Genetic Differences in Mating Success and Female Choice in Seaweed Flies (Coelopa Frigida)

Heredity 62 (1989) 123—131 The Genetical Society of Great Britain Received 13 June 1988

Genetic differences in mating success and female choice in seaweed flies (Coelopa frigida)

G. Engeihard, Department of Genetics, University of Nottingham, S. P. Foster and Nottingham NG7 2UH, U.K. T. H. Day

An association is described in seaweed flies, Coelopa frigida, between the genotype at the alcohol dehydrogenase (Adh) locus and mating success in pairwise mating trials. Significantly higher mating success was observed in females that carried the Adh-C allele, but no association was observed between Adh genotype and male mating success. There was heterogeneity in the success of different combinations of males and females, but only when the female lacked the C allele. Analyses of video recordings indicated that C-bearing females spent longer mounted by males and that they less frequently rejected males. Evidence is presented for mate discrimination by females not carrying a C allele. The significance of there being genetic differences in both mating success and in female discrimination are discussed in the context of previous results on mating behaviour in natural populations.

NTRODUCTION on this latter point are experiments showing that the exercise of choice results in increased repro- Darwin'stheory of sexual selection (Darwin, 1871) ductive success in zebra finches (Burley, 1986), proposed that during the mating process males tungara frogs (Ryan, 1983) and seaweed flies compete with one another for females, and females (Crocker and Day, 1987). Mate choice may also choose between the available males. The evidence lead to the production of progeny better able to for male-male competition is convincing in a wide survive under competitive conditions as in fruit diversity of species, but it is only during the last flies (Partridge, 1980) and seaweed flies (Crocker decade that equivalent evidence has been found and Day, 1987). The experimental support for for female choice. Female widowbirds (Andersson, female choice being inherited is more limited. Only 1982), bowerbirds (Borgia eta!., 1985), sedge in the two-spot ladybird beetle, Adalia bipunctata, warbiers (Catchpole, 1987), zebra finches (Burley, does direct evidence exist for a polymorphism with 1986) and pied flycatchers (Read, 1986) all exercise a genetic basis (Majerus et a!., 1982, 1986; choice on the basis of some male attribute. O'Donald and Majerus, 1985; Majerus, 1986), Examples from other animal groups include although convincing indirect evidence has been sticklebacks (Semler, 1971), frogs (Ryan, 1983) obtained in Drosophila melanogaster (Heisler, and several species of insects (see Thornhill and 1984) and the Trinidad guppy, Poecilia reticulata Alcock, 1983). (Breden and Stoner, 1987). We report here an The demonstration that female choice occurs association in seaweed flies between mating suc- is only the first step in understanding how sexual cess, female choice and the possession of a par- selection operates over evolutionary time. This ticular allele at an enzyme-determining locus. problem has been tackled from a theoretical point All populations of the seaweed fly, Coelopa of view in the models of Fisher (1930), O'Donald frigida, are polymorphic for a large inversion on (1980), Lande (1981) and Kirkpatrick (1982). It is chromosome I (Butlin et a!., 1982a; Day et al., clear that for female choice to evolve and be 1983). This chromosomal rearrangement, which maintained, both the preference itself and the pre- involves over 10 per cent of the genome (Aziz, ferred character should be inherited. Furthermore, 1975; Day et a!., 1982), has profound influences there should be some advantage to females being on the flies' biology. The two alternative forms (a choosy which may be manifest as increased fecun- and/3) are associated with differences in develop- dity, or greater survival of the progeny. Bearing ment time (Day eta!.,1980), adult size (Butlin et 124 G. ENGELHARD, S. P. FOSTER AND T. H. DAY al., 1982b), adult longevity (Butlin and Day. 1985) tion was necessary prior to video trials lasting only and larval survival (Butlin et aL, 1984). There is 45 mm. Checks for virginity indicated that in no also differential male mating success between case had any female been fertilized before the start genotypes (Butlin et a!., 1982b; Day et a!., 1987) of trials. and populations exhibit negative assortative mat- The experimental procedures for pairwise mating with respect to the af3inversion(Day and ing trials and for the video recording of mating Butlin, 1987). A recent analysis of mating (Crocker behaviour were