Bombax Ceiba </Emphasis>
29:1-
FLORAL STRUCTURE AND STAMENS IN BO2~IBAX CE[BA L.
BY T.A. Daws [,Ldian Statistical Institute, (,,'a[cul!a-35
From the aestivation ofBombax ceiba, two kinds ol'flowers (left-handed and right-handed) are distinguishable, and in a plant Llle two types Occur ill almost the same proportion. The mean number of stamens per flower ranges fi'om 62.72 to 102-00, and this variation is mostly clue to the variation in the number of stamens of the five peripheral staminal bundles of a flower. The central bundle has almost invariably 5 long stamens and | 0 slightly shorter ones, and variation in their number is the least. There is no significant difference in the number of stamens between a left-handed and a right-handed flower and their fi'uits weigh almost the same. Also the mtmbers of their seeds do not differ statistically significantly.
NOMENCL&TURE OF l~onlb(lX ceiba
Bombax ceiba, belonging to the Bombacaceae (included under Maivaceae by Hooker, 1872), has drawn much attention from systematists, and consequently the species has had several names. To the genus Bombax, Linnaeus assigned three species--B, ceiba, B. penta~a'trum and B. rel@iosum (Species Plalitarum, 1753, pp. 511-512). Recent work by Robyns (I961, 1963) advocates tlmt this species should be called Bombax ceiba. For further information on the nomenclature of this species, the works of Bakhuizen van den Brink (1924), Green (1929), Furtado (1939), Dugand (1945), Chattmjee and Raizada (t950, 1954), which are discussed by Robyns (1961), are useful. Some important synonyms for the species are Bombax heptaphyt[um, B. malabaricum, Salmalia malabarica, Gossampinus rztbra, G. heptaphytla and G. iVialabarica. Whether it is to be regarded as an American species or an Asiatic one, Bombax ceiba grows luxuriantly in the tropical regions of India and Burma except in arid tracts. It flourishes in the sub-Himalayas upto 3,500 ft. (cultivated at 6,000 ft. in North-west HimaIayas, S,'ilim All, 1932). It is a very common plant throughout Maharashtra and Saurastra and all the forest districts of Madras and Andhra, and to a limited extent in Kerala. It is one of the most conspicuous trees of West Bengal, particularly during January-April every year on account of its very bright, large red flowers. The species is also distributed in Ceylon, Malaysia, Indonesia, Qneensland and North Australia. In India the plant is known as the red silk-cotton tree to distinguish it fi'om Ceiba pentaIMra, the real silk cotton (Kapok or white silk cotton) tree.
AESTIVATION OF 'F~[E COROLLA
The corolla of a flower of this plant is clearly twisted in the bud, which condition is more distinct when the flower opens. A striking peculiarity of the aestivation is that in about half the number of flowers of any individual plant or any large flowering shoot T. A. DAvis 295
i'I..XTE 2 I
FMw,:rs of Bom/,a.x ~;ba 'h~,wh~g ti!2hbha~Mcd (3 fl.x~cr~ *,t, tit,- L,'t* r-wi. and h.fld~zmch'cl pr'lal-I Wislg. T. A. DAvis 297
of a pImat, the petals are twisted clockwise and the rest in a comm-:r-clockwise l~Lshion, a situation conspicuous in tim Malvaceae (Davis, 1964a; Davis and Selvara3, 196-1-). Fig'. 1 shows tin'ee right-handed flowers compared with a similar munber of left-handed ones. The individual petals are somewhat asymmetrical (the two lowest flowers in. the figure reveal this) and this character according to Schumant~ (1886) and Rendle (1959) is correlated with twisted aestiv/ttion. When viewed apically, the flower is considered left-handed (clockwise contorted aestivation) if the inmr margin of a petal curves cloekwisely towards the periphery and right-handed, if it curves counter- clockwisely. The stigmatic lobes usually show twisting in accordance with that of the corolla. In a dr?, flower where all the petals are held together wiLh the stamens, the petals remain slightly coiled according to the aestivation. Apart from Bombax ceiba, right- and left-handedness was observed in the following 14 species ,,f the Bombacaceae; Admlsolzia digitata, Bombax albidum, B. ins@his, Ceiba pentandra (])avis and Kundu, 1965), Ceiba rosea, Chorisia bzsignis, C. speciosa, C. grandiflora, Pachira aqz~atica, P. c~,athophora, P. insigl6s, P. lolNifolia , Pseudobombax grandiflorum and Salmatia insignis. Lawrence (1.963) assigns 22 genera and 140 species to this family.
PROPORTION OF LEFT- AND RIGHT-HANDED FLOWERS THROUGHOUT THE BLOOMING PERIOD
In order to ascertain whether the left- and right-handed l-low,.rs are produced in equal proportions on a tree throughout the blooming period, three t~ees standing within 200 metres from each other, very near the Indian Statistical Institute, Calcutta were marked and all the fowers that bloomed during the 1962 season e::amined. The data are preseated in Table 1. The flowers ofBombax ceiba possess largf quantities of nectar which attract quite a lot of bii-d species and on one tree 1 counted a~ many as 17 species on a day. The house crow (Corvus splendens) has a tendency to carry away a few flowers to an adjoining tree before tearing a petal or two to enable its bulk.~ beak to reach at the nectar. I may have lost in this manner upto 3 per cent of the fl~,wers of these trees. Whc table accounts tbr the rest of the flowers. Tim blooming period of tree I lasted 46 days and those of trees 2 and 3 lasted 3q. and 38 days respect!vet). I( may be seen that there are more left-hande,s than right-handurs in all three trees but the difference is significant for trees 1 and 3 only with Xe s (each with 1 d.f.} being 6"78 and 22' 18 respectively (P--0"01, 10-s). For tree 2, Xz is 4-67 ~P-=0"I0). The heterogeneity X" with 2 d.E to test the hypothesis that the proportion of left- handers is the same for all the three trees turns out to be 8'44 (P:=001). It can be concluded that the percentage of left-handers is not the same for all '~hree trees. The blooming period, which is nearly the same lbr all three tr~"s, was divided into 4 equal quarters (Table 2) to see if there was any indication th~,t the perccutagc of" left-handers changes with time in tim blooming period. Since tl~ele does not secm to bc ai~y great difference among the percentage of left-handers ft,- the tlu-cc trees during lhe period as a whole, the data for the three trees can t~,: pooled and tested for dift'crences in the percentage of left-handers. This seems only reasonable as trees of the same species standit~g close to each other are not expected {~, show any marked 5 298 Biomeby of Bombax ceiba Flowers
