THE GENETICS OF SOME MIMETIC FORMS OF , BROWN, AND PAPILIO GLAUCUS, LINN.

By C. A. CLARKE AND P. M. SHEPPARD The Universit),, Liverpool. (Received, june 3, 1957)

Bates (I862) put [brward the hypothesis that some anilnals obtain protection from ~heir predators by resembling or mimicking unpalatable or otherwise protected species and in consequence are mistaken for them by the predators. Mtiller (1879) suggested that even protected species would gain by resembling one another. These two hy-po- theses have been discussed and enlarged upon by many people but comparatively little work has been done either to determine the extent of the protection so afforded or to ascertain the evolutionary steps by which the has been brought about. The genetic work that has been done in this field (m.ostly with ) suggests that the differences between various mimetic and non-mimetic forms of polymorphic butterflies are controlled by simple mendelian mechanisms, often a single allelomorphic difference. This has led some people, notably Punnett (I915) and more recently Goldschmidt (i945) to maintain that, because the allelomorph must have arisen at a single step by mutation, the mimetic resemblance must also have arisen fully developed from the beginning. This view leads to ~various theoretical difficulties which have been discussed by Ford (1953). Because of these, both Fisher (1930) and Ford (Carpenter and Ford 1933, Ford 1937) take the view that when a mutant pro- ducing some mimetic resemblance is established in a population the resemblance is improved by selection for a gene-complex in which the original effect of the gene is altered towards more perfect mimicry. The truth of these hypotheses can be tested by suitable race crosses as has been pointed out by Ford (L953). Goldschmidt's view leads to the conclusion that when a resembles different of a model in different areas, the various forms will be controlled by a multiple alIelomorphic series, whereas in Fisher and Ford's view the same alldomorph will be responsible in all the areas but the mutant's effect will be modified by the presence of a different gene-complex in each region. Thus in P. dardanus Brown there are two very similar female forms, hippocoon in West Afi'ica mimicking niavius niavius and in South and East Aft'lea hippocoonides mimicking dominicanus. On Goldschmidt's view the difference between the two would be due to a different allelomorph of a single gene, whereas on Fisher and Ford's the same gene would control these forms and the difference wouId be due to modifiers at other loci. To distinguish between the two hypotheses unequivocally, it is first necessary to know the mode of inheritance of the mimetic tbrms within the subspecies to be crossed and to develop stocks of known genetic constitution. This paper" reports the results of an investigation into the genetics of some of the forms of the mimetic butterflies and :f~apilio glaucus. C. A. CLARKE AND P.M. SHEPPARD 237

~"IA'I.'ERIALS AND METtlODS I. Papilio dardanus

In this butterlly we have been concerned onl.v with race ceaea, which inhabits South Aft'ira northwards to Delagoa Bay. Four principal female forms occur :--1) f. cenea, mimicking Amqtlris albimaculala, 2) Ji hippocoonides, mimicldng Amauris niavius dominicanus, 3).f. trophonius, mimicking Danais chr),sippus, 4)f leighi, non-mimetic although tailless and not resembling the males. [There are also "two other forms, natalica and salaami, which are considered to be rarities, but Wells (see below) estimates that natalica is as high as 10% in his district.] The males in race cenea, as in all others, are always tailed and monomorphic. (see Plate 1).

DISTRIBUTION

In the southern part of South Africa, cenea is much the commonest, while north- wards hippocoonides is most frequently found. Our material was obtained from Natal, in the districts round Durban, Pietermaritzburg and Eshowe respectively. In these areas there appears to be considerable variation in the proportions of the female forms. Thus Leigh (Poulton 1928) gives the following distribution for Durban, obtained by breeding fi-om wild larvae :--

cenea 129 (85 %) hippocoonides 14 (9%) trophonius 6 (4%) l~i~hi 3 (2%)

On the other hand, Dickson (personal communication 1955) reports that in Durban in recent years sometimes cenea and sometimes hippocoonides appears tobe the com- moner, that tr@honius is scarce and that hc has never seen leighi. [With regard to fi leighi it is of interest that Poulton thought Leigh's figure of 2% much too high.] In Pietermaritzburg, Wells (personal communication 1955) estimates the proportion of the female forms as follows :--cenea 40%,, kippocoonides 35/o,o~ lrophonius 15% and natalica 10%. In Eshowe, Swanepoel (personal communication 1955) writes that l: hippocoonides is by far the commonest, ceuea comes next, and that gr@honius is very scarce. The original wild stock was sent to us in England early in 1954 by air mail and the occasional material received subsequently has helped to overcome the dangers of inbreeding. (These families are indicated in Table I by the word "wild".) It was found that eggs and caterpillars travelled we/l, but that pupae often died, apparently beiag abnormally sensitive to changes of temperature. In England the butterflies were bred in a g,'eenhouse at a ininimum temperature of 60~ and under these condi- tions they were continuously brooded throughout the year, the approximate time from egg to perfect being 2~ months. Pairing was always elTected by the method ofhandmating, (see Clarke and Sheppard 1956) and everlr effort was made to Inate the same male with more than one form of female. It was tbund that the males 238 Genetics of Papilio glaucus and P. dardanus

usually copulated more readily one to two days after emergence. After mating, the females were confined in large sleeves of black silk organza tied on to branches of growing citrus plants; the were fed on sugar and water every two days. Laying began a day or two after copulation and the best results were obtained at a temperature of between 70~ and I00~ alternating with periods of cooler conditions. Some of the insects did best in the evening under artificial light, but in general nothing stimu- lated them so much as sunshine. There was very great variability in the fertility of the females; sdme did not lay at all, and when they did the numbers of eggs varied fi'om very few up to about 80. The caterpillars were fed on citrus plants, but in many broods there was a high mortality among small larvae, which died before even starting to eat.

H. Papilio glaucus This butterfly has two forms of female; one is yellow and non-mimetic and the other is black, mimicking Battus philenor (see Plate 2). The males are always monomorphic and yellow (see Plate 3). We obtained the original stock of pupae of this butterfly fi'om the Chicago district where Dluhy (personal communication 1955) states that about 90~ of the females are black and 10% yellow; northwards there is a steep dine into Canada, where 100% of the females are yellow. To the South very nearly all become black, except in FIoi-ida, where there is a race consisting almost entirely of yellow females. P. glaucus does not occur west of the Rocky Mountains except perhaps in the extreme North where its western relatives are not found. From the western U.S.A. we were able to obtain a few pupae of two of these monomorphic relatives, P. rutuhts and P. eurymedon, neither of which has a black female form. We bred the butterfly at room temperature or a little above throughout the spring and summer months but overwintering pupae were kept in an unheated attic. Under these conditions tlae butterfly was double-brooded in 1955, but in 1956, owing to the late start to the season, we only obtained a partial summer brood. Liriodendron (tulip tree) w;as tbund to be by far the best food-plant for the larvae, but magnolia may also be satisfactory, though we have as yet insufficient evidence on this point. When lilac was used the larval mortality was always high. Pairing was invariably carried out by the method of hand-mating and we made every effort to mate the same male both to black and to yellow forms of female. The females laid on Liriodendron and magnolia and after several techniques had been tried it was found that sleeving them on the food plant, as with P. dardamts, was the best method. There was again marked variability in the fertility of the females, but neither form appeared more fertile than tlle other.