those of Crocker and Day (1987) and Day, 1987) suggested that female seaweed flies and Day ei a!. (1988), with the minor modification might be choosing their mates on the basis of that one female and three males were observed in inversion genotype, and that there might, in addi- each trial. Males were distinguishably marked with tion, be an element of male choice. small spots of paint applied to the dorsal surface This study by Crocker and Day used a labora- of the thorax. Just over 100 trials were conducted tory population that had been intentionally bred each using different animals. On the completion to have reduced genetic variation. In particular, of trials, animals were stored at —25°C and sub- the population only included individuals possess- sequently scored for their genotype at the alcohol ing B and D alleles at the alcohol dehydrogenase dehydrogenase (Adh) locus using starch gel elec- locus (Adh). This locus is known to be associated trophoresis (Butlin et a!., 1982a). with the cr/3 inversion system such that BB homozygotes are always aa homokaryotypes, and DD's are always /3/3's (Day et a!., 1982). A third RESULTS allele, Adh-C, is also commonly found in natural Mating success in pair trials populations. Preliminary mating tests involving females carrying at least one C allele suggested Virgin flies were paired at random and the mating that in these cases mating was random. This admit- success of each pair assessed by whether or not ted the possibility that BB, BD and DD animals they produced larvae. After the trial the Adh were mating discriminately, but that BC, CC, and genotypes of both adults were determined. The CD females did not exhibit mate choice. Here we mating success of males, regardless of the females report evidence in support of this interpretation, with which they were paired, revealed no and suggest that, as in ladybirds, there are genetic heterogeneity between genotypes (table 1). In con- differences in female choice. The data derive from trast there were differences in the mating success pairwise mating trials in which the results of of females. B(, CC and CD females were the most Crocker and Day, (1987) are extended to include successful, and if the data from these three the Adh-C allele. Observations of mating genotypes are pooled, there is a highly significant behaviour provide direct evidence that females not difference between the C-bearing genotypes and carrying the Adh-C allele mate in a discriminating the combined non-C's (x=1171;P<0001). A fashion. similar comparison of C and non-C males reveals no difference (x= 048;P =049).It appears that METHODS possession of an Adh-C allele is associated with greater mating success in females, but not males. Animalswere collected from a natural population Is the success of males (or females) dependent at St. Mary's Island on the north-east coast of on the genotypes of the two animals considered England and were used in experiments within two together? The success of each male genotype gen- generations of being collected. The maintenance erally did not vary with differing females, although of cultures in the laboratory has been described DD males were an exception to this (table 2). by Day and Buckley (1980). Adults were collected When male C's were pooled and non-C's were shortly after eclosion and males and females stored pooled, the two sets of data were similar. The separately at 4°C. When required for pair trials reciprocal comparison yielded a strikingly different flies were transferred to fresh seaweed (Fucus ser- result. While the success of B(, CD and CC ratus and F vesiculosus) at 29°C overnight. For femalesdid not vary with the male genotype, there video recording, animals were left for a further was the strong suggestion of heterogeneity between two days before use. Preliminary experiments BB, BD and DD females. Furthermore, showed that sex-isolation for one day resulted examination of the pooled data indicates that suc- in the appropriate level of sexual activity in 5-h cess was homogeneous with C-bearing genotypes, mating trials (appropriate, that is, for analytical but heterogeneous with non-C females. Although purposes), but that a longer period of sex-starva- non-C females are convincingly heterogeneous, GENETIC DIFFERENCES IN FEMALE MATE CHOICE 125

Table IMating success of different genotypes in pair trials

Per cent mating success Adh genotype

BB BC BD CC CD DD P

Males 651 571 583 438 567 566 331 065 No. of trials (83) (42) (503) (16) (90) (228) Females 610 648 538 692 746 521 1541 0009 No. of trials (105) (91) (513) (13) (63) (163) x2 values (in this and all subsequent tests) were calculated from the original numbers of succesful and unsuccessful pairs.