Table I. Bombax ceiba: Daily bloomblg in 3 trees
Tree I Tree 2 Tree 3
1962 Flowers Lefts ]~'lo;vers Lefts Flowers Lefts operled opened opened.
January 30 6 5 31 10 5 February 1 16 8 2 20 12 3 14. 6 4 42 21 t 1 5 t84 100 1 I 6 211 110 7 5 1 1 7 250 121 10 5 - - 8 202 95 12 4, - - 9 252 138 7 2 I ! I0 571 293 21 ll 1 1 11 708 344 43 25 4 1 12 729 373 66 37 4 4 I3 1124 581 102 49 7 5 14 1873 990 121 67 37 20 15 1472 730 129 63 83 51 16 1082 498 199 114 201 t08 17 978 457 144 71 172 102 18 1016 501 153 76 196 102 19 1370 687 176 90 358 208 20 1518 778 158 80 360 226 21 1788 943 205 116 395 187 22 1328 644 104 56 424 227 23 1118 555 85 42 458 228 24 843 445 84 42 526 271 25 1042 550 86 47 564 302 26 740 393 70 39 763 380 27 466 224 44 19 885 463 28 578 318 32 16 824 440 March l 267 140 21 8 ~.75 230 2 239 145 17 I0 358 193 3 285 152 22 12 293 i53 4 226 124 16 9 171 80 5 I88 88 II 4 125 6i 6 166 84 10 4 i47 70 7 135 69 7 6 107 56 8 102 48 I I 59 32 9 52 23 1 1 56 26 10 48 25 22 12 11 51 28 28 18 12 32 17 16 10 13 17 9 10 6 14 12 6 9 8 15 6 4 4 2 16 1 1
Total 23378 11888 2166 1133 8145 4285 Percentage Le~s: 50.85 52.31 52.61 T. A. Davls 299
Table 2. Bombax ceiba: Period of blooming and % of Lefts
Period Total Lefts % Lefts Tree 1 1st Quarter 1,778 914 51.41 2ud Quarter 13,658 6,882 50"39 3rd Q.uar ter 7,132 3,690 51.74 Final Quarter 810 402 49.63 23,378 11,888 50.85
Tree 2 1st Quarter 168 91 54.17 2nd Quarter 1,182 610 51.61 3rd Quarter 710 377 53' 10 Final Qtrar ter 106 55 51-89 2,166 1,133 52.31
Tree 3 1st Quarter 138 84 60.87 2nd Quarter 3,090 1,659 53.69 3rd Quarter 4,4.59 2,302 51.63 Final Quarter 458 24.0 52.40 8,145 4,285 52.61 All trees 33,689 17,306 51-37
interaction with the quarters. )i 2 with 3 d.f. turns out to be quite low (2.49) and does not, therefore, indicate that the percentage of left-handers changes with time dm-ing the blooming period. The overall X2 with 11 d.f. to measure differences in the percentages of left-hander~. for the 12 categories (3 trees >,4 quarte,s) turns out to be 19-97 (P=0"046) thus sup- porting the above conclusions. Data on the blooming of" trees I and 3 on i\u'ther examination show a very curiou~ day-to-day irregularity. For tree 1, on 33 days over 50 flowers opened (days witt, smaller number of flowers are ignored). X]~.=58-48 (P=0"003) which imptics ver,. significant heterogeneity. For tree 3, there are 23 days on which over 50 flower, opened, and ,~,~= for heterogeneity turned to be 4.3-20 which is also very significm~, (P=0-005). The conspicuous day-to-day variation in the proportion of lefts ar/d rights is very difficult to explain (Davis I964b).
NUMBER OF ~TAIVIENS IN LEFT- AND RIGIIT-HANDED I?LO~,VERS
TaMe 3 shows data on the total number of stamens for left- and right-handed flower> of B.ombax ceiba. Small samples of flowers were drawn at weeldy intervals from thre, trees during the 1962 season, and stamen numbers counted. The mean s{amen mlmbc~ for the 540 flowers sampled was 70.17. Approximately half the flowers sampled wet-,:: left-handers and the mean stamen number for left-handers was 69.93. This numbe~ 300 Biometry of Bombax ceiba Flowers
Table 3. Bombax ceiba: Number of stamens in Left- and Right-handed flowers (Data on 3 trees, season 1962)
Source Flowers Mean Stamens Free+2 Double examined Free Double Mean variance
1. All flowers 540 67-49 1-34 70'17 30"01 Left-handers 272 67'35 1"29 69"93 38"4.7 Right-handers 268 67.63 1.38 70.40 39.60 2. First week 150 68.97 1.88 72.73 60.48 Left-handers 75 68.65 1-72 72-19 55.78 Right-handers 75 69.19 2'04 73.27 65-41 3. Second week 150 67.86 I-4.1 70.69 41.32 Left-handers 75 67"91 1"45 70.81 46- I 0 Right-hal(ders 75 67-81 1.37 70-56 37.06 4. Third week 147 66-68 0"93 68.53 20-18 Left-handers 74 66.26 0.91 68-07 18.86 Right-handers 73 67-11 0.95 69.00 21-36 5. Fourth week 93 65-78 I'00 67.78 10.54 Left-handers 4B 65' 96 0"98 67.92 t 3.14 Right-handers 45 65-60 1"02 67'64 7'96 does not differ significantly from the mean for right-handers (70.40). At weekly intervals also, both kinds of flowers bore almost the same number of stamens, although the mean per flower decreased as the season progressed. Further data on the pheno- menon of reduction in stamen number with time are presented in Table 7. In the case of Hibiscus rosasinensis also, the number of stamens of left-handed flowers did not differ significantly from that of the right-handers, but the stamen number varied significantly with time (Davis and Ghoshal, 1965). A perusal of the values in the last column of Table 3 will reveal that the variances steadily decrease with time (within the flowering season extending for less than two months) from 60.48 to 10-54.. This is in marked contrast with the phenomenon recorded by Roy (1963) on the petal numbers of Nyctanthes arbor-tristis where the variances rose steadily with time during the four-month blooming period, and were from 20% to 40% greater at the end of the season than the beginning. Similar increases in variance of petal number were also observed by him in Jasminum multiflorum.
CAPSULES FROM LEFT- AND ~--,.IGHT-HANDED FLOWERS
During the 1963 season, 38 fully developed capsules were harvested from one tree just when they showed signs of dehiscence. Of these, 21 were fi'om left-handed flowers and the rest from right-handed ones. The capsules were sundried uniformly for a fortnight and weighed. The numbers and weights of seeds were also estimated. Not all seeds of a capsule were fully developed. The fertile (fully developed seeds) and the defbrmed ones were accounted for separately and the entire data are shown in Table 4. T. A. DAVIS 301
Table 4. Bombax ceiba: Data on fruits from Left-and Right-handed ,/Towers I. Left-kandeds
Weights (gin.) of Number o| seeds FrLtils Wt. of capsule (gin.) seeds cotton Fertile Detbmaed
1 2i'5,1. 5-09 5.93 t44 39 2 22.51 4.73 6'27 118 0 3 22"86 5"62 6-01 161 2 4. 26"22 6-41 6-96 211 25 5 20"13 3-93 5'14 254 18
6 22.95 5.81 5.94. 177 27 7 22-12 5.69 5-98 213 24 8 14..06 2.65 4-03 96 17 9 15.44 3-15 4.19 105 10 10 I4.09 2.48 3-65 87 0
I1 13-15 2"88 3"42 93 0 12 12-87 2"t7 3-25 110 7 13 16.45 3,25 4'53 I10 32 14 18.71 4,04 4.86 125 15 15 17.63 3-56 4.51 121 9
16 17-42 3-84 4-02 121 7 17 13-27 3'12 3.29 104 0 18 13-26 2.21 3'31 119 7 19 13"32 2"85 3"30 80 0 20 17"43 3-67 4'17 107 2
21 18.28 3.95 4.69 138 0
Total 373.7 ~ 81.10 97-4.5 2794 2-tl ),'[can 17,796 3.862 4'640 133.05 11 '4.8
IL Right-handed~
l 154.2 3'13 4-17 9o 0 2 1655 3"03 4"64 106 4 3 I 8.3ti 3-35 486 112 0 4 14.9i 2.50 3.78 14-6 16 5 1535 3.11 424 99 12
6 17-38 3 55 4.'89 I i 7 14 7 13-1.I 3-05 3-88 96 0 8 1625 3'62 4-56 19-t- 31 9 2082 3'56 6-02 133 I t 5 It) 25-21 6-51 6'58 197 45
11 I,I-83 2-73 ,I.'35 89 4 12 16.23 332 4..60 111 0 13 18'92 ,t-.20 54.1 154. 5 I 14 19.72 ,I-.07 5.45 I ,I 3 5 I 15 16.26 3-32 4-73 105 32
16 21.57 ,t..57 5-73 153 23 17 13 1 I 2'33 4.09 107 20
Total 29,14)5 59.95 81-98 2157 418
Mean 17.297 3"526 4"822 12~188 2't "59
L--l~. oa mean 0-499 0"336 --0" 182 6' 17 - 13'1 I 302 Biometry of Bombax ceiba Flowers
The differences in the weights of the entire fl-uit, or of the seeds and flint between the two types of fruits are not significant. The left-handers possess 4"86 pex" cent more seeds than their counterpart. ']'he difference between rights and lefts with respect to the number of sterile seeds is largely due to capsule 9, with 45 per cent sterile seeds. Even excluding this, however, the excess of capsules with sterile seeds on the right is significant (X~ =37-6). If the capsules are divided into those with more and less than 10% sterile seeds, one finds the following.
