M. GIBERNAU,]. ALBRE, 2008 101 Size Variations of Flowering Characters in Arum italicum (Araceae) Marc Gibernau· and Jerome Albre Universite Paul Sabatier Laboratoire d'Evolution & Diversite Biologique (UMR 5174) Bat.4R3-B2 31062 Toulouse cedex 9 France *e-mail: [email protected] ABSTRACT INTRODUCTION In Arum, bigger individuals should An extreme form of flowering character proportionally invest more in the female variations according to the size is gender function (number or weight of female modification, which occurs in several flowers) than the male. The aim of this species of Arisaema (Clay, 1993). Individ­ paper is to quantify variations in repro­ ual plant gender changes from pure male, ductive characters (size of the spadix when small, to monoecious (A. dracon­ parts, number of inflorescences) in rela­ tium) or pure female (A. ringens) when tion to plant and inflorescence sizes. The large (Gusman & Gusman, 2003). This appendix represents 44% of the spadix gender change is reversible, damaged length. The female zone length represents female individuals will flower as male the 16.5% of the spadix length and is much following year (Lovett Doust & Cavers, longer than the male zone (6%). Moreover 1982). These changes are related to change these three spadix zones increase with in plant size and are explained by the plant vigour indicating an increasing size-advantage model. The size-advantage investment into reproduction and pollina­ model postulates a sex change when an tor attraction. It appears that the length of increase in body size is related to differen­ appendix increased proportionally more tial abilities to produce or sire offspring than the lengths of the fertile zones. On (Policansky, 1981). For entomophilous average an inflorescence counts 156 male plant species, as Aroids, the Size-advantage flowers and 61 female flowers which model predicts a female-biased gender result in a male-biased floral ratio in A. expression with increased size (Klinkha­ italicum. mer et at., 1997). The numbers of male and female flowers In Arum species, inflorescences are increased significantly with the spadix size monoecious (e.g. hermaphrodite) bearing but differently according to the gender, the both male and female flowers and such number of female flowers increasing faster dramatic gender change does not exist. But than male: on the other hand this effect was the size-advantage model should also marginally significantly (p = .08). This apply and thus bigger plants should pro­ relative gender difference of flower num­ duce more female-biased inflorescences. ber increase is visualised by a significant This should be represented in Arum by a decrease of the maleness floral ratio with proportionally bigger increase of the fe­ spadix size. male function (number or weight of female flowers) than for male flowers. Interestingly the only study on Arum KEYWORDS related to this topic gave contradictory Flower numbers, resource allocation, sex results (Mendez, 1998). In Spanish popula­ ratio, size-advantage model. tions of Arum italicum, the number and 102 AROIDEANA, Vol. 31 size of inflorescences increase with plant width of leaves (at the level of the petiole). size, estimated by the dry weight, as well as Leaves were approximated as a triangle the total of flowers (Mendez, 1998). The with respect to their surface. For some of numbers of male & female flowers were these inflorescences, the number of male positively correlated to the inflorescence and female flowers per inflorescence (n = size but with the same relation (Le. slope). 67) were also counted. The floral sex ratio Finally the floral sex ratio (66-72% of male (number male flowers/(number of male + flower) was not correlated to the plant size female flowers» was calculated for some of or the number or size of inflorescences these inflorescences (n = 64). The floral (Mendez, 1998). Thus the reproductive sex ratio varies from 0 (pure female) to 1 effort is related to the plant vigour but with (pure male). no differences among gender and thus no Data analyses consisting of linear regres­ female bias in bigger inflorescences. A sion analysis and slopes were compared complementary study showed that in terms with an ANCOVA analysis. Mean size of biomass, male and female flowers each differences between inflorescences were represented only 6% of the spadix biomass tested with a t-test. A distribution test to but when considering a range of different assess if the floral sex ratio was non-biased flower biomasses there was a dispropor­ (equal to .5) with a Kolmogorov-Smirnov tionate allocation to the male part versus one sample test using a normal distribution female (Mendez, 2001), which is a result (mean: .50, standard deviation: 1). All these