Table 2Mating success of different genotypic combinations of males and females

Per cent mating success of males when paired with females of the following genotypes

Male genotype BD BB DD BC CD CC df P

BD 571(261) 500 (58) 532 (77) 537 (41) 636 (22) 667 (6) 178 4 078 BB 568(37) 774 (31) 600 (10) —(0) —(2) —(0) 3.33 2 019 DD 513(119) 563 (16) 326 (43) 786 (14) 750 (7) —(1) lOll 3 0018 BC 556 (18) —(0) [667 (6) —(1) —(2) —(0)] FET* 090 CD 353(34) —(1) —(4) 571(14) —(3) —(1) 195 1 016 CC 200(10) —-(0) [—(1) —(0) —(0) —(0)] FET* 055 Non-C's 554 (417) 590 (105) 469 (130) 600 (55) 613 (31) 571 (7) 520 4 027 C's 387(62) —(1) 545(11) 600(15) —(5) —(1) 275 2 025

Per cent mating success of females whenpaired with males of the following genotypes

Femalegenotype BD BB DD BC CD CC df P

BD 571(261) 568 (37) 513 (119) 556 (18) 353 (34) 200 (10) 1080 5 0056 BB 500(58) 800 (30) 563 (16) —(1) —(1) —(0) 748 2 0024 DD 532((7) 60.0(10) 326(43) 667(6) 250(4) —(1) 549 2 0064 BC 537(41) —(1) 786(14) —(1) 571(14) —(0) 272 2 026 CD 636 (22) [—(2) 429 (7) —(2) —(3) —(0)] FET* 081 CC 714(7) —(0) [—(1) —(0) —(1) —(0)] FET* 033 Non-C's 553 (396) 662 (77) 472(178)560 (25) 359 (39) 273(11) 1626 5 0006 C's 586 (70) —(3) 636 (22) —(3) 611 (18) —(0) 019 2 091

* Whensample sizes (given in round brackets) were very low, trials have been pooled as indicated by square brackets, and a Fisher's exact test (FET) performed on the data. the statistical test for homogeneity of C-females remain mounted long enough for insemination. In is not very powerful, since a large proportion of order to study various aspects of mounting, trials these females were paired with a single male were conducted with a single female and three genotype (BD).Theseresults should not, there- males. The behaviour of animals was recorded for fore, be taken as strong evidence for the mating 45 minutes and the interactions quantified during success of C-females being independent of male slow play-back of the tapes. It should be noted genotype. In passing we may note that the patterns throughout this section that none of the data were of non-randomness shown by BB, BD and DD normally distributed and neither did any transfor- females do not appear to be the same, a point that mation render them so. The data have therefore will be considered in a later section. been analysed non-parametrically using either the Mann-Whitney U-test, from which the normal Theopportunity for insemination deviate was computed, or the Kruskal—Wallis test in which H is distributed as x2withfive degrees Clearly,before a pair of flies can achieve a success- of freedom (since six genotypes were being com- ful mating, the male must mount the female and pared). 126 G. ENGELHARD, S. P. FOSTER AND T. H. DAY

Table3 Comparison of female genotypes with respect to various aspects of mounting

Pooled Pooled Female genotype C's non-C's (BC CC(BB, BD BB BC BD CC CD DD and CD)and DD) Mean time to first mount (minutes) 26 24 17 39 08 38 24 29 (s.c.) (13) (18) (05) (24) (03) (20) (10) (0-9) Mean number of mountings per female 180 301 167 127 293 169 248 169 (s.c.) (39) (59) (18) (4.9) (51) (18) (35) (12) Mean duration of mounts (seconds) 289 375 376 567 258 352 369 361 (s.c.) (63) (48) (29) (7.7) (44) (43) (34) (2.2) Number of mountings 14 22 85 10 19 48 51 147 Per cent of trial spent 288 377 237 325 298 206 338 232 mounted (s.c.) (111) (59) 032) (85) (5.4) (3.4) (36) (2.4) Number of females 5 10 35 7 7 20 24 50