Over lO% U~2der 10% steri~ s~ri~
Right 10 7 I7 5gt 7 14 2l
17 21 38
~=0"42. This is insignifican.t.
DEVELOPMENT OF THE COROLLA
Serial sections were prepared from the apex of the fleshy pedicel upwards of several left- and right-handed flowers to examine how the aestivatiort in them develops. Fig. 2 shows six main stages before the initials of the corolla are well-defined. The numerous vascular traces are seen arranged as a ring, and at various intervals, a few traces are seen div&ging tO the periphery giving connections to minute appendages seen on the surface of the thalamus just below the fleshy calyx cup. These perhaps represent the reduced eplcalyx present in many malvaceous species. The conti~auity of the ring is ~ost (stages 2 and 3) when some traces proceed to the periphery. Ia stages 4 and5, they are brought together as a wavy ring. In stage 6, initials of the gynoecium are just formed. The ring of vascular traces at this stage breaks into five curved strips whiIe several traces diverge to the periphery to reinforce the calyx cup. In Fig. 3, the de- velopment of the five strips which form the petaI-stamen cord upto the stage where the petals are clearly distinguished are shown. "/'hose along the left row are from a right-handed flower and those in the right row are of a left-handed flower. At stage 1, the five broken cords form semicircular arcs with their convex surface.~ facing the lJeriphery of the flower. The arcs become more pronounced in the next stage. At stage 3, each arc has broken into a unit of three. The distal ends, curving in, detach fi'om the mai~ arc. The main are develops into a petal and the inner two portions of the petal-stamen cord develop into stamens. The relative positions of the two inner traces to the bigger outer one has some significance. If a line is drawn (indicated in two figures) fi'om the centre of the flower right through the gap between the two inner traces, the petal-trace (or most of it) will be seen remaining either to the left or the right side of the line. This is the situation in all five units. This lies more to the left in a right-handed flower and vice versa. In the next stage, each petaI-trace T. A. DAvis 303
Fig. 2. ,Stages iu the (l~.~,clopmcnt of petal-stamen cord of a B. ceiba flower. 304 Biome6y of Bombax ceiba Flowers
Fig. 3, Difli2rentiation oF tile corolla fi'om the petal-stamen cord in tL~ right-handed (left-row) and left-handed flowers of B. ceiba. T. A. DAVIS 305 is seen dividing radially and the petals just begin to be formed. With further radial divisions of these traces which ultimately form almost a ring, the five petals are clearly distinguishable, their fi'ee ends overlapping the petals to their right or left according to the type of flower.
THE ANDROECIUM
The androecium of the Bombacaceae is interesting and highly variable from species to species. The number of stamens varies from five to about one thousand in different species, and Robyns (1963) has given considerable importance to this numerical varia- tion in his classification of the genus Bombax. Bombax ceiba is even more peculiar in that a flower has three kinds of stamens (Davis and Mariamma, 1965). The stamens are fused at the base into a short monadelphous tube to which the bases of the succulent petals are attached in such a way that when the flower withers, the calyx cup, the corolla and the stamens shed together. The androecium consists of five peripheral staminal bundles which form a ring around the internal verticil. The number of stamens per peripheral bundle varies considerably and in a flower all five bundles do not necessarily bear the same number of stamens. The central verticil consists of two kinds of stamens and they differ also in length. The five innermost stamens though free almost to the base, envelop the style. Each of these stamens must be regarded as a double stamen, since it bears two distinct anthers. There is a prominent groove on the filament which is further suggestive of the stamen being a fusion of two. In many such stamens, the filaments separate into two at the tip, each carrying an anther. There are ten more stamens at the verticil which are slightly shorter than the inner five, and they partially cover the latter. Each of these stamens has a single anther which is reniibrm or more often coiled, the ends overlapping each other. The anthers of the centralmost five stamens are slightly bigger than the others and their coiling is more pronounced. The pollen grains of the innermost stamens are larger than those of the other two. A peripheral bundle may be considered as sharing three stamens of the central verticil of which two belong to the shorter kind and one to the.longer. The number of stamens of the central bundle is invariably 15 and the variation in the number of stamens per flower is due mostly to the peripheral hundles. A peripheral bundle curves sIightly and the convex surface facing the petal (as can be expected) has a greater number of stamens. Stamens are pluriseriate and are commonly united at the base. Each branch usually forks into two, each ultimate unit bearing an anther, and thus a normal stamen may be regarded as split: into two halLstamens. Such a splitting of stamens is termed ehorisis' by Velenovsky (1910).
VARIATION IN STAMEN NvMm.;v.
Flowers fl'om 53 trees of Bombax ceiba were examined f,_w tlle number of stamens during tile last five flowering seasons (1962-66). The trees are fi'om tile 24 Parganas and Hoogh/y districts of West Bengal covering almost a rectangular region measuring 306 Biometry oj" Bombax ceiba Flowers
about 150 km .'<50 km. In Table 5, data are given on 34 trees where flower samples not exceeding one hundred were drawn only once and the stamens counted. Ill the peripheral bundles two stamens sometimes fuse almost upto the anther and these are considered as double stamens.
Table 5. Bombax ceiba: Variation in the number of sta~nens (per flower) for different trees (Samples drawn only once)
Mean slamen nnn-tber "J.'rees Age Observed Flowers Range Variance (in years) on sampled Free Double (F+2D) (F+2D) '
1 10 3.3.63 50 62-52 0.10 62.72 57-65 10-25 2 20 "'27.2.63 I00 63.26 0.01 63.28 55-67 7.44 3 40 2.3.63 50 64..42 64-42 55-72 5.56 4 8 3.3.63 50 63.38 0.58 64-54 58-72 8.60 5 20 4.3.63 50 64.90 -- 64.90 62-67 0%5
6 15 3.3.63 50 64.82 0.06 64.94 57-71 2.83 7 15 2.3.63 50 65.48 0.04 65-56 63-75 4.21 8 15 22.2.64 50 67.38 - -- 67-38 65-73 5.00 9 45 24.2.63 50 67.52 0.10 67.72 65-75 6-41 10 35 24.2.63 50 68.08 0.12 68.32 63-75 8-84
11 4 2.3.63 13 68,46 0.04 68.54 65-79 22.48 12 20" 11.2.64 50 69.34 -- 69.34 65-75 11.62 13 30 14.2.64 50 69.54 -- 69-54 65-75 11-85 14 13 4.3.63 50 69-56 -- 69.56 64-78 9-30 15 25 2.3.63 50 69.64 0-02 69-68 65-81 11.67
16 18 27.2.63 I00 69'97 0-02 70.0l 65-77 13'49 17 25 1.3.63 81 70'59 0'01 70.61 65-80 14-23 18 25 25.2.63 50 70"64 0'06 70.76 65-75 7-03 19 17 20.2.64 13 69"00 1-00 71-00 67-75 6-92 20 15 20.2.64 50 71.16 -- 71-16 65-75 7"69
2l 30 7.3.64 18 7I'44 -- 7I'44 65-79 18'77 22 7 28.2.63 50 7I'52 0'I0 71-72 65-78 2'18 23 12 2.3.63 50 72'32 0-08 72.48 65-80 I4-30 24 12 3.3.63 50 73.60 0-02 73-64 65-75 !3"76 25 26 3.2.64 26 74"92 -- 74"92 73-77 1'00 26 8 24.2.63 50 74'42 0-30 75.02 66-87 15"12 27 23 17.2.63 50 75-26 0.04 75"34 73-78 0-30 28 30 2.3.63 50 76-94 -- 76"94 70-84 9-30 29 35 10.2.64 32 79'75 -- 79"75 75-87 9"I9 30 18 22.2.64 50 82"16 0'12 82'40 73-91 10'97
31 8 2.3.64 50 82-90 --- 82'90 74-89 11"45 32 20 20.2.64 50 89"52 0"04 89'60 79-102 14.32 33 25 28.2.64 50 92"30 0.02 92-34 85-98 8'86 34 28 3.2.64 15 102"00 -- 102.00 95-t 11 11"73
The data show that the mean 'standard' number of stamens per flower varies for different trees from 62.72 to 102-00. Trees with over 80 stamens seem to be relatively rare. There is also no indication that trees bearing greater numbers of stamens produce more double stamens. The one with the maximum number of stamens per flower (102) has ao double slamen at all. In another set of 24- trees fl-om North Calcutta, I W. A. DAWS 307
oo~oo ooo~ ooooo ooo~o ~oo6 66666 66666 66066 66666 oooo
..... m+... m- F- &
< 66666 66666 66666 66~66 666~
r cD
i+ _0
B~
,~"G ud
o~'~ d"~ E', + lJ m+~
--=+-,.~2 ~Z
OI c-I ~-c Ol C,I 3{)8 Biometry of Bombax ceiba Flowers collected flowers at weekly intervals during a flowering season and counted their stamen numbers. The data are given in Tabh: 6. The mean mtmbers of stamens for the entire perio~l for individual trees are also given in the Table. The mean for a ttcc varies from 65"971 to 99. 186. 'Frees 2:1 and lg produce high percentages of double stamens. A perusal of column (g) of the Table veil[ show that the modes for 19 of the 24 trees l'all at 65, 75 o, 85. At least two more trees wiI[/'all into the above group if the variation in the central bundle is ignored. Since the central bundle has the rather co~,stant number of 15 stamens, excluding these, the above trees will be seen to possess 50, 60 or 70 stamens per flower and these stamens are fi'om the five peripheral bundles. If these numbers are divided by the number of bundles, a peripheral bundle will have 10, t2 or 14 stamens on an average. This peculiarity is further brought out by the graph (Fig. 7) and Table 14. The 532}~ flowers account ~)r '4.00139 stamens or rot~ghly 75 stamens pet" flower.
