299

NON-GENETIC POLYMORPHISM IN A C U3 'IINA TA L

BY K. R. DRONAMRAJU Indian ,5'lalislica[ Institute, 6'alcttlla-.'~5 (Received i0-11-59)

INTRODUCTION

In the year 1958 Professor J.B.S. I-Ialdane suggested that I should look for heterostylism in Indian where it had not previously been observed. I found a condition resembling it on a bush ofBa:~hinia acuminal,a L. This however differs fi'om the hetero- styIism so far reported in three respects. First, long and short styled are found on the same plant; secondly the lengths of the filaments of the two types are not negatively correlated with the style length; and thirdly, most, if not all of the short styled flowers, are female sterile, i began measuring the styles on this bush in tl~e middle of the flowering season, and the results encouraged me to measure them on four other bushes of the same species. A bush can produce up to 50 flowers in a day, so it was possible to compare the results on different bushes, and on the same bush at difl'erent times.

]~ATERIAL

The members of the species Bauhblia acumiTmla L. are leguminous belonging to the subfamily Caesalpineae. They have woody upright stems growing to a height of 12 feet. The consist of 2 leaEets joined to form a single with two lobes at the apex. The flowers are white and solitary, and very conspicuous, making the bush attractive in a garden. The 5 petals are slightly unequal in size. The number 10, but those of the inner whorl are usually smaller and sterile (Fig. I a, b). The pods are thin and flat, and contain 10-14 fiat seeds. Two bushes (Nos. I & II) ~fB. acuminata L. growing in the compound of 902 Barrack- pore Trunk ]Load, of the Indian Statistical Institute, and three (Nos. 111, I~v( & V) growing in 204 Barrackl)ore Trunk Road, were numbered in the order in which collec- tions ~a,ere started. All the opened flowers produced on each bush were picked daily between 08.30 and 09.30 hours in thc morning, and the styles were measured while the flowers remained fiesh. The periods d.uriug which these daily collections were made are shown in Table ] and iaclude the last days ofthe flowering seasol~. After removing the petals and stamens thc flower was placed horizontally on a millhnetre scale; and the distance between the end of the ovary where this is seen to narrow and the stigmatic expansion was measured. The length was recorded to the nearest millimetre. The style was straightcned fully on the scale before reading the measurement. In flowers where the style was much distorted m- ben t, a thread was also used to aid in the measure- mere. In 1960 a few petals and stamens were also measured in the same way. 300 F/ora[ pol3,morphi.s'~ it~ Bauhinia acuminata

Fig. I. Baultitda ac~uuiuala L

(a) long s@ed flower

(b) shorl styled flowtr

Table I

Tree Number I II [II IV V

Date /st examined 27.7 27.7 !I.8 I3.8 I4..8

No. of days examined 115 115 69 39 79

1-~ESULTS

The flowers are grouped according'to their styJe le~gths as shown in TabIe 2, and the bimodality demonstrated in Fig. 2. There were no flowers at 11 ram, and the division: K. R. DlZONAMRAJU 301

Table 2

.~.-'urfJ'Dcr of Flowers on Slyle Iength in ( millimctres ] II II1 IV V

27 0 ] 0 0 0

"26 1 4 0 0 0

25 I7 14 0 0 3

24 33 29 6 ] 3

23 72 61 20 3 8

22 151 120 35 10 38

21 254 180 40 17 47

20 316 242 71 3t 72

19 141 93 39 28 29

18 112 82 25 18 ]3

17 IGO 56 17 6 2

16 24 12 0 0 0

15 41 25 0 0 0

14. 10 7 0 0 0

13 1 0 0 0 0

I2 3 1 0 0 0

Total with long styles 1276 927 253 114 215

mm 11 0 0 0 0 0

10 2 I 0 0 0

D 0 0 1 0 0

8 18 4 5 1 1

7 35 15 5 1 3

6 33 3 7 1 1

5 100 37 12 4 2

4. ~7 25 21 9 3

3 83 20 13 t0 '2

2 49 10 10 6 2

Tolal wldl short styles ,I07 115 74 32 14.