The data are presented as means averaged over all females of a given genotype. For the calculation of mounting duration, mountings followed by an immediate female rejection or male dismount were excluded. In this case means derive from averages of extended mountings of a given female genotype.

The time elapsed from the beginning of the the mean duration of each mount was calculated. trial to the first mounting was measured (see table In this case there are significant differences 3). There were no significant differences between between genotypes (H =1208;P =0.034),but the any of the female genotypes, nor were the pooled differences are not associated with the possession C's different from the pooled non-C's (U=532, of a C allele (comparing C's with non-C's gives d =079;P =043).However, the vast majority of U=3576, d=049; P=062). The analysis sug- females were mounted within the first 60 seconds gests that CC's are mounted for long duration, of trials, a rapidity that is probably a consequence whereas the mountings of BB and CD females of the long period of sex-starvation to which all tend to be short. animals were subjected. If the animals had been The combined effects of all these aspects of less sex-starved it is possible that differences in mounting were explored by calculating the total mounting time might have been observed (but far time spent by females with a male mounted upon fewer data on mountings would have been avail- them. The three C-bearing genotypes all spent a able for analysis). larger proportion of the trials mounted than did There were significant differences in the mean BB's, BD's or DD's (see table 3), and when com- number of times each female was mounted (H = bined, there is a highly significant difference l2l7;P=0033), but only a small difference between C's, and non-C's (U=418, d=259; P< between C"s and non-C's (U=560, d=177; P 001). 0009). Nevertheless, there is a strong suggestion The conclusion from this analysis is that there that BC's and CD's were mounted more frequently may well be greater opportunities for insemination than other genotypes. This result needs to be inter- of C-bearing females compared with non-C's. preted with care since many mounts resulted in an Furthermore, we tentatively suggest that the three instant dismount by the male, thereby making the C genotypes gain their opportunities in slightly female available for almost immediate re- different ways. At one extreme CD females are mounting. In contrast, a few mounts were of mounted with little delay, subsequent mountings exceedingly long duration—indeed most of the occur frequently but each is of short duration. In trial—so that it was impossible for that female to contrast CC individuals are mounted more slowly be mounted many times. and less often, but each mounting is of long dur- For those mountings not terminated by ation. These three attributes combine in their immediate male dismounting nor female rejection, various ways to result in all three C genotypes GENETIC DIFFERENCES IN FEMALE MATE CHOICE 127 being mounted for about 15 minutes of the 45 enquired whether any relationship exists between minute trial. BB's, BD's and particularly DD'sare the Adh locus and female rejection, and if so, mounted for a distinctly shorter time. whether it could contribute to the pattern of mating Similar analyses were also performed on the success observed in pair trials. six male genotypes but the data are not presented When all mountings were considered (table 4), in detail here. Suffice it to report that no differences there were no differences between female were observed in the times to first mount, total genotypes in their rejection rate (H=675; P number of mounts, mean duration of each mount, 0.24), although DD's showed a distinctly higher or in the total time spent mounted. This lack of rate of rejection than other genotypes. However, any association between male genotype and a review of the original data suggested that a few various aspects of mounting is consistent with the individual animals were contributing dispropor- similarity in mating success of males observed in tionately to the number of mounts (males) and the pair trials. number of rejections (females). Furthermore, the behaviour of many females changed during the Malerejection of females trial (see below). In an attempt to remove these effects, only the first interactions between each Afrequent consequence of a mounting is that the male and female were scored. The data (table 4) male dismounts with no apparent female rejection reveal large differences between females (x = response. If for some reason certain females are 173;P=0004), with BC females rejecting half more "attractive" to males, those females should as often, and DD'srejectingtwice as often, as the be rejected less often. There are in fact no differen- remaining genotypes. Overall, the non- C-bearing ces between female genotypes in this respect (H = genotypes rejected about twice as often as C's 323; P=0.67),although BB and BC females were (x=63; P=0012), although this difference is rejected slightly less frequently than other largely due to DD females. genotypes. There were also no differences in the Having demonstrated differences in the willing- rejection rate by different male genotypes (H = ness to mate, we may now ask if some, but not all, 46;P=0.47). There appears to be no association females are also "fussy", or discriminating in between genetic variation at the Adh locus and which males they reject. If females are classified male rejection. as consistent rejectors—rejecting each different male who mounted—consistent acceptors— Femalerejection of males accepting every male who mounted—or mixed rejectors/acceptors (i.e., possible discriminators), Thefemale rejection response involving kicking, a further difference between genotypes is revealed. wing lifting, abdominal curling and eventually One third of C genotypes (8/24) exhibited mate lateral rolling of the complete body has been discrimination while over a half(29/54) of non-C's described elsewhere (Day et a!., 1988). We now were discriminating. Collectively BB, BD and DD