TEXT--TABLE 1 The table gives the fbrm (C=f. cenea, H,=f. hippocoonides, L=f. ldghi, T=f. trophonius), the brood number of the female, and Ihe brood number of the male where known. When the male has been successfillly mated more than once it is given a Greek letter, so that broods having a common father can be recognised,. Where wild parenls have been used the letter "w" is entered against them. Where the insect is of unknown parentage the letter "u" is used, C. A. CLARKE AND P. M. SI-IEPPARD

Table I. Breeding data in Papilio dardanus

C ~l: 6e'tlga~ H=f. hippocoonides, L=L leighi, T--l: trophonius, w=wild, u-~unknown. Greek letter denotes multiple mating of males.

OFFSPRING OFFSPRING

O 0 6 o~ 5 "o ~'O ~ G"a- ~ O+ Z ZC Z~ ~-o ~O'" ~o ~o ~ -~ o ~ o o ,'-

wild H 440 larvae 2 0 0 1 693 601 601 2 0 0 0

wild H 460 Iarvae 1 i 0 0 694 601 601 2 0 2 0

C H 467 w w 5 0 0 3 702 609 u ? 0 0 3

C H 469 w w 0 0 1 0 705 609 601 12 0 4. 1

H H 498 w w 3 1 0 4 708 609 620 0 0 0 l

H H 500 w w 0 0 0 3 713 609 601 10 0 0 25

H H 503 w w 3 1 0 3 714 609 u B 0 1 I

C }I 532 460 440 0 0 0 1 715 614 601 1 0 0 3

L H 60I 4,69 467 17 0 6 8 723 609 u 3 0 0 0

FI H 609 467 4,98 10 0 0 12 736 w w 4 2 0 1

H L 613 u u 19 0 0 19 741 684 684 3 0 I 1

H L 614 u u 4 0 0 7 75I 685 694- 1 0 1 0

]-I H 620 500 u 3 0 0 2 758 685 689 3 0 0 3

'[] [~ L 683 601 601 5 0 0 II 764 69q. 683 10 0 9 4.

"fJ ct H y 684. 60l 601 11 0 7 3 769 702 702 3 0 0 3

L 8 I[ 3' 685 601 60'1 4 0 4 1 771 683 702 2 0 0 0

H # H" 689 60I 60I 2 0 0 0 785 713 705 1 0 0 I 240 6'etzelics oJ'Papilio glaucus and P. dardanus

Table I (contd.) Breeding data in Papilio dardanus

,...... ods; A- !i ..... OFFSPRING

-g Ol" "~ (.)~" ~'o -q~" f-If" z z:: -o +, 71 z zc C31- ~ ~ (}b- ~-O

~ I <

C H 796 w w 11 1 I 0 4 1t 16 890 875 1 0 0 0

L L 797 705 758 I5 0 3 5 1 t28 889 886 9 0 5 3

H L 831 785 7-tl 8 0 4 5 1129 875 881] 3 0 2 2

C tt 85I 796 796 6 0 O 0 1130 886 886 15 0 0 18

C 8 H 852 796 796 5 1 0 I 1131 886 890 15 0 0 17

H 8 H 853 796 796 6 1 0 2 1133 886 886 0 0 0 1

C T 860 796 796 1 2 0 1 1142 w 1 0 0 0

wild C i~ 861 larvae 3 0 0 1 1154 100l 1001 11 6 0 1

L L 875 797 797 6 0 1 2 1157 1020 1001 12 1 4 0

H H 886 797 797 13 0 0 15 I158 1001 1020 2 2 0 0

L .i L 0 889 797 797 1 0 2 t 1159 1020 1001 14 8 .t- 0

H C 0 890 797 797 4 0 0 3 1165 I001 1001 4 2 0 3

H e L -q 100t 831 861 12 2 0 2 1166 1020 i020 1 0 ') 0

L ~ L 1020 831 861 5 1 4 2 1176 u 100I 17 4 6 5

L H 1035 831 85I 0 0 0 l 1191 1075 1020 18 10 0 5

L L 10017 831 851 0 0 1 1 1192 1075 1020? 7 4 1 2

H L" t 861 831 1I 0 5 4 1193 1075 1075 11 0 I4- 1

H t,' U 1083 853 852 4 0 0 3 1194 1075 I075 2 0 0 0 C. A. CLARKE AND P. M. SHEPPARD 241

I Table I. (contd.).--Breeding data in Papilio dardanus

OFFSPRING I OFFSPRING

C+ ' r C+ Of r~ Of 0+

......

L C 1195 1075 t083 28 0 10 8 1255 1159 i157 2 1 0 0

H H 1198 1083 1075 10 0 4 7 t268 1176 1159 2 0 0 0

L g 1200 1055 1083 9 0 9 4 1269 1176 i191 1 0 1 0

L L 1203 1075 1075 2 0 2 0 1273 1176 1212 one trophonius

L H 1205 I075 1083 17 0 9 9 1274 1191 I212 0 1 0 1

wild C 1212 larvae 4 2 0 1 1275 119I 1191 3 0 0 0

L 1 0 3 2 C 1215 1129 1116 kl inte~ex. (L/d) 1282 I212 1191 3 0 0 0

tI L I216 1131 1128 3 0 0 3 1295 I192 1212 1 1 1 0

H C 1219 1130 1130 3 0 0 1 1302 I192 1195 0 0 t 0

L H 1225 1128 1129 5 0 3 I 1306 1191 1192 1 0 0 0

H H 1226 I128 1129 3 0 0 2 1309 1198 1193 1 0 0 1

H ~ L I236 113t 114.2 I 0 0 0 t310 1193 1195 2 0 I 0

H ~ L I238 1130 1/4-2 2 0 0 0 1324- I193 1165 0 0 0 t

C H 1250 1159 1130 3 0 0 0 1327 1165 1194 4. 0 1 2

L L 1254 1157 1130 7 (I 1 0 1328 1200 1191 6 3 4. t -- ......