Percentage wilh short styles 24..2 J: l "2 I 1.04-I 0.9'7 22'6:L:- 2.3 21-9:L-3"4 6.1 -:" 1.6 JO_" 9 Floral I}o!~,mor2hs il7 Bauhilaia acuminata

'2 3 4 5 6 7 ~ 9 10 1l I2 13 l,I. 15 !6 [7 18 19 20 2t '22 23 2:t- 25 Jqg. 2. Bimr~dat clislt'ibul.lt~ns of slvle IengtI~s in f~ushc~ [-V. N[iIlirnctvcs K. R. D~ONAM~,JV 303

G'J ~

//

i

J J \ \ \ \ \ \ ,\..< \ \ \ \\\\

p,. N ,

\\\\

,\\\

/

Fig. 3. Perct:mages o['lvl~g ~lyicd llo',vr in 10-day periods. Bushes I (,--) ar, d I1 (' - - - ) comp,-',,rcd. 304 F/oral pd.y,~orphis,z itz Bauhinia acuminata

t~o !\

\\\

-, \

, /,/-ll r, 2

I "7 /y'jr//

f / J //

I / / f I. / f I / /

/ f / r I F p r ,

Fi S. 4. Mean Icngd~s o~iong styled tlowcrs in i0-day pc~'i~ds. Bushes I (--) and II (- - -) COITl[}~.l.l'od' K. R. DRONAMRAJU 305 into long and short styled flowers is based on this observation. This is fiu'ther supported by the occurrence of very few flowers of style lengths around 11 ram. Thus the total lmmbers of flowers obtained fiom all the five bushes were I, 4, 3 and i for styte lengths 13ram, 12ram, 10ram and 9ram respectively. Bushes I and II flowered for about t 15 days fi'om the date they were lirst examined, and produced i683 and 1042 flowers respectively. They thus provided enough data for an analysis of the variation on the same bush and between the two bushes. The flowering period was divided into groups of I:ep. days; and the mean maximum and minimum tempm[atures, the mean rainfaI1 and the two mean style lengths of the two sorts of flowers were calculated for each period (Table 3). Similar analysis of data has been made for bushes III, IV and V, but is not pubIished and is less meaninjul because of the smaller numbers of flowers measured. The proportion of long styled flowers altered much through the season (Fig. 3) and on bushes IV and V no short styled flowers were produced for periods of ten to twelve days consecutively. The lowest percentage 0flong styled flowers was found on bush I, and the highest on bush V (Table 2). When all the five bushes are compared over the same period plants I, III andlVdo not differ significantly, but II and V differ from the rest and from one another significantly. The mode for the lengths of tong styles is at 20ram in all the five bushes (Fig. 2). The mode for short styles is at 5mm in I and II. In III this is at 4ram, in IV at 3ram and in V at 7ram and 4ram (Fig. 2). This deviation of the modes for short styles in bushes III, IX/" and V from those of I and II is possibly because very small numbers of flowers were obtained from them. This is partly due to the fewer days of flowering period observed and partly to the differences in the fi'action of short styled flowers produced by the plants. The mean lengths of the long styles in ali the five bushes rose at first and later fell to a minimum at the season's end (Fig. 4.). In I, II and V the highest mean corres- ponds to the hottest period in that part of the flowering season observed. The mean lengths of the short styles in all the five bushes fluctuated a good deal through the period 0f observation. No relation was noticed between the raini2dl and the mean style length and its variance. No order has yet been recognized in tlle position of the two types of flowers on any of the five bushes. As the season progressed the mean length of long stvJes s rows a slight rise and later a Fall which shifted the distribution nearer to thc mean lengths of short styles. If we cousider Bush No. I (Fig. 5), there were i14 flowers at 20ram and 8 flowers at 17ram for the first 20 days of the flowering period obser4-ed. But in the last 20 days there were 49 flowers at 17mm and only 10 flowers at 20ram.