Table 4 Relationships between female genotype and female rejection

Per cent of total Per cent of initial Number of females exhibiting mounts ending in mounts ending in female rejection female rejection Consistent Some rejections, Consistent Female genotype (se.) (number of females) rejection some acceptances acceptance

BB 26(17) 308(13) 0 3 2 BC 25(14) 12.0(25) 0 4 6 BD 46(12) 28.6 (84) 3 16 11 CC 2.6(2.1) 28.6(14) 0 2 5 CD 2.8(2.2) 22.7 (22) 0 2 5 DD 143 (4.2) 53-8 (52) 5 10 4 Pooled C's 26 (1.0) 197 (61) 0 8 16 Pooled Non-C's 77 (17) 376 (149) 8 29 17 Tests for heterogeneity between C's andNon'C's d= 142 63 =90 P=0-16 P=0012 P=0007 128 G. ENGELHARD, S. P. FOSTER AND T. H. DAY females were significantly more choosy than mating, and began rejecting males she had earlier females carrying a C allele (table 4). accepted (e.g.,trial41). Is there any evidence that the pattern of dis- Trials 83, 49 and 41 involved non-C bearing crimination varies between female genotypes? In females who showed both acceptance and rejec- this further breakdown of the data, the only infor- tion. While most C-bearing females were con- mative genotypes were BD and DD—both puta- sistent acceptors, a few did reject as well. Trials tive discriminators. BD females rejected 33 per 21 and 100 are two examples. In these, and the cent (sample size 18) of BD males, but 75 per cent other six similar trials, individual males are not (8) of DD males. In contrast, DD females rejected repeatedly rejected or accepted. In other words 71 per cent (7) of BD males but none of the three acceptance or rejection does not seem to be related DD males. In spite of the small sample sizes there to the males' genotype. We have tested more for- is significant heterogeneity in rejection (x= 82; mally the proposition that C females who both P=0042: Lewontin and Felsenstein (1965) have accept and reject are doing so regardless of which shown that this test for heterogeneity is valid even male has mounted, whereas non-C females are with exceedingly low expectations). It appears, genuinely discriminating between males. 2x3 then, that at least BD and DD females are dis- tables were produced for each trial summing the criminating in who they reject, and if this rejection number of acceptances and rejections for each of influences mating success, it predicts there should the three males. For example, trial 83 yields the be positively assortative mating with respect to the table: Adh locus. This evidence for female discrimination comes MaleI Male 2 Male 3 from considering only the initial mount by each Acceptances 5 6 0 male. If the sequence of rejections and acceptances 0 0 5 during the course of trials is examined, further Rejections indication of discrimination is apparent. In table whereas fortrial21: 5,detailsof five such sequences are shown. in trial 83 there is clear evidence that the BB female repeatedly accepted the CC and CD males but Male1 Male 2 Male 3 rejected the BD male. Such consistent behaviour Acceptances 2 3 3 was not typical of most females. It was more com- Rejections 3 2 5 mon for a female initially to reject a male but subsequently accept him (e.g.,trial49). The gener- It is obvious that the statistical heterogeneity in ality of this phenomenon is seen in the overall the first contingency table is far greater than in the rejection rate being very much lower than the rate second. For each of the eight such tables from C of rejection following initial mounts (table 4). females, and 19 for non-C females, the probability Occasionally the female seemed to get fed up with of obtaining such a set of data was calculated using