Table 7. l?,ombax ceiba: Weekly mean number of stametzs (per.flower) in 24 lrees
~r intervals
Trees 1 2 3 4 5 6
1 67"36 67.04 65.08 63.98 64.25 . . 2 67.54 66-54. 65-,1.0 ...... 3 67.56 67-38 69.2-1- 67-00 65.92 66.27 4 69-20 67.22 65.20 ...... 5 67.08 67-9,1- 67"70 68.00 .... 6 69"80 69"36 67..54. 65.72 .... 7 73-28 71"56 70.20 70'25 .... 8 74"40 72"52 70'72 67"70 .... 9 75-22 75'05 72-40 67"88 66"06 . 10 74.36 73-68 72.32 66.50 65"00 . . I 1 73-94 72-36 74.00 70.01 .... 12 74. I2 74-44 73.02 73' 14- . t3 74.16 7,1.68 73.56 73.84 68.50 I4 74.12 73"90 72-80 ...... 15 76-34 75.68 72.76 70.00 .... 16 85.63 78" 18 71-52 71.90 .... 17 78-84 78.74 74..00 72.00 .... 18 81.94 80"1-6 76.12 72"56 70.92 .. 19 76-29 77-36 ~i.06 ...... 20 80.,1-2 80-,1'8 76.63 ...... 21 80.02 81.88 83-14 80.07 .... 22 83-54 81-93 83.04 83-62 .... 23 89.82 90.13 89-38 89.6~f - 24 103-18 102-76 99.46 96.44 94'66 95"63 T. A. DAws 309
NUMBER OF STAMENS A'r ~,A,rEEKLY INTEIIVALS IN ~d" TREES
Depending on the duration of" blooming of a tree, flower samples were collected fi-om the 24 trees and the sampling ranged fi'om three to six weekly intervals (vide Table 7). In ten trees the mean steadily dropped with time: The fall in the mean is particularly striking for trees 9, 10, I6 and 18, wlaieh is as much as 12'2, 12-6, 16"0 and 13-4 per cent respectively of the initial number of stamens. Only in one tree (19) does the mean steadily increase with time and this trend is seen for the first three weeks for tree 21 which again shows a drop in the fourth and last week. No obvious pattern is seen for the rest of the trees and the means vary irregularly.
Table 8. Bombax ceiba: Stamens per jIowerfrom three main shoots of the same tree
Tree Shoot Mean \;ariance
I 68.32 8.22 2 68.39 6.54 3 67.16 4-05
1 96.32 179-18 2 96.00 24-60 3 94.86 18-72
1 7t.18 12-31 2 70-,I-4 7-85 3 70-44 8.05
NUMBEI-~S OF STAMENS IN FLOWERS ON DIFFERENT SHOOTS OF A TREE
To see whether the flowers on differel~t large flowering shoots of a tree prodt~ced the same number of stamens or not, flowers from three trees were examined il~ 1964. In each tree 50 flowers each fi-om three large branches were collected and the total number of stamens counted. The data are presented in Table u.~ The variation in the mean numbers of s~amens between shoots is not statistically sigrfificarLt. Snmc of the branches of tree 24 (Table 6) spread right over another tree whose flowers have only 70 stameJ~s each and this i~dicates that the effect of the environment on the mmlber of stamens per flower is not great.
INTENSITY OF FLO'WERING AND ]N~-UMBk;R OF STAME'NS
Table 9 givc.~ data on the number of stamens per flower for three trees for 1962, 1963, 1964, 1965 arid 195(5. While tree 2 produced almost the same number of flowers during the five seasol~s, trees 1 and 3 behaved very differelnly in 1964. These trees stand wilhin 200 metres fiom each other and under apparently similar conditions. Tree 3 produced only about one per cent of the normal number and tree 1 about seven percent. The- cause lbv the severe drop in production in the two trees is not known; tim trees are perfectly healthy-looking. Periodicity in flower production as met with in Jl,lmlgifera 310 Biomet O, of Bombax ceiba Flowers
Table 9. Bombax ceiba: Intensi~ qf.flowe~:ing and nlonber of slamcns
Season Particulars Trees A 1 2 3
bloomed 2,166 23,378 8,145 1962 No. of flowers examined 150 200 193 Me:m number of stamens 64-67 67.89 75.70 bIoomed 2,200 25,000 8,500 1963 No. of tlowers examined 270 341 263 Mean number of stamens 65"85 67'28 77.94 blooined 150 25,000 80 1964 --No. of llowe;'s examined 112 302 63 Mean number of stamens 65.33 68.34 77.27 bloomed 2,420 25,000 9,350 1965 No. of flowers examined 263 300 283 lX.fean number of stamens 65-89 67.65 78.38 bIoomed 2,200 25,000 9,000 1966 No. of flowers examined 160 196 162
Mean number of stamens 65.32 66.85 79.13
N.B. The s(:rial numbers given for the above trees are different from those given in Table 5 or 6.
indica or to a lesser degree in Cocos nucifera is not known to exist in Yombax ceiba. While the mean number of stamens for tree 2 during 1964 was slightly higher than the pre- vious years, n'ees 1 and 3 produced almost the same numbers of stamens as they did in 1963, althc, ugh their productivity fell very abruptly in I964. During 1965 while tree 2 produced the same number offlowers, trees 1 and 3 produced about 1Q per cent more than their production in 1963 and 1962. The mean stamen numbers remained the same, as can be seen in Table 9. It is clear that the mean stamen numbers.pm~ flower for the three trees do not vary significantly witl~ density of blooming. The figures for 1966 only [brtify the above conclusion. While examining the stamen numbers of the three trees over five ?:ears, a gradual increase in the number of stamens is clearly noticed for tree 3, although no such trend is seen with trees 1 and 2.