FURTHER COMPARISON OF TI-IE 'r~WO KINDS O1,' FLO\.VERS

[[: had. seemed tha t other parts o[" short stylcd flowers were also relatively smaller than tl~ose of long styled {lowers; so in 1960 I made a few mcasureme:nts of petals and stamens (Table q.). The length of a petal was measured alol:g its :midrib. The breadth was measured on a line perpendicular to the midrib, passing through its mid poil~t. Tlm ftlament length was measured after removing the anthers. Table 4 gives the means of 306 Flottd 2u~,}~o~/~is]~z i~t Bauhinia acumim~(a

-H -II -I-I -t-I -H -~-! -I-I -H -H

c.'.,I r C:+ "~ ~ ~ C.O "if+ C<3

F~. II ~-I H -H -l-J -H -i-I -I-I -I-{ -H -If -F!

p-..+

II ~i -ll -H II ~! ~i -i.l -~ _~ o. ~ ~ ~.~,= . o ~? ~ ~

s E~ F1 -!-1 -H -II -I+1 --H ~1 +1 -IG -H -H -H

d >

•% t",- C+.~ u+3 CO O0 ":~ ,...++ + + + + + d -~-~-H +~-~-~-~ +~ +~ -~ -H

d-I H +H -i-I +1 -H -H +i -H :-H -! -fl d~

C~ "~ -- ~'~I ~'~

0 ~ 0 0 0 0 0 G C' G K. R. Dr~oNA~r~A]v .307

1 x , r

1-.9

~n

ko

o

i J r d

i

I I I _ m_.-i

I I I I- ......

I m ] [ [ 1 I

i

I I,,~ I I f

l,'[g. 5. Bim6dal distributions of' sL),Ic lcr~gLhs in bush I, ~t the beginrdng and the end o.f Q]e flawerlng period observed. ~24.. 7.58 to ]5. 8.58 ...... 4.H.58 tQ 23.li.58 308 Floral polymorp/dsm in Bauhinia acuminate

']"able 4 (Bush 1)

Me:ms in millimeters of: Style tcngd~ in pctzd petal lilam,:n t length nllll lengths breadths hmg slmrt

21 38 "2 24"2 22.4 13.4 20 36.-I 23.8 22.4 12"8 20 37.0 24-0 21.8 12.0 20 34.6 2'1-.4- 22.2 13.0 20 34-6 24"6 22.6 14.0

Total 36.16 24.20 '2'2"28 13.04

5 31 "6 20"4 19.6 i 1"2 4 35.8 21-4 2 ] .6 i2.6 3 30.0 21.2 21.0 I 1.4 3 32"2 20"6 20.8 10.6

2 31.6 18.8 21.6 11.6

Totat 32.24 20'48 '20.92 11"48 these measurements for 5 short styled and 5 long styled flowers. There is very little variation between the different parts of a single flower, and some of this must be due to details of their detachment and separation; also measuring to the nearest mm is too coarse a grouping for statistical analysis to be.meaningfi.tl. For example the 21 measure- mencs made on the flower with the largest range of petal lengths are given in Table 5.

J Table 5 (Bush I)

Style Pctal Pctal Filament Icngth length length breadth long short mm mm mm mm mm

...... 20 35 23 '22 i t 38 24 20 t3 38 2,1. 20 10 37 24 24 14 37 25 23 12

In cvery flower there were 5 long lilaments (i.e. over 19ram) in an outer whorl and 5 short (i.e. under 19ram) tilaments in an ironer whorl. This is not comparable to the .style dimorphism as there were 5 beIonging to each r in both long and short styled K. R. DaoNAmzAiu 309

tlowers. From Table 4 it is clear that both the petals and the filaments of short styled flowers are smaller than those of long styled flowers. Thus there is no suggestion of an inverse relation between style and filament lengths such as is characteristic of heterostylic ftol~,.,e rs. The stigmas of short styled flowers were narrow and unexpanded in contrast to those oflol~g styled flowers, and no short styled flower was ever noticed to set seed on the pla~lts. Forty-two were tested systematically, abou~ half by simple bagging of the flowers and about half by bagging after emasculation followed by artificial pollination using polten fl'om both kinds of flowers. None of these experimental flowers set seed. However, among the short styled flowers dissected for measurement a few had a more developed ovary than usual, leading me to suspect .that an occasional Fertile short styled flower may yet be recorded. Also because measured flowers cannot be subsequently tested for fertility it is not knows whether the functional difference coincides completely with the qualitative division based on style lengths, i.e. whether sbme flowers with style lengths of 12, 13, and i4mm may also be sterile, and conversely whether flowers with style lengths of 10, 9 or 8mm may sometimes set seed, anti may be more likely to do so than flowers with shorter styles. One long styled flower on bush ] yielded seed when fertilized by the pollen from a short styIed flower of the same plant. Bagged long styled flowers yielded seed.