Table S Sequence of female rejection/acceptance in individual trials

Trial 83 female: BB;male1: CC;male2: CD;male3: BD Ia,3r 2a, Ia, 3r, 2a, 2a, 225, 3r, 3r, Ia, Ia, 2a, Ia, 3r, 2a. Trial 49 female: DD;male1: BD;male2: Bil;male3: DC 2r,3a, 2a, 2a, 2r, Ir, Ir, 2a, Ir, 3a. Trial 41 female: BD;male1: DD;male2: CD;male3: B!) Ia,3a, 3a, 3a, 3a, 2r, 3a, 3a, 2r, 3r, lr, 3a, 2r, 3a. Trial 21 female: CD; male I: BC;male2: BC; male3: (!) 3a, Ia, Ia, 2a, 2r, 3r, 2r, lr, lr, 3r, 3r, 2a, 3a, 2a, 3r, 3a, Ir, 3r. Trial 100 female: BC;male1: BB;male2: B!);male3: CD Ia,3r, 3a 3r, 3r, 3a, 3r,3r,3a,3a, 3r, 3r, 3r, 3r, 3a, 3a, 3r, Ir, 3r, 3r, 2r, 2r, 3r, 3a, 3r, 2r, 3a, 2a, 2a, 2r, 2a, 2r, 2r, 3a, 3a, Ia, 2a. Rejection is represented by r", and acceptance by 'a". Trial 83 began with male I being accepted, then male 3 was rejected, then male 2 accepted and so on. Male dismounts have been omitted from these sequences. GENETIC DIFFERENCES IN FEMALE MATE CHOICE 129