ANNUAL VARIATION IN THE NUMBER OF STAMENS
Stamen numbers for a few trees Were estimated during the 1963 and 1964. seasons and the figures for five trees are presented in Table 10. Only one tree (15) produced more stamens ia 1964. For the rest, the differences arc not significant. T. A. DAvxs 911
Table 10. Bombax ceiba: A4ean mtmbers of stamens per flower of five trees during conseculive years
Stamens (mean) at weekly intervals Trees Season For the Vari- 1 2 3 4 5 6 7 season ance
1963 67.56 67.38 69-24 67.00 65.92 66.27 . . 67-28 6-81 3 1964. 68.54 68.24 68.66 68.56 68.32 67.84 66-00 68.34 10.40
1963 .. 67.08 67.94 67.70 68.00 .. ... 67-73 7-30 5 1964 69.65 69-30 69-26 68-82 67-63 66-79 . . 68.69 13.01
1963 74..36 73.68 72-32 66.50 65.00 . . . . 71.89 15-88 10 1964 71.60 71-74 70.18 68-59 65-60 .... 70.58 12.90
1963 74.16 76.68 73-56 73.8,t. 68.50 .... 73-77 9.60 I3 1964 73.90 73-30 74.12 73.27 71.47 .... 73-36 7.79
1963 76.34 75.68 72.76 70-00 ...... 74.09 14..91 15 1964 78.30 76-26 73.95 68-67 ...... 76.03 14.02
Table 11. Bombax ceiba: Aroma! variatiolz bz the number of stamens per flower for one lrec
Weekly interval 1962 1963 I96,1. 1965 1966 Mean for 5 seasons
First .. 67"56 68-50 68"36 .. 68' 14
Second 69-56 67.38 68"24 68.44 67.14 68.15
Third 68"20 69"24 (~8'66 66"92 66"90 67"98
Fourth 67"50 67'00 68-56 67'30 65-62 67'20
Fitih 66-30 65-92 68.32 67.68 67.80 67-20
Sixth .. 66.27 67-90 66"58 .. 66.92
For the season 67-89 67-28 68"36 67.65 66-85 6,7.65
(No. of flowers) (200) (270) (300) (300) (196) (1266)
In Table 1 1, data have been given on the mean numbers of stamens per flower at weekly intervals for one tree for five consecutive seasons. The overall mean is 67.65 stamens per flower. Although for 1964 the mean is slightly higher than for the other seasons, the deviation ti'om the overall mean is not great and the difference not signi- tican t.
"VARIATION IN THE NUMBER OF STAMENS PER PERIPHERAL BUNDLE
Tile peripheral bundles of 50 flowers fl'om 24 trees were examined during the tirst week of their bloom in 1964, though much larger data were collected over three seasons. " 9 Biometry of Bombax ceiba Flo+c,ers
I
cr~ ~ cr'+'I ~ C'I
Pa
c-I- ~" -+ 0+ I I .0 ~ o+ +. ~ ] t
L~. B co ze+~ B~ CL~
"+ ~ '~ +" I I [ B~ ~tj L~ +~ gi ~ +, ~o - - ] i.~ +u
1,0 +, r~
~o
P~ t,,.,.. "- §~+- .... i I U";.
q ++r~,
Lr" u ,'+ +U 0
c~ 0 !i +'~ ~ ~ ~ ~ ~ ~ I r
-~ I I c'4 M
+++.,+ T. A. DAvis 313 dlowers in which all five bundles bore the same number of stamens were separated out. Also, those flowers with fore" bundles bearing the same numbers and the fifth a different number were accounted for separately. There are five more possible situations as seen in Table 12, and the flowers are grouped accordingly. The 24 trees are also arranged according to the ascending order of their stamens per flower. It may be seen that in about 30 per cent of flowers, all the peripheral bundles bore the same number of" stamens. Only 3 flowers out of 1200 had all dissimilar peripheral bundles. Flowers with four simiIar and one different bundles are almost as frequent as the firsL category. A perusal of the Table will suggest that flowers with smalter numbers of stamens have a tendency to possess bundles of the first two categories. Tree I0 is an exception. Another striking peculiarity observable fiom the data is that trees 18 and 24 have the least number of flowers under the first two categories. It may be recalled that these are the two trees which produced the maximum number of double stamens (vide Table 6). This suggests that the production of double stamens is correlated with an unstable number of stamens per bundle.
Table 13. Bombax ceiba : Flozoers at weekly intervals according to nature of peripheral stamina! bundles
Trees Nature of bundles Weekly intervals Percent- age of 1 2 3 4 5 6 7 Total total
5 alike 14 22 17 8 18 22 14 115 33"72
4 same 1 different 15 12 11 14. 16 15 14 97 28'45 3 same 2 different 8 6 13 17 6 4 2 56 16"42 3 3 FI ,L1 7 5 4- 9 4 5 4" 38 11'14 2 q- 2 -b I 6 5 5 '2 5 4 5 32 9"38
'2q- 1 + I-1- 1 ...... 1 .. I 2 0"59 5 dissimilar ...... 1 I ~ 0'29 Total 50 50 50 50 50 50 ,11 34l
5 alike 15 6 10 8 18 2t 78 26'00 4 same 1 diffc,'ent 10 18 22 14. 15 15 94 31'33 3 ,~ame '2 different 12 l0 9 14 ~; 6 57 t9-00 3 b 1 "- I 9 I0 0 3 6 5 39 t3.0[) 2 q-2 + I 2 5 '2 8 [5 1 ~ 23 7.67 2-{. I q- 1-F 1 '2 1 1 '2 .. '2 8 2.67 5 dissimilar ...... t .... ! 0'33 T,,tal 50 50 50 50 50 50 300 314 ~r of Bombax ceiba Flowers
Table 13 (Col~,td.)
Trees Natttre of but~tdes Weekly intervals ~r age of 1 2 3 4 5 6 7 Total total
5 atlke .l.l 28 34 22 24 7 156 ~50.00 4 same 1 diffcma 7 15 [ l 15 19 2 69 26-54 3 same 2 difl'ercia~ ! 4. 2 5 4 .. 16 6-15
I0 3+ I + 1 1 2 1 4 l I l0 3.85 2 -I- 2 -" 1 .. I 2 4 l .. 13 3.08 2 + 1+ 1 q- 1 ...... I .. l /J-38 5 dissimilar ...... TotN 50 50 50 50 50 10 260
5 alike 19 12 17 10 14 8 80 29.63
4 same 1 dift~re~.t 15 20 16 18 15 6 90 33.33
3 same 2 differen: 13 7 9 9 I0 2 45 16.67
'13 3-1- 1 -[- 1 7 4 3 8 8 2 32 I 1.85 2+2+1 1 7 4 4 3 2 21 7.78
2+1+t+{ .... 1 1 .... 2 0'74 -
5 dissimilar ...... Total 50 50 50 50 50 20 270
5 alike 2 2 3 3 4 l 15 5.70 4 same i differcm 9 9 9 I5 9 3 54 20"53 3 same 2 differen~ 6 6 8 5 9 2 30 13.69
i8 3+I+l 12 14 13 10 15 2 66 25.10 2-I-2+ 1 12 7 7 I0 10 4 50 19.01 2+l+I-I 9 II 9 7 3 .. 39 14-83
5 dissimiiar .. I I .... 1 3 1.i4 Total 50 50 50 50 50 13 263
Table 13 shows further data on this phenomenon. A sample of 5 trees (out of 24) was selected and the nature of the peripheral bundles recorded at weeki 7 il~tervals. Over 250 flowers fi'om each tree were examined during six or seven weeks. The figures for the first week are the same as in Table 12 and the data of Table 13 help to compare the bundles during subs,:quent weeks. In trees 9, i0 and 13, the bundles bccaine more dissimilar with time. Tree 18 maintained almost the same condition throughot~t, and in tree 3, flowers with similar bundles increased with time with a rew-rsaI duri1~g the fourth week. Tree.~ 9, l0 and 13 may be regarded as demonstrating t~tigue-eflEct. T. A. DAvis 315
C3
tl co N-
III
c~
c~
u-j
X |
~ o 2m X
tt
tli • ""d Xll X • 11o X x X 1t0 X .,4 ~3 • fix X X • X
X II X X It
X X
X 1 • I • % Q) <2) C3 O~ r '.31G Biometry of Bombax ceiba Flowers
VARIATION IN THE ('~,ENTRAL BUNDLE
'l/hough the centrai bundle of stamens in l~ombax cdba is generally a constant possessing !5 stamens, three trees showed striking variations. Trees i8 and 22 bore extra :-,LzLmens in some flowers while tree 23 had less than the usual number. The in- ,.:~case in the number of stamens is brought about ci~.her by the duplication of the ~t,,~rter stamens or by the splitting of one or more of the 5 innermosL stamens. Stray c:~.,es of variatiotl in the central bundie were encountered in six other trees which must I)c ~'egarded as freaks. In tree 18, one flower had ]css than 15, though many had c:.:cess stamens. Tree 23 showed the reverse situation. In flowers where the central b,mdles show a reduction in number, the peripheral bundles also show a corresponding dillkrence. But there is no indication that the peripheral bundles show a corresponding ,:{ii:~'erence wheye the central bundles had greater numbms.