DISO,USSION

Variation in the number or fertile stamens has been recorded in a number of Bauhb,ia species such as B. pz~JJurea L. and Jg. variegala L. The extreme case of reduction in the number of fertile stamens is seen in B. mmzal~&a Kurz which is characterised by having only one ['ertile in each of its ICiowers. But so far as I know, there is no previous record of the variation in the in the genus. Hvoker (1S79) gives the length 0fthe style in B. acumDzala L. as 0'5 i~ch, (12.7ram) with no Estimate of variation. Since { have found only one with a style of 13ram and only four of 12ram among 3,-1.27 flowers, it is possible th.at Hooker gave an average length of a few short and long styles. Such 1,~ unexplained average hinders the use of his description lbr iclentilication, because :his description implicitly denies a conspicuous (and physiologically important) differencc between flowers on tile same tree. Roy (1959) studied the variation of petal numhers in the flowers of.,Vyc[a~th,:.s arbor- 'risUs. This work is very similar to this on Ba~hiMa, as both were concerned with the :ollection of data regarding the variation in a certain organ or the plant with time. [-tlowever, the present study includes measm'ing as well as counting, whereas Roy's ,vorkwas concerned with counting ouly. Price-)'ones (1933) found tlaat the red blood ;Ell diameters in patients with pernicious a~mcmia were not only Iarge but also more Iariable than in normal peoplE. Attfield (1951) made a similal study on mice and ,llowed that red eel1 diameters were more variable in armemics thal~ in nornrals i~ stocks egregating for two of the dominant spotting genotypes. In Roy's data the variancc ,vasshown to increase at the end ofthe season. Though Ilbund that the lengths of the 31.0 Flora! polymorphism i~z Bauhinia acuminate functional styles decreased at the end of the season, no increase of variance was observed. Because of their part in fertilization, styles are nlore important sexual organs than petals which are the m'gaus of sexual display; therefore this is a possible reason why styles should, be more rigidly canalized (Waddington 1957) in their development than petals, if !,:deed this is a general rule. There is no evidence tidal the variation in style length in Bauhinia ac,mip~ata L. will have any evolutionary flttme. It may be merely a faihne el'developmental honleostasis which is common among horticultural plants and domestic animals. However, in perennial plants with no mechanism lot" wicle dispersal of seeds there may be selection pressure in f`avour of devoting more of the available materials to 1brining pollen, because the pollen will t,J sonre extent compete with pollen from other plants, while the seeds will cto so to a much less extent. Such selection is to be expected especially. in species [ike Bauhinia acamilm/a L wlaich have seeds with large food reserves that germinal:e under the parent plant and therefore compete with it. Such variations must however have been essential steps in ttle evolution of both diclinous populations and dioeciuus populations fiom hermaphrodite populations. A reduction in tl~e function of the stamens in long styled flowers would be a step towards dicliny. The advantages, if any, ofdic[iny are not very obvious to me unless there are ve,-y f`ew flowers to a plant and the sexes tend to flower at different times. In any species where heterozygosity is an advantage, the failure of one sexual attribute of a flower will Ncili:ate cross-pollination, provided other flowers provide a sul:lqcient supply of the reciprocal gametes. Though the condition orB. acuminata L. here described provides only a trifling facilitation of cross-pollination, if" combined with hewers having a simiIar abortion of the male parts, this facilitation would be greatl?.~ increased. This will become genetically more and more effective as the two kinds of incomplete flowers are associated with diKerent genotypes as would be expected if the}; are themselves genetically determined. Such_ variation in the degree of development of male parts is indeed characteristic of the genus Bauhinia (Hooker 1879) though not yet reported in B. acumilzata L. It is difficult to think that the tendency of a plant to produce incom- p[ete flowers is not due tO its genotype, a,~d also that the proportions of two or more types oft:lowers are not dependent on the genetic milieux and therefore sensitive to selection.': Though I have not discovered what determines whether a given style is canalizedas long or short it is unlikely tiora previous experience that this is due to somatic mutation i~a a wide sense. If heterozygosity is as important as modern results and ideas lead one to 5elieve, it is surprising that dioecy is not as common among plants as bisexuality is among animals. Spurway aJ~d Callan (!960) suggested that plants have been u~mble to evolve whole populations ofgenomes which maximisc the hybrid vigour for which they are responsible when heterozygous, but can be inadequate when homozygous because plant genes while in the haploid person have to control some developmental processes, whereas among animals no epigenetic fl.mctioning seems to be denlanded of the genomes when these are hemizygcms, except for genes on differentiated sex chromosomes. Alternatively, the very fact that some genes are functional in the haploid gmmration K. I'(. DI~ONAMRA}'t~ 31 1