.P~ESULTS

P. dardanus The results are given in Table I. The table gives the brood number, ttm form of the female and the number of the brood of her origin where known, the brood number of the male where known, and the offspring of the mating. Because the po[ymorphism 242 Ge~zelics of Papilio glaucus amt P. dardanus

is confined to the females, the same male was mated wherever possible to different forms of 5 male. To indicate this in the table a male fercile with more than one femaie is given a distinctive (;reek letter in addition to the number ot" the brood whence he came. ~Iims in brood 683 a female hippocoonides li'om brood 601 was mated to a male 601, and ihe offspring were 54'6~ and 11 hippocoonides 9~- In brood 684 the salne~ was mated to a Mghi 9 also of brood 60i and the offspring were 1i ~5',5", 7 leighi ~ and 3 hippocoo,~i&s99. It must be pointed out that matings such as the above, which give much gel..,:tic in[brmation, are fi'equently impracticable since the stocks carried are not large, ,uitable butterflies do not always emerge at the same time, and the appropriate couple ofi,~ refiJse to be mated.

6'el~etics ofibrin cenea

The data reviewed by Ford (1936) strongly suggest that the difference between hippocooTffd, r and cenea is controlled by a single autosomal gene with cenea dominant and sex ccmtroiled. Our breeding experiments are in accordance with this view. For example, the direct ancestors of brood 831 had never segregated for cenea at any time whih" in our possession (four generations); neither had any of their relatives. Nevertheless in brood 1020 a female teighi fi-om brood 831 produced a cenea female among its progeny when mated to a male of wild stock. Moreover, when this same male was z~tated to a femaIe form hippocoonides, also from brood 831, cenea again appeared (brood t0LII) showing once more that it was the male which was carrying cenea and this form must be dominant to hippocoonides (unless X-linked). The fact that the males carry the .~ene responsible for cenea excludes total Y-linkage. Tota! X-linkage also

Table II. Analysis of families in which f. hippocoonides is the mother aud which segregate 2 forf. cenea andf. hippoeoonides

C L cenea "H=K hippocoonides Brood C H Total C. expected 498 1 4 5 2"5806 503 I 3 4 2"1333 736 2 1 3 1-7143 853 1 2 3 1.7143 100I 2 2 4 2.1333 [158 2 0 2 1-3333 or 2"0000 ll91 10 5 15 7"5002 1274 1 l 2 1'3333 k. Totals 20 18 38 20.~1.3 or 21.109 C. A. CLARKE AND P. M. SHEPPARD ~409~

seems unlikely, for if the locus was sex-linked in this way at least one hippocoonides in brood 1 I59 would be expected as leighi is not sex-linked (see below). To summarise the data quantitatively we can calculate the expected numbers of cenea in segregating broods where hippocoonides is the mother and compare these with the observed number (Table II). The formula for calculating these expected numbers is obtained as fo.llows. If in a very large family the expectation of producing cenea is -~- (as it will be in such segregating broods if cenea is dominant) then in n families of s members we should expect ~- ns cenea. But a t?action 2 -s of these families would contain no cenea, and would not be included in our list. Thus, we should only expect to find n(1--2 s) families containing at least one cenea. In those which did so we should expect to find ~- ns cenea, or on an average

1_ 5" or ---2 s-I s 1_2-~ , 2S_l

This is the number tabulated in the right hand column. However, in 1158 the father may have been homozygous for cenea. If so we expect 2 cenea. It will be seen that the fit to observation is very good on the assumption that cenea is not an autosomal recessive.

Genetics of form lelghi

It is shown by the data that the difference between leighi and hippocoonides is con- trolled by a single autosomal gene with leighi dominant and sex-controlled. Leighi is a rare non-mimetic form whose genetics have not been investigated before. There are several lines of evidence which indicate that ldghi is not recessive to hippocootddes. 1) Whenever a 9 leighi has been used, leighi has always appeared in the offspring if a reasonable number was produced, whereas if it were recessive this would not be expected. 2) Brood 601, from a leighi mother, must have segregated in a 1 :I ratio for &ighi and hl.ppocoonides if leighi is recessive. A To hippocoomdes of this brood, which would on this hypothesis have been a heterozygote, gave 11 hippocooni&'s ~ when mated to a sib (brood .683) whereas the same ~ mated to a leighi sib gave 7 leighi and 3 hippocoonides (684.). Had &ighi been recessive, brood 683 would have been expected to give a ratio of 3 hippocoo,ides to I leighi which it did not do. Moreover, the discrepancy is on the borderline of significance. Using the same argument with broods 797 and 886, a 3 : 1 ratio of hippocoouides to leighi would have been expected in brood 386, whereas I5 hippocoo~zidcsg~_ were obtahmcl. This shows conclusively that leQ,hi is not recessive. The question arises whether tim "leighi" heterozygotes and homozygotes resemble one anoiher, and brood 1193 suggests very strongly that they do. This mating was between a leighi 9 and a sib and produced 14 leighi'.?9 and 1 !@pocooMdes. Not only is there a significant difference fi'om a 1 : l ratio, showing that the c~ was carrying lcighi 244 Genetics of Papilio glaucus a~d P. dardanus

(as can be confirmed by the results of broods 1 194 and 1327) but it is even more extreme thana 3 : I ratio, suggesting that the brood is not a 2 : 1 but a 3 : 1, thus demonstrating that leighi is dominant. The genetic proof that some of these individuals were homo- zygous for leighi must await the results of broods now being reared. Knowing that leighi is dominant to hippocoonides we can see fi'om broods 6133 and 684. that the gene is not on the X chromosome, tbr the .,3" was not carrying leighi and yet the 9 [eiehi pro- duced this form among its offspring. Moreover, the data give abundant evidence that. it is not l--linked--for example, brood 1193 shows that the ~ was carrying the gene. Table [[][ gives the obsevved and expected number of leighi calculated in the same way as for cenea. It will be seen that the fit is very good on the assumption that le(~hi is an autosomal dominant.