VAR.IANCES AND ~'V]~EANS OF 58 .[*REES
['he variances and means of stamens pe," tree for die 34 trees from where flower ~>'_~:ptes were drawn only once (vide Table 5) are represented in Fig. 4 by crosses. qhe dots represent those for the 24 t,-ees from where flowers were studied at weekly itm:rvals. Of the 58 trees, all but 9 have means betweml 62 and 80 and variances be,we&t 0"3 and [5"88, that is standard deviations between 0.55 alad 3.99. Two c:.:,:eptions have high means, six high variances aad ,:,lie has both. From the figure, iu is ctear that a much higher percentage of trees where flowers were sampled at weekly itttcrvals show high figures for variance compared to the rest of the trees. Thisis because in most trees the mean stamen numbers changect (decreased in most cases) with ~ime. There is no obvious correlation between mea~.s and variances.
Fig. 5. A peripheral staminal bundle of B. ceiba bearing two sterile stamens.
STERILITY IN ANDROEGIU~.~
Of the 58 trees examined, two trecs exhibited partial sterility of tim stamens. Sterility ira these trees is manifested by the production of stamens devoid of anthers or evela fully T. A. DAvis 317
developed filaments. They are reduced to staminodes about half the n~,rn-ial size. Fig. 5 shows a peripheral bundie with 2 sterile stamens. Each peripheral i
Period of bloom Flowers Perceiztageof examined flowers with sterile slame~is
First week 50 84-00 Second week 100 63.00 Third week 50 44-00 Fourth week 50 22.00 TotaI 250 55.20
The mean stamen number (including sterile ones) per flower of this tre,: is 81.87 and timt for ~-~lowers showing one or more sterile stamens is 82"14. This di!i,:rence is not statistically significant. But another tree (16) shows a different situati,.:~ . ()nly 12 out of 191 flowers sampled bore sterile stamens. Except for one Ilowev .,,,,ith two sterile stamens, the rest had only one each. The mean nmnber of stamens !wr flower with sterile stamens is 80.50, and that for the rest 76'12 (o," for all flowers 76-3!~ . Here the difference is statistically significant, and even if the sterile stamens ar~ ignored, the remaining normal stamens in these flowers are significantly more .tha~ in those without sterile stamens. in these two trees, flowers with sterile stamens are quite common and only :he peri- pheral stamens are reduced to staminodes. But there is another stray case of :, ce~tral shavt stamen reducing to a staminode in tree 18.
I~NTuMBER OF STAMENS PER PERIPHERAL ~UNDLE
In Fi~. 6 the number of peripheral bundles is plotted against the numlmr o~ -tamens per bundle for 24 trees. Bundles with as few as six and as many as 31 stamc;~s were also recorded. The mode is I2, the next l'requent number being 10. It is ~:!:o clear fiom thi~ figure that bundles with even numbers of stamens m'e commnner tlt~:n those wilh odd numbers. The odd nulnbers arise out of further duplication of a stature, w fiom a stamm~ not having the primary lbrldng. As stated ea,'lier, the stamens in a I>': ipheral 318 Biometry of Bombax ceiba Flowers
4o" ---4/
9X J,2"
DtSTfPIDI/T/ON' OF Sr.4b, IE~,'S / OUNI3I. E 8A'I,.' L/IN 5 pEf?IPt-IE:PAL ~UHDLES'j~
7 x IJ"
% ~'X tJ,
4xlE-" --
3xtO - !
2X~O ~
'~ i t i 200
*9 8 JO t2 r t8 ~8 20 2f 50 j:t SYAMEN$ PER [3UNDLE Fig. 6. Number ofperlpheral stamina! bundles against number of stamens per bundle. bundle appear in pairs, suggesting that a normal stamen splits into two, each bearing a single-celled anther. Thus, presuming that a bundle originally had 5, 6 or 7 stamens, when they all forked each into two, the bundles would have an even number of stamens. This phenomenon called chorisis was explained in detail by Velenovsky (1910). T. A. DAvis 319
c~ ~k
ua 0.
o
r ~S ta
~a <3
q ~o cc .z
S % U ro s
C5 cO "'
0 C=
t'. t~
r~
-- r
, I I i ..I i ! , I I I I ...l
S~l~07J ~0 "ON 320 Biometry of Bombax ceiba Flowers
-I I I I I I ~-
~q
I i 1
L~
C~
- I I
-~~,~~~~§ I ~-
C~
C~
T. A. Davis 321
co '~
c-!
N
.... i ] i -~
N 2
I I I I I- N -
@q
i ~ ".;4 "~
~,2"1 322 Biome#y of Bombax ceiba Flowers
A somewhat similar condition may be seen in Fig. 7 which represents the number of flowers according to total number of stamens pet" flower. The most obvious difference one can see is that flowers with odd numbers of stamens are commoner than their even numbered neighbours. This apparent reversal is due to the centraI bundle which has usually 15 stamens. Thus, under each column, if 15 stamens are subtracted, it will be seen that flowers with even numbers of stamens have greater frequencies. The two principal modes in this figure are at 75 and 65, and if the 15 stamens of the central bundle are excluded, they wilI be at 60 and 50 stamens per flower. Hence, one peri- pheral bundle will bear one-fifth of these numbers, i.e. 12 and 10, which are the modes of Fig. 6. The distribution of 5328 flowers of 24 trees according to number of stamens per flower is seen in Table 14.
Tab)e 15. Bombax ceiba: Seeds per capsule and stame1~s per flower
Trees Capsules Seeds/capsule Stamens] Length of Wt. of tested good sterile total flower capsule 1000 seeds (mm.) (gm.)
A 8 94-38 .. 94.38 71.18 70.00 33.22* B 49 94.24 34.06 128.31 .. 71.29 15.27 13 35 137.66 0-71 [ 38-37 73.77 91.57 26.20 1 50 136.78 7.58 144.36 ~55.97 93.70. 28.75 24 48 160.85 0.67 161.52 99.19 85.23 30.03 3 4.0 134-63 28.48 163-10 67.28 89-60 18.65 9 45 133.69 38-I8 171.87 71.89 66.87 21.23 11 45 175.53 0.38 175.91 72.5g 99.47 39-62 18 50 t88.42 2.96 I91-38 77-95 9I .52 27-54 19 I5 190.93 16.53 207.47 78.97 I08-27 24.95 10 10 332.60 6.90 339.50 71.90 95.30 32-86
*Based on the weight of 500 seeds.
NUMBER OF SEEDS PER CAPSULE
Capsules not exceeding fifty per tree were opened and their total seeds (including underdeveloped ones) counted for 11 trees of which 9 are from the list of 24 trees (Table 6) whose stamen numbers are aIso known. The number of seeds per capsule varied fiom 94-38 to 339.50 between trees (Table 15). The variance among capsules of the same tree is also quite high. Some trees produce sterile seeds in appreciable numbers--one has just above 20 per cent while two others produce about 30 and 35 per cent. One tree has no sterile seed while four others produce less than 2 per cent underdeveloped seeds. T. A. Daws 3_a
There is some indication that the immber of seeds per capsuIe between trees vm-ies with the size of capsule; the number varying with the size of the capsule within a tree is vividly demonstrated in many trees. The size and shape of capsules varies appreci- ably between trees, and so does the weight of seeds. The shape of seeds and the seed coat colour also vary between trees. Tree 10 appears to be very different fl:om the rest by possessing the maximum number of seeds per capsule and rectangular b,'own seeds. It may be mentioned that this is the only tree producing pale orange cotoured flowers. The weight of seeds also varies significantIy with trees. There is no indication that seeds fi'om capsules producing smaller number ot" seeds weigh more. There is also no correlation between the size ofa capsuIe and the weight of seeds.