contrary to the size-advantage model. In analyses were performed using Systat 11 Arum the pattern is expected to be (2004) software. complicated because an individual can produce several inflorescences and in fact RESULTS the last-produced (and smaller) inflores­ cence showed a female-biased floral ratio Morphology and plant vigour compared to the first and larger one The mean sizes of the different parts of (Mendez, 1998), which is a result in the inflorescences are shown in Table l. accordance to the size-advantage model. The appendix appears to be an important The aim of this paper is to characterize part of the inflorescence as its length the reproductive characters of Arum itali­ represented 43.1 ± 8.2% of the total spadix cum in the south of France and to quantify size (66% when considering only the fertile variations in reproductive characters (size zones). The female zone is about 2.5 times of the spadix parts, number of inflores­ longer than the male zone. cences) in relation to plant and inflores­ A positive relationship was found be­ cence sizes. Do more vigorous plants tween leaf size and spathe length (R2 = .33, produce more and bigger inflorescences? F1.67 = 33.1, P < 10-5). Moreover, the Do bigger inflorescences bear more flow­ mean appendix size increased with spathe ers? Does the floral sex ratio change size (R2 = .24, F .62 = 20.27, P < 10-5). For according to the inflorescence size? 1 plants producing two inflorescences, the size of the second appendix was smaller MATERIAlS AND METHODS than that of the first (t = 2.28, df = 29, P = During the spring of 2001, 90 inflores­ .03). But there was no difference in spadix cences of Arum italicum Miller were sizes between the inflorescence of plants harvested in order to measure the spathe flowering once and the first inflorescence length and different parts of the spadix, Le. of plants flowering several times (t = .26, df fertile male and female zones and appendix = 33.3, p = .79). Finally an inflorescence length. Plant vigour was estimated by produced an average of 155.6 ± 42.7 male counting the number of inflorescences flowers and 60.8 ± 19.2 female flowers. produced and by measuring the height of Thus as an overall, the floral ratio in A. the longest petiole, and the limb length and italicum was male biased (.72 ± .04) and M. GIBERNAU, J. ALBRE, 2008 103 Table 1. Reproductive morphological measures (mean ± SD) in01;Uimetres of Arum italicum inflorescences in south-western France(n = 90). The sum of the lengths of the appendix, the male zone and the female zone are not equal t9. the spadix length, as the appendix stipe and the two sterile zones ..were not measured. Spadix Appendix length Appendix diameter Male zone Female zone 101.7 ± 29.7 44.8 ± 17.2 10.6 ± 3.4 6.4 ± 1.5 16.8 ± 5.4 statistically different from .5 (p < 10-5). In both the fertile male (slope = .03) and average, male flowers were 2.5 times more female (slope = .15) zones increased with numerous than female flowers. the spadix size, with the female zone increasing proportionally more than the Intra-plant relationships male zone (Fig. 1). The number of male and female flowers Both the sizes of the appendix and of the increased significantly (respectively: R2 = two fertile zones were positively correlated .28, F1,62 = 19.73, P = 4 X 10-5; R2 = .39, with the spadix size, which is logical due to F1,60 = 37.68, P < 10-5) with the spadix the autocorrelation of these traits. But size (Fig. 2). The two slopes on log­ interestingly, when comparing the slopes, transformed data were marginally signifi­ it appears that the appendix (slope = .51) cantly different (F1,124 = 2.28, P = .08), the increased proportionally more than the number of female flowers (slope = .76) fertile zones (Fig. 1). In the same way, increasing faster than for male flowers 100 90 80 -E 70 • • E • • Appendix 60 -.c • [J Female C) 50 • -c • • Male .!! 40 CDc • 0 30 N ' .....Ii.' · [J 20 10 i=: i:1i~ e~~:. ~ :~ o+-----~----,_----,_----,_----,_----~----~: : 50 70 90 110 130 150 170 190 Spadix length (mm) Fig. 1. Relationships between the lengths of the spadix and the appendix, the male flower zone, and the female flower zone. Each of the measured parts increases with spadix length (appendix: R2 = .81, F1,87 = 373.85, P < 10-5; male zone: R2 = .48, F1,87 = 83.95, P < 10-5; female zone: R2 = .73, F1,87 = 226,39, P < 10-5). But the slope for the appendix regression is higher than for any of the fertile zones (F1,175 = 75.50, P < 10-5). In the same way, the fertile female zone increases more than the fertile male zone (Fl;175 = 69.45, P < 10-5). 104 AROIDEANA, Vol. 31 250 : • • .. 200 . '"~ •• 0 150 ;;::: • male 0 -.. 0 o female CD Jl 100 E ::;, 0 c 50 0 0 0 50 100 150 200 spadix length (mm) Fig. 2.