Table Ill. Analysis of families in which Ji hippocoonides is the mother m~d which segregate for f. teighi al~df. hippocoonides

L=E ~hi H=f. hippocooni&s Brood L H Total L. expected 694 2 0 2 1"3333 or 2"0000 705 4 I 5 2"5806 714 1 1 2 1.3333 83I 4 5 9 4"5088 I075 5 4 9 4.5088 1198 4 7 11 5-5027 1327 1 2 3 1,7143 Totals 21 20 41 21.482 or 22-I4.8

Relationship betweert cenea and Ieighi

We are unable to say fi'om the present data whether the difference between cenea and leighi is controiied by a pair of allelomorphs. However, there is some informa- tion on the relationship between the two forms. Brood I159 segregated for leighi and cenea only. If the difference between the two forms is controlled by a pair of allelomorphs then unless the [eigai mother, which cannot be homozygous for this gene, was carrying cetzea, at least one offspring in four would be expected to be hippocoo,~ides, the ,3" being heterozygous for this allelomorph. However, there was none in twelve individuals. Thus the leighi 9 was probably carrying cenea, so that leighi would be dominant on this hypothesis. However, there is an alternative, namely, that the two forms are controlled by genes at independent loci and that there is an epistatic effect. If this is so, then if the teighi 9 was not carryir~g the allelomorph for cenea, again one olCt}pring in four should be the double recessive hippocoonides. Therefore k this brood makes it probaMc that le@hi is either dominant to cenea or that there is an C. A. CLARKE AND P. M. S'HEPPARD 245 epistatic effect such that if the allelomorph producing h'ighi is present the insect will be leighi regardless of its genetic constitution with respect to cenea and hippocoonMes.

TEXT--TABLE IV

1. The table gives, the colour (B=black, Y=yelIow) and brood number oF the female, and the brood number of the male where known. When the male has been successfully mated more than once it is given a Greek letter, so that broods having a common father can Ee recognised. Where wild parents have been used the letter "w" is entered against them. 2. "When atl the pupae have not yet emerged the number of those remaining is given in brackets --thus, in brood 963, 54~J, 0 black ~9 and 7 yellow 99 have emerged and 3 pupae remain. 3. The six families which have segregated for fcmale colour are given in bold type.

Table IV. Breeding data in Papilio glaucus

Y=yellow, B=black, w=wild, E=P. eurvmedon Greek letter denotes muhiple mating of'males.

I. OFFSPRING OFFSPRING

O jl 2 b O+ G-L Z ~E @ ~t,_, vO O+ >

o K e "-44 .

Y B 102 w w 0 0 1 502 w w 8 8 1

Y Y 108 w w 1 0 I 506 w w 16 2 23

B Y 112 w w 2 2 0 512 w w 1t 0 16

Y Y 138 w w 2 0 0 522 w "v 5 0 8

B Y 172 w w 1 0 0 523 w -.215 1 0 2

B B 173 w w I 0 0 63'1. 501 501 10 16 3

Y Y 182 w w 1 0 0 63~J 512 502 24 0 13

Y B 215 w w 2 0 "1 6"1-0 502 506 28 35 0

B B 237 w w 5 0 0 64.2 502 512 18 ltl I

B J~ B 252 112 i 12 4 2 0 643 502 502 2 2 0

Y Y 501 w w 1 1 2 646 512 ~502 2 0 5 246 Get~efics of Papilio glaucus and P. dardanus

"Fable IV (contd.). Breedingdata in Papilio glaucus

'-d OFFSPRING OFFSPRING o

d CW u 0+ Z 'co O~- > o N -.go ~~ o "d ol- o oo , ~:1 >-

B B E 647 502 506 2 I 0 968 6r w 5 o o (5) Y Y 648 506 502 1I 0 12 970 669 640 0 0 2 B 649 502 506 7 6 0 972 640 640 7 13 0(121 Y B 650 512 502 4 0 1 973 6,1.0 640 0 1 0 Y -/ Y E 655 528 5O2 1 0 0 975 669 w 7 0 0 B B 656 502 512 3 I 0 977 642 669 2 8 0 (7) Y Y E~ 659 512 512 1 0 2 (1) 978 669 w 0 0 0 (~_) Y y B 662 501 502 I0 0 7 1005 673 673 2 3 0 Y B 663 506 506 4 0 10 1008 673 671 14 4 I Y B 666 522 522 2 0 I 1014 642 662 1 0 0 Y 1., Y 66w 512 502 18 0 9 1015 662 642 20 0 12 B B 670 506 522 4 7 0 1039 656 648 2 0 0 Y B 671 522 522 1I O 5 1040 673 648 17 I0 0 B g Y 673 502 522 13 5 O 1041 648 642 0 0 I(1) B Y 899 643 643 23 26 0 1047 648 648 o o o (4) B B 9t9 640 64-0 1 0 I048 640 663 o I o (i) B B 920 640 640 3 4 0 (S) I062 649 648 3 7 0(17) Y B 934 639 639 8 0 6 1073 634 648 5 3 o (4) Y B 939 639 639 1 0 3 1078 670 64.8 l0 4 0(I7)

B E:. Y t 954 64O ;v 4. 0 0 (2) 1079 648 650 2 o o(191 Y B 956 669 640 4. 0 9 1085 649 648 0 0 0(191

Y Y L 961 639 640 i6 0 17 1087 648 650 0 0 4 Y 7t B 670? 963 639 640 5 0 7 (3) I088 649 671 ? 1 0 0 B Y 0 967 642 642 I I 0 1089 w 648 4 0 3(14) C. A. CLARKE AND P. ~.SHEPPARD 247

Genetics ofibrin trophonius

We have had only one family (brood 1273) in which f. tr@honius appeared. 'A leighi fiom brood 1176 mated to a wild c~ gave a f~ trophonius. As trophonius had never appeared in our broods before this date it seems likely that the gene came in from the ~, in which case our, tr0phonius 9 is a heterozygote and trophonius cannot be recessive to all othe/forms. This interpretation will not be proved until further breeding work has been clone, for a recessive could be carried for a number of generations in the stock without a homozygote being found.