Fisher's exact test. It must be pointed out that most preference and preferred character may have been of the trials were not nearly as extreme as the two of significance during the evolution of female examples given above. Nevertheless, when the choice. probabilities are combined, using the method of There is still some doubt over the tightness of Fisher (1934), none of the C genotypes is this association. While recombination is exceed- heterogeneous (BC:x=22, P=O98;CD: ingly low between the a and /3 forms of the inver- 17,P=079; CC: x=06, P=073),whereasthe sion, Adh-C, as identified by starch gel electro- non-C's are (BB: x=206, P=0002; BD: x= phoresis, can be found on either inversion (Day 260, P=O05; DD: x6=522, P<<0001). The et al., 1982). It is clearly of consequence to know combined heterogeneity with non-C females is whether one or both inversion sequences carry the very considerably greater than with C's (non-C's: allele for non-preference. Whatever the outcome, X38=987, P0.00l; C's: P=099). one could expect the genes responsible for female This provides yet more evidence that for those preference to recombine with at least one of the females exhibiting mixed acceptance/rejection inversion types. Why then do we observe associ- behaviour, the non-C's are genuinely discriminat- ation between the loci for Adh and preference? ing, whereas the C's are not. Possibly the two loci are exceedingly tightly linked, We consider the critical observation in many or there may be selective reasons why these, and of the trials to be that females rejected and accep- perhaps other loci, should remain in linkage dis- ted males in successive mounts; they were phy- equilibrium. siologically able to carry out both types of Regardless of the solution to this genetical response. For C-females this behaviour was ran- problem, we can make a prediction concerning dom, but for other genotypes it was highly dis- behaviour. If the non-preference allele is located criminating. It seems as though both willingness on only one of the inversions, it should be possible to mate, and lack of mate discrimination is confer- to make a much more clear-cut distinction between red by the C allele, or by alleles in coupling at choosy and non-choosy females. The category of closely linked loci. non-dicriminators as presently identified may well include females with both types of behaviour. The genetic determination of female choice is being DISCUSSION studied further. Consider next the patterns of mating success. Itis widely assumed that female choice is geneti- Several factors, both genetic and non-genetic are cally determined yet direct experimental support known to be associated with mating success in only exists for Adalia bipunctata (Majerus et a!., seaweed flies. These include female choice on the 1986). In the two-spot ladybird a single dominant basis of the af3 inversion, male choice on the basis gene appears to be responsible for female prefer- of size (of which this same inversion is a major ence for melanic males, with homozygous recessive determinant), male-male competition and the females mating randomly. Preference for non- forced mating of unwilling females (i.e., rape). In melanic males has not been observed. The results such a complex situation it is to be expected that reported here provide further evidence that female a complete understanding of mating will not choice has a genetic basis. Seaweed flies possessing readily be achieved. The results presented here are an Adh-C allele exhibit no mate discrimination not obviously consistent with previous data. For whereas other genotypes appear to exercise mate example, the least successful males in conditions choice. We are still a long way from identifying of mass mating were Adh-BB homozygotes, and the gene or genes responsible for this behaviour, Adh-CD's were startlingly successful (Day et a!, but there is currently no reason at all to believe 1987).In contrast, pair trials conducted in small the alcohol dehydrogenase locus itself is crucial. enclosures revealed no significant differences in The Adh locus is associated with a large chromo- male success, and if anything, BB's were mar- somal inversion and any of the 200 or more genes ginally more successful than other genotypes (see in this inversion could be responsible. What does Table 1). In mass matings female success wa:s appear to be true, however, is that whatever genes homogeneous whereas heterogeneity was observed determine female choice they are genetically linked between genotypes among the results reported to the genes determining the male character being here. Furthermore, the pattern of mating was pre- chosen—namely genotype at the Adh locus, or dominantly negatively assortative in the results of loci closely linked to it. Such association between Day and Butlin (1987) whereas in the present 130 G. ENGELHARD, S. P. FOSTER AND T. H. DAY experiments the flies tended to exhibit positive Acknowledgements We are grateful to the N.E.R.C. for their assortment. financial support (Grant No. GR3/4931A to T. H. Day), to Dr