h~s made possible the evolution by plat~ts of several different meclmnisms which prevent self-fertilization as completely as dioecy, and are more subtle than the bisexuality of animals, in that they also seIectively sterilize tlae matings of close relatives. It is vex'?' i~teresting that sex determining mechanisms comparable to s~, s"~ sa..incompatibility aIlelic systems have evolved in Hymenoptera (Whiting, t91-3; Mackensen, 195I) a group where the males are haploid, and therefore their development and physiology must bear a relation to their genotypc.~ compa:-able to that of a pollen tube t(~ its geno- type,

I am indebted to Dr. H. Spurway and Professor J. B. S. Haldane for various sugges- tions during the work and much help in preparing this paper. Mrs. P. C. Mahalanobis kindly permitted me to collect theflowers of bushes IIl, IV and V fi'om 204, Barrackpore Trunk Road, Calcutta-35.

SVM~aRY

The lengths of 3,-'!-27 styles from five bushes of J3auhinia acuminata L. ranged from 2 mm to 27 mm with modes at 5 and 20 millimetres. No flower had a style of ll mm in length and only 26 flowers had style lengths between 8 mm and 10 ram. The fractions of S}'10rt styled flowers obtained fl'om the 5 b~ashes were o,.1..9o.,~._ _/o, 11.,Io/1/o, 9o.~o/-- ~/e,

21"9 ~ and 6.1%. No order has vet been ~ecoomseo O" ' in the position of the two types of flowers on any of the five bushes. The mean lengths of the long styles on d:e first two plants show a significant rise and later a decrease as the day temperature rose and fell during tl~e season. The periods of flowering of the tatter three were insufficient for such comparison. The short styled flowers are female sterile, therefore such a variation could play some part in the evolution of both dioeey and dicliny.

REFERENCES A'ruTmLD, MVRIEL (1951). Inherited macroc'/tic anaemias in the house mouse. III Red blood cell diameters. J. Gem.t., 50: 250-1263. tlooKzr,,.)-. D. (t,379). FIovaorBritisb , Vol. II. L. Reeve & Co. Ltd. MACl,:ENS~, OTTO (1951). Viability and sex detern;~nation in the honey bee (Apis mdl~'ra L.), Gener 36: 500-509. *I>RmE-JONES, C. (19331!. Red blood ceil diameters. OxfordUniversity Press. P,o'~', S. K. (I959). ]t.egulalionof _Morphof_-~nesis in an Oleaceous tree, .~yclatIIhcs arbor-tristis. Nature~ 18.3: I410-141l. Sl'up,w.~', I-I.; (:ALLAN',H. G. (1960). The vigour and male stevility orhybrids between species T~ilurlts z,tdgari~ and T. hdr~ctictts, ,7- Genct., 57 : 84-118 "~'VADDINGTON,C. ]~[. (1957). The stt'ategy of the genes. GeorgeAlien & Unwin, London. WUtTIr~f~, ]~. V~r. (19~t3). Multiple alleles in ct,mplcmen.tary sex detenninatlon of][abrobracon. Genelics, 2:8 : 365-382.

*not seen in original.