Genetics of P. glaucus

The results are given in Table IV which is constructed in exactly the same way as that for P. dardanus. It has been suggested on insufficient evidence (Cockayne 1932,. Remington 1954, Ford I953, Clarke and Sheppard 1955,) tt~at the difference betw-een the yellow and black forms of the female is determined by a single pair of autosomal allelomorphs, with the black form dominant and sex-controlled. The data contradict this view. In nearly every brood the female offspring are of the same form as the mother regardless of whether she is black or yellow- and also regardless of the origin of the male. On rare occasions there is segregation. However, when this occurs the exceptional females behave genetically like their sisters in the sense that there is no very great tendency for the progeny to segregate, their offspring usually being of the same form as themselves. Brood 506 is an example of one which segregates, there being two black and twenty-three yellow female offspring. In contrast to this another yellow ~ mated to the same 6~ (brood 512) produced sLxteen yellow and no Mack ~. From this one might assume that black is recessive despite the fact that there are too few black individuals. However, brood 634 from a black ? gave sixteen black to three yellow, which if black were recessive would have to be a 1 : 1 ratio which it clearly is not. Thus the difference between the forms cannot be controlled by a pair of autosomal allclomorphs whether black be considered as recessive or dominant. Moreover, the character cannot bc controlled by a locus confined tn the X chromosome because in butterflies the ? is the heterogametic sex, and thus ~9 would sometimes be expected to give no 9 offspring like th.emselves. Segregation should also occur frecluently in our broods. Neither of these two requirements is fulfilled by the data nor can the gene be totally 2"-linked because there is segregation in some broods. At first sight, partial sex-linkage might appear to be a likely explanation, but even this does not fit the facts. To take onl? one example: on this hypothesis broods 506 and 512 show that yeIIow would have to be dominant to black, for if black were do,ni- nant and there was partiaI sex-linkage thc d' must have been a heterozygote to produce any black offspring as neither of the ~? to which he was mated was black. But. if he was heterozygous, black and yellow" ~ offspring would be expected in equality which they obviously are not. Therefore, black must be recessivc to yellow and the two black off'spring in brood 506 might have resulted fl'om a. cross-over. But if black is recessive, then in brood 634 thc d" must have been hcterozyg'ous for yellow. One 248 G'enelics 0fPapilio glaucus alzd P. dardanus would therelbre expect equal numbers of black and yellow ~ offspring, but in fact there is a considerable excess of })lacks. Thus the hypothesis of partial sex-linkage cannot be correct. The next possibility is that our results are due to cytoplasmic inheritance, in which case segregation for a form unlike the ~ parent could be due to cytoplasmic factors carried in the sperm. If this hypothesis were correct there must be an "all or nothing" effect, otherwise we should have got intermediates rather than distinct classes. This suggests that if the cytoplasmic factors exist, they are particulate and that one being carried over in the sperm is sufficient to switch the balance. Edwards (1884) figures a :_~ P. glatwus one side of which is black and the other yellow. Such an effect is extremely unlikely on the hypothesis of cytoplasmic inheritance, unless only very few particles are involved. However, if this were so one would expect ttlem often to fail to pass into both cells at mitosis, in which case mosaics would be common, which they apparently are not. It therefore appears unlikely, though perfectly possible, that we are dealing with a case of cytoplasmic inheritance. It seems theoretically possible that purely external environmental causes may on occasion be able to switch the cytoplasm from one "state" to another, one giving the black and ttle other giving the yellow form. If the :'states" were semi-stable we might get the results we have observed, but again as segregation is not rare we shouId expect mosaics to be common as they are known to occur, at least one bilateral mosaic having been reported. Another possible explanation could be that reproduction was usually by a partheno- genetic mechanism but with the sperm being necessary to stimulate the egg to divide. This hypothesis can be rejected with a fair degree of certainty. We have obtained crosses between glaucus and rutulus and glaucus and eu,ymedoT~ which produced un- doubted hybrids as judged by their morphology. Furtfiernaore, tbur sib matings of the offspring of 640 were made, and in three of them an abnormal form appeared (see Table V). This is characterised by the light yellow submarginal Iunules on the

Table V. Appearalwe of abnormal recessive in q~prbzg oJ" sib matings of Brood 640

OFFSPRING Brood Colour of Wild type Mutant Wild type Mutant No. mother 6"~ d~ ~0 .,:r 919 black I 0 0 I (black) 920 black 2 I 1 (black) 3 (black)(3) 972 black 7 0 13 (black) 0 (12) 973* black 0 0 0 1 (black)

*The father of this brood was also mated to yellow fr 639, (brood 963, see tab!e IV) which gave no mutant types among the 12 ofl'spring which have so for emerged. The mother and titther of all these broods were wild type with respect to the character under discussion. Figures in brackets ddnote numbers of" o~,erwintering pupae. C. A. CLARKE AND P. M. SHEPPARD 2i9

upper and underside of both tbre and hindwings spreading outwards to reach the margin of the wing, (see plate 2, fig. 5) and it has occurred in five black females and in one male. We have not been able to find a report of this form of P. g[aucus but it is of interest that in the British Museum (Natural Igistory) there is a similar aberration in B. philenor whicli has been given the name of var. wismu:hii. An apparently similar variety Ivts been described for P. rululus (Rothschild and Jordan, 1906). This ab- normal form in P. glaucus has been seen in no other of out" broods and must there- fore be judged a recessive for the present, despite the fact that there seems to be an excess of them (for data see Table V). Using this recessive we can test whether it can be transmitted by the c5', for if it can, fertilisation must have taken place. In spite of every care in breeding, the possibility of mistakes must always be con- sidered. In the present experiments precautions were taken against this, and when a larva was misplaced, which occasionally happened, it was rigorously separated fi'om all the others. It seems very unlikely, therefore, that our segregating broods are due to errors. Again, that a 9 of one form can give rise to ~ of another has been observed on severaI occasions. Thus Edwards, (1884) reports as follows :--"In June, 1875, I obtained eggs by confining several black females upon the limbs of a tulip tree, and there resulted therefi'om, in August following, 9 ~c?, 2 black 99- Part of the chrysalids passed the winter, and in the spring there emerged 9 ~, 5 black 9~. In the spring of 1872, (here emerged from chrysalids o/" the previous year, the eggs having been laid by black females, 15 c~, 7 black _0.9, 2 yellow ~Q. In the spring of I877, fi-om eggs laid by black, 21cirri, 7- black c~9. From eggs obtained in the same way fi'om yellow females, there resulted in the spring of 1872, 8 c~c~', 3 ~cx~, alI yellow. In August 1875, 3 ~c~~, 12 .,~,on all yellow, and fiom part of the chrysalids of this lot which passed the winter, there appeared in the spring of i876, 4c~c~, 12 yellow 9~, I black 9- This is the only instance out of many broods raised, in whiell a black has come h'om a yellow mother, though on two occasions, besides the one above mentioned, a yellow female has come fiom black." It will be seen that Edwards unlbrtunately did not keep iris broods separate, but this does not matter tiom the present standpoint. Edwards (I884) also reports a further example (bred by Akhurst) in which a yellow female from the New York area gave two black females among its large progeny, all the other females (number un- stated) being yellow. Now the black form is uncommon in this area, and therefore it is unlikely that this result was due to contamination fl'om eggs or larvae brought in on plants used to feed this brood. Finally, the lkct that a bilateral mosaic has been obtained suggests that. a 9 may occasionally give rise to a 9 unlike herself, because if this can happen for half of a -9 it is not unlikely that it can sometimes happen for a whole one. However, to ensure that contamination cannot occur we are in future going to separate the two kinds of 9 by I0 miles after tlmy have copulated, but before they have had a chancc to lay. A~tother possibility is that the locus concerned is in fact }~-linked but that our original stock carried a chromosome abnormality of some kind which gave rise to abnormal meiosis and the occasional production of a 9 unlike the mother Jn appearance. 9 250 Genelics o2/"Papilio glaucus and P. dardanus

