proceedings (In Philodendron species from two differ­ press). ent subgenera (Araceae). Bot.]. Linn. Gons;alves, E. G. 2005. A revision of the Soc. 139(1): 79-86. genus Dracontioides Eng!. (Araceae), M. GIBERNAU, M. CHOUTEAU, K. LAVALLEE, D. BARABE, 2010 191

including a new species from Bahia, Uvaria elmeri (Annonaceae) in a . Aroideana 28: 21-31. lowland mixed-dipterocarp forest in Hay, A. 1989 ("1988"). Anaphyllopsis: A Sarawak. Amer j. Bot. 84(2): 208-213. New Neotropical Genus of Araceae­ Tanaka, H. 2004. Reproductive Biology of Lasieae. Aroideana 1(1): 25-31. Lysichiton camtschatcense (Araceae) ---. 1992. Tribal and subtribal delimi­ in Japan. Aroideana 27: 167-171. tation and circumscription of the Uemera, S., K. Ohkawara, G. Kudo, N. genera of Araceae tribe Lasieae. Wada & S. Higashi. 1993. Heat-pro­ Ann. Missouri Bot. Card. 79(1): duction and cross-pollination of the 184-205. Asian skunk cabbage Symplocarpus Hoff, M., G. Cremers, H. Chevillotte, J.-J. de renifolius (Araceae). Amer. j. Bot. 80: Granville, V. Guerin & J.-F. Molino. 635-640. 2007. Base de donnees botaniques Vlasakova, B., B. Kalinova, M. H. G. Aublet2 de I'Herbier de Cuyane fran­ Gustafsson & T. Holger. 2008. Cock­ ~aise (CAY). http://www.cayenne.ird. roaches as pollinators of Clusia aff. fr/aublet2 sellowiana (Clusiaceae) on inselbergs Mayo, S. J., J. Bogner & P. C. Boyce. 1997. in French Guiana. Ann. Bot. 102(3): The genera of Araceae. Kew, The 295-304. Trustees, Royal Botanic Gardens. Zhu, G. & T. B. Croat. 2004. Revision of Nagamitsu, T. & T. Inoue. 1997. Cockroach Dracontium (Araceae). Ann. Missouri pollination and breeding system of Bot. Card. 91: 593-667.