DISCUSSION
The mean stamen number in Bombax ceiba varies significantly with individual trees, and this 'standard' number for a tree does not vary appreciably between branches of the same tree and during different periods of the same blooming season or between seasons. Also this number does not fluctuate with the varying intensities of flowering during different years. But trees standing very close to each other can differ greatly in their mean stamen numbers. The foregoing information suggests that the variabi- lity in stamen number in this species is genetically controlled which means that indi- vidual trees differ from one another in the degree to which the development of the flowers is canalised (Waddington, 1942, 1957). In 1962 when I sent data on only three trees (trees 1, 3 and 1.8 of Table 6) to Professor C. H. Waddington, F.R.S., he informed me that though the data provided good evidence to demonstrate canalisa- tion of development in 13ombax ceiba, a sample of only three trees was rather small, and he suggested that it would be valuable if I could sample some more individuals, or still better to do the genetics of these trees in detail to produce evidence that the varia- tion is genetic. But as the latter project will take 6-8 years, this aspect has been given up. Instead, more samples from different populations involving over half a million stamens were collected to see how far they differ genetically f'rom the cartier popula- tion. After scrutinizing the fi-esh data given in Tables 5 & 6, Professor Waddington made the following comments: 'In ttle wider sample you have now made, t notice that there is also some difference in variance between the different individuals, but that is not so large as it was in the original small sample. I am not quite clear what con- clusions can be drawn fi'om your data. They certainly show that there is some difference in canalisation between different trees, but there is, of course, no way of telling whether this is a genetic property since no breeding experiments have been done'. I give some further variation among trees which may or may nc~t be due to genetic differences. Tree 10 has very pale orange flowers which separates this tree fl'om the rest that bear deep crimson flowers. This is further unique in that the nurnber of seeds per capsule (339-50) is over 63 percent in excess of the next largesl seed-bearer. Its seed coat is brown (seeds of most others are black) and the seed is rather rectangular. Another tree is distinct by producing unusually large number 324 .Biomet~y oj" Bombax ceiba Fb.~wers
of sterile stamens. A further tree produces petaloid ~tamens. *~ret another tree produces fruits which are peculiarly dumb-bell shaped. A seedling growing under tree 1 (suspected to be the mother) commenced flowering in 1965 with 23 flowers. The stamens in aIl the flowers were counted and compared with those of the probable mother. Sl(,'met~ j:er f/aHat~ce ~:\'o. of
Moth,er 65.60 9.71 1t,8 Pro2eT~y 65'2 i 1" 14 23
Inspire of din conspicuous difference in the age of tl-c parent and progeny, the number of stamens per flower they bear is almost the rome. From data given in Tables 5 and'6 it will appear that the number of stamcn~ per flower and the age of the plant are not correlated. However, one of the tr,'es in Table 9 registered a progressive increase in the number of stamens during five years. In anot;her centre, two young plants were located growing very close to each other near a house- The flowers of these trees looked atike i~ size as well as the lighter shade of the petals. 25 flowers f'rom each tree were sampled and their stamens coun- ted. The mean stamen numbers did not differ significan/ly as seenbelow.
,57mne~sf']lower :.,ariance
Pla~zt 1 64-72 I8"31 Plm~t 2 64.'04 10-44
The late Professor J. B. S. Haldane, F.R.S. considered Bombax ceiba to be a nice example to demonstrate the variability of canalisation, and ideal material on which to study normalising selection. A somewhat similar phenomenon was observed in Euca!,.'plu.s globulus by Davidheiser (1965). The capsules of this species have usually four or five carpels, and those with three or six are rare. The more capsules a t,'ee produces of either of the extreme types, three or six, the !ess it tends to produce of the otlmr. The ratio of the fours and fives is also correspondingly shifted. Although this trend appears to be common to all the trees, the trees do not fall into a single series. Each tree appears to have a different ratio. There was no evidence to indicate that the ratios are different on different branches of the same tree. Four trees standi~ag very close to each othm gave capsules with different ratios. Davidheiser concludes that the ratios nf carpeli per capsule in Eucalyptzt.s globzdus are dependent upon a number of genetic factors. In a majority of Bombax ceiba plants sampled, the number- of stamens per flower dropped, though not significantly, with time. The peak blooming for a tree is durilag the second week, or for some trees between the second and third weeks. Thus, it may appear that there is a negative correlation between the number of stamens (per flower) and the intensity o~'[tower production. However, ti~e two trees producing sterile T. A. DAws 325
stamens behaved differemlv since the percentages of flowers in them bearing stcrile stamens and the ~aumbel ,_,f sterile stamens per flower diminished with time. In Hibiscus rosasincnsis, the m~mber of stamens per flower was found to vary significantly with time and the mean per flower increased as the flower production decreased (Davis and Ghoshal, 19(;5). During monsoon (June-October), flower production was at its maximum. BL~ ~.he mean stamen number per flower was higher during the rest of the year. The right and leR-ham!cd twisting of the corolla is also observed in a few families other than the Bombaca,:cae. All the 34 species of the Malvaceae and 14. of the Bomhacaceae observed s!~owed this floral asymmetry without an exception. Also a few species of the Lin::ceae, Caricaeeae, Plumbaginaceae and Cochlospermaceae examined behaved in a like manner. However, in Euphorbiaceae, Sterculiaceae and others, some species showc d this floral asymmetry while in others the petals are either valvate or imbricate. In a few palm species, the inner whorl of perianths representing the petals show regular t~isting (both directions) as well as imbrications within the same tree (Davis, 1964b; Davis and Kundu, 1966). The left-handed flowers i,~ alt the trees of Bombax ceiba investigated were significantly in excess of their counterp:Lrt although the proportion did not change with time durillg the blooming period. Among Malvaceous species, Hibiscus rosasinensis and Abutilon indicum produced significm:dy more left-handed flowers (Davis and Selvaraj, 1964). With regard to the foliar ~:s>mmetry in Cocos m:cifera, while the data from the Northern hemisphere showed an e~cess of left-handers, the right-handers were in excess in the Southern hemisphere (Da\is, 1964b). The North-South hemispherical difference was statistically significan~ at the 5~ level. Since the foliar asymmetry in the coconut is not genetically determi~ed (Davis, 1962), one might attribute some physical cause for this. Dr. Grotc Ikeb~ : of Tasmania, on the basis of available data, was inclined to believe that foliar asymmetry was influenced slightly by the earth's magnetic field. Since the data o~ /:ombax ceiba and the two Malvaceous species showing a significant excess of leR-i::,nders are fi'om the Northern hemisphere (mostly liom Calcutta), it may be presu~,~cd that some physical force more favourable for left-handed- ness is operating in this i,~_qaisphere. IL may be mentioned that Kundu and Sarma (1966) observed a slight e::c~ss of left-handed Corchorus capsularis plants in a population of 16,408 at the Jute Agri,:ultural Research Institute, Barrackpore near C,alcutta. In an earlier experiment al ille same centre reported by Sarma (1965), 54..31% of C. ca]:sularis improved str~i~: JRC,-212 were left-handed. No regularity in the pr,,duction of left- and right-handed flowers in the ultimate units (fascicles) of the infi,.~~:sccnce could be observed in Bombax ceiba. From Eichler's (1878, Fig. 115A, p. 284) diagram of the section of an inflorescence of Matva sy[vestris having three flowers, it m.:~ appear that a right-hander is alternated by a lei't-hander which in turn is followed l~>, another right-hander. To verify this, flowers flom several shoots of Hibiscus cam:abim,.', H. sabdari..ffa, H. csculenl~.~m, lJ. rosasincnsis and Ceiba pentandra (Davis and Kundu, 1965~ were examined. No doubt, some shoots of these species bore the two types of fl<,~,'crs alternating with each other upto five continuous leaf 326 .Biome#y of Bombax ceiba Flowers
axils. Bm analysis of the data from large numbers of shoots in difl'erent species did not at all suggest that the flowers are arranged in some regular o~-der. However, in a variety of Althav'a rosea, a slightly different situation was met with. Almost every leaf axil in this plant bore two flowers (once the flowering started~, the first and the axillary (main) one being a right-hander. The second flower, an extra axillary one located always towards the right side of the axillary flower was a ieft-hander. Thus in this variety ahnost all the axillary flowers (first flower in each leaf axil) were alike in their petal-arrangement. However, other 'varieties' ofAlthaea ro~.,a behaved like the rest of the Malvaceous species.