Professor J. B. S. Haldane and Dr. H. Spurway have propounded a hypothesis which fits the ['acts as we know them. They suggest that the genes for black and yellow pigmentation are on the X and Y chromosomes of this species. They cross over with a fi'cquency of about 5~ (a very rough estimate). That is to say, their fiequency of recombination w;th the sex-determining gene or gene-complex is about 5%. Only the gent carried by the Y chromosome determines the colour of the butterily. This may be so for several reasons, the most probable perhaps being that it is the result of a "position eH~.ct", this being particularly likely in the sex-chromosomes where there may be a good deal of inert chromatin in the 7". If we call the gent tbr black colonr B, and its norma! allelomorph q-.., there are 7 genotypes to be considered :--

B~:( _, BX d', ~_.X ~. These ca,mot be distinguished. g2 'j' 4-'~,u BY

§ B-X 9~ do not resemble their mothers. This hypothesis can be tested. For exampIe, there are popuIations in which all the "%~'s are yellow and therefore the d):~s and 9~s must be of the constitution o~2~k,. Such females shotfld not give exceptional black daughters when crossed with males fiom a pure black or mixed stock. But some of their yellow daughters shouId do so, even if crossed to mates from the pure yellow stock. +," Similarly some yellow females fi'om the mixed stock should be B~7~" and would have occasional black daughters ever, if crossed with males fi'om the pure yellow race. Of ali the hypotheses, this seems the best to adopt for the moment as it is ca'pable of experimental investigation. The sima.don we have encountered in glaucus where no simple mendel.inn hypothesis fits the facts as we know them is by no means unique. For example, Mattingly (1956) reports that in species-crosses between the mosquitoes Aedes aegypti L. and Aedes albopictus Skuse the offspring pearly always have the morphology of the species of the mother whichever way the cross is clone, but occasionally an insect with the form of the d" parent appears in one of the later generations which result fi'om crossing the "hybrids".

DIscusSION

p.r dardanus

The genetic data confirm tile view that the difference between the mimetic tbrms is controlled by a switch mechanism which reduces the probability of getting intert mediate forms. This would be expected, tbr there woukl be strong selection agains- C. A. CLARKE AND P, M. SHEPPARD 25I

shch forms in nature. Mather (1955) has discussed this type of selection which operates when there are two or more forms which are advantageous, but when the interniediate types arc disadvantageous. He has called this selection disruptive and has pointed out that it could lead to or to speciatioa depending on conditions. Darwin (1872) suggested.a very similar mechanism, to explain the evolution of two distinct castes of sterile worker ants which can be found in the same nest. The P. dardanus data do not throw any light on whether the mimicry evolves gradually or whether it arrives at a single step, as suggested by Goldsehmidt, for no race crosses have yet been made. 'However, a cross between the form hippocoonides fi'om East Africa which mimics Anzauris niavius dominicanus with hippocoon fi-om the West coast which mimics Amauris niavius niavius wouid give the appropriate information. On Goldschmidt's view, the two mimics should differ by a pair of allelomorphs, while according to the other hypothesis the difference should be mu!tifaetorially controlled. If the non-mimetic sub-species from Madagascar, the Comoro Islands o," ti-om Somali land could be obtained, an even more conclusive answer might be arrived at. On back-crossing each of the mimetic forms to these races one should, on Fisher and Ford's view but not on Goldschmldt's, alter the effect of the gene responsible because the non-mimetic races cannot have had experience of the mimetic forms for a very long time and therefore will not be adjusted to them. If the mimetic form is altered by the process of back-crossing and the new form so produced is the same for all the mon- mimetic races, this will be good evidence that the new form is in fact close to that produced by the mutant when it first appeared. For it seems extremely unlikely that a gene-complex modifying the appearance of a mimetic tbrm in the same way will have been independently evolved in the three races in .the absence of the mutant. The presence of [.eighi in some areas shows thatl a non-mimetic form of.P. dardanus can be maintained in a polymorphic state. This raises the question of whether the evolution of a new mimetic forrn need wait for the appearance of a new mutant.. It seems possible that a mimetic polymorphism could be evolved from a non-mimetic one if suitable models became available. For example, if a protected species even remotely resembling le@hi should establish itself in South Aftira, leighi might well evolve into a mimic of that species.

P. glaucl~s

It is by 110 nleallS certain how tile black mimetic and the yellow non-mimetic female forms of P. g{a~t.cu.~ are controlled. However, it is dear that an eflScien~ switch mechauism is again in operation because intermediates are rare or absent. Moreover, the effect of the gene-cemplex on the mimetic pattern has not been ascertained because no race crosses between the southern race and the northern one, whict~ is non-mimetic, have been made in this investigation: also the emergence of the expected black fi:mates in the species cross P. glaucus • P. eur~,mcdon has (at best) been delayed. Deficiency or absence of females, the heterogametlc sex in le'pidol)tera, is in agreement with pre- dictions made by Haldane (1922). " 252 Genetics of Papilio glaucus and P. dardanus