SUMMARY
Tile coro!la of Bombax ceiba, typical of the Bombacaceae, is cont,-,rted, all petals in a t'lower twistiI~g clockwise or counter-clockwise. Tile two kinds of [towers are almost equally common in any plant and even on a large flowering shoot. The entire flowers produced in three tt:ees during the 1962 season were examined, :.rod 51.37 per cent of them (out of 33,689 flowers) were found to be left-handers. I'he percentage of left-handers (or right-handers) did not change signifieantIy with time during the blooming period. No significant difference in the number of stamens between the two types ,.,f flowers was noticed. So also the weight of the capsule developed fl'om one type of flower and the number of seeds per capsule did not differ appreciably from those of their counterpart. A study of the internal structure of the flowers at various stages of development suggested that the petal-stamen cord even at a very early stage m:mifests asymmetry, and the distribution and deveIopment of the petaI-primordia in the r_wo types of flowers are reversed. A flower of Bombax ceiba has five staminal bundles arranged izl a ring which are united at the base and applicated to the lower end of the corolla. Each bundle has pluriseriate stamens whose number varies fl'om 6 to 31. Enclosed by the flve bundIes and surrounding the style is another bundle, the central verticil, bearing invariably 15 stamens of which the longer five have two anthers each. The short ten stamens of the central verticil and the peripheral stamens bear only one anther each. The mean stamen number varies slightly between flowers of the same tree and significantly between trees, and this is mostly due to the variation in the periphera[ stamens. Of the 58 trees from which flower sampIes were examined (involving over hail a million stamens), the mean per tree varied from 62"75 to 102.00. In many trees the number of stamens dropped sligEtly with time. Flowers on different shoots of a tree b,)re the same number of stamens. No significant difference was noticed in the number of stamens per flower in a tree during different seasotts. Also the mean stamen number did not vary with the variation in the intensity of flower production in a tree dtwing different seasons. Sterility of stamens was noticed conspicuously only in two of ~he 58 trees. The percentage of flowers bearing sterile stamens and the number of ~terile stamens per T. A. DAVIS 327
flower diminished with time during a blooming season. The number of seeds pet" capsuIe, estimated tbr I 1 trees, was found to vary significantly between trees.
AGKNOWLEDGEMENT
It was the late Professor J. B. S. Haldane, F.R.S., who induced me to look into the variation in the stamen numbers of Bombax ceiba. I wish to record my profound gratitude for the unflincMng help he rendered me for over seven years of my association with him. i thank Miss K. O. Mariamma for the anatomical preparations, Shri S. K. De for the drawings, and my wife for the help in counting the stamens.
REFERENCES
BAKHUIZI-N VAN DEN BRINK, R. C. [19241. Revisio Bombacacearum. Bull. Jard. Bat. Buiten-., Ser. 3-6, 161-240. C~'.AT'r~Nv.E, D. AND R.~.IZADA, M. B. (19501. Correct name of Indian silk cotton tree (Semat). Ind. For., 76, 154-155. CJ-LaT'rERJEE, D- AND RA:ZADa, M. B. (19541- Note on the typification of the generic name Bombax L. Taxon, 3, 102. DAVIBIIEISER,B. (1965). Eucalyptus fruit. Turtox News, 43, 66-67. DAvis, T. A. (i9621. The non-inheritance of asymmetry in Cbc0s nucifera. J. Goner., 58, 42-50. Davis, T. A. (1964a). Aestivation in Malvaceae. .u 201, 515-516. DAVIS, T. A. (1964.b). Possible geo-physieal influence on asymmetry in coconut and other plants. Procd. FAO Working Part.), on Coconut, Colombo, 2, 59-69. DAxqS, T. A. A.'-;D GI-mSHAB, K. K. (19651. Variation in thc floral organs o[" Hibiscus rosaJinel~sis L. J. Ind. Bat. Sac., 44. DAvis, T. A. AND KUNDV, A. (19651. Floral structure and stamens in Ceiba pentantha (L.) Gacrtn. J. Bombay .Nat. Hist. ,S'oc., 62, 394-41t. Davis, T. A. AND KUIXDU, A. (19661. Aestivation t~fperianths of Areca catcchtl Linn. fi'uits. J. Bombay jVat. Hist. Sac., 63. DAVIS, T. A. AND MARIA~.~.~,~A,K. O. (19651. The three kinds of stamens in Bombax ceiha L. Bull. Jardin Botanique de PEtal, Bruxelles, 35, 185-211. Davis, T. A. ,',tad SELVAR:XJ, C. (I9641. iVIoral asymmetri,' i2J MMvaeeae. J. Ban,be9, .,Vat. Hisl. Soc., 61, 4-02-409. DUGA,~ID, A. (i9451. Rcvalidacion dc Bombax ceiba L., coIno especie tipica del gene,'o Bombax L. y deserip- clan de Pseudobambax germ. nov. Caldasia, 2, 47-68. Eiciit.~r% A. W. (I878). Blitthendiagramme, II. Leipzig. IVuICrADO, C. X. (19391. Ti~e typilication of Bombax, Gassam/,itms and Salmalia. Ga~d. Bull.; Straits Settl., 10~ 173-18I. GlaeEN, M. L. (19291. Proposals by British Botanists, Ii~tcraational Botanical (_:engross, Cambridge (Ei~gland), 1930. Nomeuclature, I55-195. HooK~mJ. D. (I8721. FloraofBdti.*h hMia. I. L. Reave a,d C-: London. K.UNDU, B. (3. AND .%~.P,M.X, N'|. S. (19661. Direction -fl,:af spiral~ in Corchorus capsula~is L. (Under publicati.n). L,'~wl~et~<.:t~, G. ][. M. [1.9fi31. Taxoamny o['va~r Thc MaeMillml C,,: New York. R~..XDLI~, A. B. (19591. Thechts:iflcation of/7oweringptants. II. Dicol.rlcdom, University Pres~: Cambridgc. I~-OB'~'NS, A. (19611. Contribution ;., l'dtudc monographiqu~ do genre Boml,ax s. I. II La typificatbm dc B. ceiba L. Taxon, 10, 156-160. ROB'C~s, A. (19631. Essai de monographie du genre Bomba.x s. I. (Bt, lnbacaceae). Ball. Jardia Botaniquc de l'Etat, Bruxelies: 33, 1-315, 328 Biomchy oJ" Bombax ceiba Flower~
Roy, S. K. (1963). The variation o1" organs of indivldtl:d plants. J. Genet., 58, 147-176. S..'iLIm ALI (1932). Flower-birds and bird fl~Jwers ia incest, j. Bombay N~,t. Hist. 5be., 35, 573-605. S.~,RMA, M. S. (1965). A prelimiliary ~tudy ol~ the direction of leaf spiral and phyllotaxy in jute. J. Ge,et., 59, 140-143. SCHU,',L~,NN, K. (i886). Aestivation der Bluten and ihrc mechanlsel~.en Ursachen. Ber. d. deutsch, bolan. Ges. 4, 65. V~.ENOVSK'~', .J. (1910). Ke~leichc,de Mv~]d~ologie dcr Pf:,zen. Verlags-l~uchhandlul~g yon Ft. Rivnac: Prag. WADDING'rON, C. g. (1942). Cailatization of dev,qopn~,:tat and the inhe,'itance of acquired characters. N(,ture, 150, 563-565. WaDDIN~ZTON, C. H. (1957). Ttw ~tmtegyoflhegenes. (_;.:,,rge Al!en and Unwin Ltd: London.