Besides tim distribution of the black pigment itself there are odmr important eiements in the mimetic pattern, particularly the distribution of blue scales o[l the upper side and the amount ot" red in tl~e spots on tile underside of tile hind wings. These are not detm'mined by tile presence of the factor controlling black for they are present in the yellow non-mimetic tbrm of the southern race. In the northern form where there is no model the blue is much reduced. The orange is variable in both races and therefore a quantitative study would be needed to know if it were reduced ov more variable in the North. The differences could bc environmental but tile hybrids with P. glaucus and P. rululus and Y. glauczts" and P. eulymedon show that both the b].ue and the orange are modified by inherited factors. Mo,'eover, both the blue and red are reduced in tile hybrids as would be expected if they had been evolved because of their contri- bution to the mimicry, for neither P. ruiulus nor P. euo;medon is mimetic and both inhabit an area where the model P. philenor is rare or absent. The data, therefore, although not proving the point, do suggest that the main mimetic resemblance is controlled by a single switch mechanism but that the detailed resemblance has been acquired by the accumulation of other independently inherited factors. Previous data obtained by Clarke and Sheppard (1955) are reIevant in this connec- tion. They investigated the genetics of the inheritance of black in P. polyxenes by means of race and species crosses. The genetic control of mimicry in such a no~> polymorphic species has not, as far as we know, been investigated before, such experi- ments having always been confined to polymorphic species; the difficulty has been that species or race crosses have to be made in such an investigation. In P. poly,'e~zes the 9 also mimics B. phile~zor as does glaucus (see plate 2), but the male, although black, is not a good mimic on its upper side although it is on its under surtkce. The difference between the black ground colour of this species and the yellow of some of its relatives is controlled by a gent at an autosomaI locus, black being dominant. Thus here again in a non-polymorphic species the main difference between mimetic and non-mimetic is controlled by a single factor. The sexual dimorphism in P. polyxenes is controlled independently fi-om black and its presence is dominant to its absence, as crosses with P. brevicauda have shown. In the latter species the females are like the maIes and also resemble the males of P. poly- xenes in pattern (see plate 3) ; it is not certain, however, if the presence or absence of this dimorpifism is unifactorially controlled. It is interesting to note that the two yellow species P. machaon and P. zelicam, have a gene-complex producing sexual dimorphism in the presence of the black, but are themselves not dimorphic, for the gene or genes concerned have no other effect on the pattern in the absence of black. However, outcrossing to the yellow species increases the amount of ])lack in the black ~brms and disrupts the distribution of the yellow patterning on the male, so that tile distinction of the dimorphism is diminished and the males are about as dark as most P. polyxenes females. The black species living in the same areas where the model is rare or absent are all nor>mimetic and lack some or all of the subsidiary mimetic patterns mentioned previously. The details are set out in Table VI which gives some comparisons between/7, phile~,or and its three mimics, P. gIaucto', P. lroihes and P. pol.rxenes C. A. CLARKE AND P. M. SHEPPAIZD 253

together with a number of tile non-mimetic relatives of P, polyxenes and P. glaucus (see plate 2). It will be seen that in P. polyxenes not only is the presence of black controlled by a single factor but there is an increase in the amount of blue compared with most of its non-mimetic relatives. Our breeding data show that this is geneticaIly controlled, is probably multifactorial, and is independent of" the black locus. The distinct sexual dimorphism is also genetically controlled and tends to be reduced on out-crossing to non-mimetic relatives. Thus the mimetic patterning of the black pigment is also genetically controlled. P. polyxenes has more orange in the underside of the hindwing than all its relatives except brevicauda and again this character is muhifactorially controlled and increases tim mimicry. It will thus be seen that the mimetic pattern is built up as a result of a large number of genetic substitutions as compared with non- mimetic relatives and therefore could not have arisen at a single step. It is true that the all-important difference, namely, the distribution of black, is unifactorially con- trolted and this alone can produce good mimicry at a single step, provided the gene-

Table VI. Some comparisons between the model B. philenor, its mimics and a number of the non-mimetic relatives of P. polyxenes and P. glaucus

Degree of Ground Blue Marked sexual blackness in.9. Orange Colour99 pattern on dimorphism (on which spots on Black-'- + hindwing in black depends the hindwing Yellow = -- pattern mimetic resemblance) ModeI B. philenor 9 + ++ ++ -- ++-F+ +++ NIimics P. glaucus 9 + or -- -t- + + + + + + + + + + -- P. troilus ? + + + + -- ++++ +++ P. po(~,x~l~cs _9. + + + + + + +++ +++ Non-mimetic relatives of P..polyxene:s

P. nitta c, + ,~ __ + + + to + P. brez&auda o -}- + + -- + + + + ..k P. bah'dff ~ + or -- _,1__1_ + + + + + + q- P. L,z&a ~. + + -- -I- P. rudkini ~ + or -- ? ? yellow or ? ? P. machaon .c.? "- -b-I- N.A. yellow + to -- P. zdicaon ~ -- + N.A. yctlow + to -- P. or,'gonia ~? -- -I-+ N.A. yellow + to -- Non-mimetic relatives of P. g[auncr P. rulutus .9. -- + N.A, yellow + to- P. eloymcclon c.f. -- + N.A. paI~: yellow -I- to -- P. mullicaudatles 9 -- -}- N,A. yellow .q,- to -- 254 Genelics of Papilio glaucus and P. dardanus complex with respect to sexual dimorphism is appropriate. When the n-todel is mainly black with very little pattern, one might well expect that any gent producing a great deal of black will produce to[erablc mimicry. It is therelbre not surprising that very littIe modification is needed. The data therefore wouId not be expected to contradict Goldschmidt's hypothesis as firmly as they might have done but it seems !ikely that in a more complicated pattern such as that found in dardanzls, more extensive modifica- tion will have occurred during the evolution of the mimicry. In the last four columns of the table the pluses indicate the degree to which the character is expressed. A minus indicates its absence & N.A. (not applicable) signifies that the character cannot express itself. We can there~bre conclude with some certainty that the mimetic patterll is built up h'om a large number of independently inherited characters and therefore could not have appeared at one step. The only way the pattern could have appeared at one step is for the subsidiary characters, which improve the mimicry, to Ilave been perfected before the allelomorph controlling bIack arose. This seems very unlikely in view of the distribution of these characters among the non-mimetic relatives of P. polyxe~zes, and the case against it is Further strengthened by a consideration of the non-mlmetic relatives of P. g[auctts. Here again the subsidiary characters of blue and orange are much reduced in P. rutulus, P. e,trymedon and P. multicaudatus.

SUMMARY

1. Genetic investigations using the hand-mating technique show that in Papilio dardanus fi'om South Afi'ica the mimetic form cenea is almost certainly dominant to the mimic hippocoonides. The form leighi which has no model but which is tailless and not like the male is also dominant to hippocoonides. Form leighi is either dominant to or has an epistatic effect inhibiting the expression ofcenea. 2. In the North American swallowtail P. glaucus there are two female Forms, one yellow like the male and ~he other black and mimetic. The difference is not controIIed by an autosomal gent with black dominant and sex-controlled as has been suggested. The results show that the Female offspring are nearly always of the same Form as the mother regardless of the origin of the male. However, this cannot be due to total or ordinary partial sex-linkage because on rare occasions (six out of fifty-four completed families) segregation occurs. But even in these the ratio is not a simple mendelian one and the tbrm unlike the mother (whether this be yellow or black) is always the rarer. Of several hypotheses one is selected for special discussion. 3. There are two reasonable hypotheses concerning the genetic origin of" mimicry: i) that it arises by mutation Fully formed at a single step. 2) that a partially mimetic form appears by mutation and that its mimetic pattern is improved by selection of a suitable gene-complex. Race and species crosses can decide between these two hypotheses. 4. ;Yhe present investigations in dardanus are concerned with obtaining genetic information in preparation for making race crosses. This in the main is true for the C. A. CLAr~KJa AND P. M. SI-IE~PPAI~.D 255

'work with P. glaucus (a mimic of B. philenor) although here two species crosses, one between P. glaucus and P. rululus and the other between P. glaucus and P. euo,me&, n also have been made. The results from these kybrids suggest that some minor pattern differences, notably the blue on the upper smface of the hindwing and the orange spots on its under surface, (essential constituents of the mimetic pattern) are indepen- dently inherited from the black. These, in particular the blue, have probably been intensified in the southern (mimetic) race of P. glaucus as a result of natural selection. 5. Previous work with P. polyxenes which also mimics B. philenor is reviewed, and suggests even more strongly that Small differences in pattern with respect to blue and orange (and also in this mimic with respect to sexual dimorphism) have been evolved independently of the black pattern and result in an improvement of the mimicry in the female. In contrast to these adjustments the main resemblance probably arose at one step by mutation involving a change fi-om a mainly yeIlow to a mainly black pattern, black being dominant to yellow.

ACKNOWLEDGMENTS

We are very gratefi.tl to ProFessor J. B. S. Haldane, F.R.S. and Dr. H. Spurway for their help in the interpretation of some of the data and for suggesting the calculations made in Tables III& IV'. We are also indebted to Mr. and l~Irs. Lincoln Brower for commenting on the manus- cript and 1"or giving us first-hand information about the American species. We wish to thank Dr. E. B. Ford, F.R.S. and Dr. Julian Huxley, F.R.S. For reading the manuscript and for making helpful suggestions. The work has been greatly aided by the cogperation of Mr. T. G. Howarth who kindly allowed us to examine and photograph specimens in the possession of the British Museum (Natural History). One of us (P.M.S.) is grateful to the Nuffield Foundation for its support during part of the investigations and to the Rockefeller Foundation whose generosity enabled him to collect material in America in 1955. YVe would also like to thank Mr. N. McFarland who ldndlv sent us pupae of P. euo,- medon and Mr. E. Dluhy for sending us pupae ofP. glaucus and P. ruluhts.

REFERENCES

B~.a'l~S, H. W. (18621. Trans. Linn. Soc. Lettd. 23, 495-566. C~k~PENTt~R, G. D. ]-I. & Fo~tD. E. B. (1933). Minficry, Mcthuen, London. C~.ARm~, C. A. & S~.l~m,:,Rr~, P. M. (1955). The breeding in captivity of the hybrid Papilio ~utulus female "< Papilio glaucus maIe. Lc]~id.Ncws, 9, 46-48. CLaret.E.C.A. & SHv,rn,z~m:), P. M. (1955). A preliminary reporl on lhc gmletics of the machao1~group of swallowtaiI butterflies. Evolution., 9, 182 201. (-',LA~U

DA~,,v~r,r, C. R. (1872). The Origin of Species (6tit edition), John Murray, London, 230--233. Eow,xl~as, W. H. (l[18,1-). Butterflies of North America, second series, IIoughton Miffin & Co. Fist-lEa, R. A. (1930). The genetical theory of natural selection. Clarendon Press, Oxford. FORD, E. B. (1936}. The genetics ofPapitio dardanus Brown (Lep.). Tra,as. R. Ent. Soe. Lund., 85, part 19, 435-466. Foa~, E. B. (I937). Problems of heredit) in the . Biol. Rec., 12, 461--503. FORD, E. B. (1953). The genetics of polymorphism in the tepidoptera. .,ldval~cesin genelics., 5, 4-3 87. GocDscrt~,nD'r, R. B. (I945). Mimetic polymorphism, a controversial chapter of Darwinism. .Odwrt. Rev. Biol., 20, 147-I64, 205-230. HALD.~,t,;e,J. I3. S. (1922). Sex ratio and uni-sexual sterility in hybrid . J. Gelzet., 12, 101 109. MATrtER, K. (1955). Polymorphism as an outcome ofdisruptl.ve selection. Evolldion.~ 9, 52 61. M.~,'rTINOLV, P. F. (1956). Species hybrids in mosquitoes. Trm2s. R. Eat. Soc. Loml., 108, part 2, 21-36. MtiLLEr~, F. (i879). Cosmos., 5, i00. POULTON, E. B. (1928). Pcoc. Eat. Soe. Lond., 3, 16-17. PUNN~'r'r, P,-. C. (1915). Mimicry in Butterflies, Cambridge University Press. RE_.',tlsc'roN, C. L. (t954.). Advances in genetics., 6, 403-=1-50. Ro'rHSCmLD: W. & JoaD*N, K. (i906). A revision of the American Papi!ios. In .iV'o~itceles Zoologicae. London, HazeIl, Watson & Viney., 13., 411-752. C, A. CLARKE AND P. M. S~-~r,P?.r

TEXT ]>I.A'l h:

~Ialc 1 shows the I.~rinripaI 1-~ m~ oF P. ctar,'.lanu~ di.~cu~cd in the paper: ]. P. &O&lnUS ~. 2. P. c/aJclanzzy nero-mimetic ~r! 3. P. da~&tnles'li' f. /li/,poco,nidc.~ q. P. darda.,llL', !:! f. ce:l,'a 5. P. r 7 l'. O@h,mi~L~ G. P. &;,V~'Hn~s 'L l. lciA;h/. Ge~:cl:cs of Pa/~i/ G/oa:u.r a;zd P. Da:da:~z~,r

TEXT PLATE 9 Plate '2_ gives the model and the mimetic species discussed in the paper. In order to show features of the mimicry, on ll~e left of each photog~aph is shown the ~.lpperside and on the right the un,:lefside of the butterfly : 1. B. pkilcn,~r % 2. P. troilus 9 3. P. pol2.x:~ % %. P. .o~/aucus 9, b[ack form. 5. P. ghzucus 9, mutant form. 6. P. glaucus :i', yellow* form. C. A. CL.aaK~ aND P. M. SHEPPARD 259

TEXT Pt.:,'r~- 3

Plate .3 Ri~cs the uppcrside oF tim males of the f;lur species whose [-cmales arc ( ep ctcd on Plate I. 1. B. pl~itrrtor. 2. P. Iroilu.r 3. P. po!l,x~',lc,~ 4. P. .~'ioac~r