At the moment we can only speculate on the D. T. Parkin for advice on 2 x 3 Fisher's exact tests, and to Drs R. K. Butlin and F. S. Gilbert for critically reading the manu- reasons for these apparent discrepancies. Clearly the mating conditions were not identical. In pair script. trials male—male competition is absent and females may find it difficult to exercise choice when closeted with a large male. There were also differences in the levels of inbreeding, genotypic frequencies, sex starvation and encounter rates, all REFERENCES of which could affect mating success. We certainly believe the previous findings of negative assort- ANDERSSON,M. (1982). Female choice selects for extreme tail ment to be valid and indeed R. K. Butlin (personal length in a widowbird. Nature, 299;818-820. AZIZ,j. o.(1975). Investigations into chromosomes 1, 2 and 3 communication) has pointed out that there are of Coelopa frigida. Ph.D. Thesis, University of Newcastle several indications that it is also operating under upon Tyne, U.K. the experimental conditions used in the present I3ORGIA, 0, PRUFVT-JONES, S.D. AND PRVETT-JONES, M. A. study. The problem will be to disentangle the (1985). The evolution of bower-building and the assess- ment of male quality. Z. Tierpsvchol., 67, 225—236. various components of mating behaviour and BREDEN, F. AN 1) STONER, c. (1987). Male predation risk deter- assess their genetic and evolutionary effects. We mines female preference in the Trinidad guppy. Nature, should not be surprised if it is the balance between 329, 83 1-833. female choice and male choice that largely deter- BURLEY, N. (1986). Sexual selection for aesthetic traits in mines the pattern of mating success. Needless to species with biparental care. Am. Nat., 127, 415—455. BUTLIN, R. K. ANt) DAY,T.H.(1984).The effect of larval say, we are actively pursuing this problem. competition on development time and adult size in the There is evidence that the inversion poly- seaweed fly, Coelopa frigida. Oecologia, 63, 122-127. morphism has been stable in natural populations BUTLIN. R. K. AND DAY, 1. i-i. (1985). Adult size, longevity and for almost two decades (Kelsey, 1969), and the fecundity in the seaweed fly, Coelopa frigido. Heredity, 54, uniformity in inversion frequencies over northern 107—110. BUTLIN, R. K.. COllINS .M.,SKEVINGTON, S. J. ANI) DAY, Europe suggests the polymorphism may be of con- 1. H. (1982o). Genetic variation at the alcohol dehydro- siderable antiquity (Butlin et a!., 1982; Day et a!., genase locus in natural populations of the seaweed fly, 1983). What are the selective forces maintaining Coelopa frigida. Heredity, 48, 45—55. I3UTLIN, N. K., REAl),I.L. AND DAY,f. H. (1982h).The effects this stability, and in particular, why is there a of a chromosomal inversion on adult size and male mating polymorphism for genes affecting female choice? success in the seaweed fly, Coelopa frigida. Heredity, 49, The results of Crocker and Day (1987) suggested 5 1-62. an advantage to being discriminating. When C'AT('HPOLE. C. K. (1987). Bird song, sexual selection and animals were given the opportunity to exercise female choice. Trends Ecol. Evol., 2, 94—97. CROCKFR, C;. AND DAY, T. (1987). An advantage to mate choice mate choice, a larger proportion of females were in the seaweed fly, Coelopa frigido. Dehac. Ecol. Sociohiol. mated and their progeny had better survival, com- 20, 295-301. pared to animals given no choice. Higher fecundity DARWIN,C.IS.(1871).The Descent of Man, and Selection in and superior progeny fitness obviously constitute Relation to Sex. Murray, London. a selective advantage for mate choice. What is the DAY,T. H.ANDBUC'KLEY, P.A.(1980).Alcohol dehydrogenase polymorphism in the seaweed fly, Coelopa frigido. Biochem. evolutionary point of being non-discriminating? Genet., 18, 727—742. Under conditions of low adult density, females DAY, T. H. ANI) IStJTI,IN, IS. K. (1987). Non-random mating in may encounter rather few males and non-choosy natural populations of the seaweed fly, Coelopa ,frigida. females may then be more likely to breed than h'eredit3', 58, 213—220. DAY,T. H., DOBSOS',T., HILLIER, P. C., PARKIN, D. T. AND those who are choosy. Results to be published CLARKE, B. (1980). Differentrates of development associ- elsewhere indicate that flies carrying an Adh-C ated with the alcohol dehydrogenase locus in the seaweed allele do indeed enjoy a selective advantage over fly, ('oelopa frigida. Heredity, 44, 32 1-326. other genotypes when the encounter rate is low DAY,T II , DOBSON, I., HI I. LIER. P. ( ., PARKI N. t). T. ANI) ('LARKE, B. (1982).Non-random association of enzyme (H, Sawyer, K. Kristou, A. Brown and T. Day, and chromosome polymorphisms in the seaweed fly, unpublished results). In seaweed flies there is a Coelopa frigida. Heredity, 48, 3 5-44. genetic polymorphism for mate discrimination that DAY, I. H., DAWE,C., DOISSON,T.ANI) HILLIER, P. C. (1983). could be maintained in a state of balance, with A chromosomal inversion polymorphism in Scandinavian discriminating animals being at an advantage due populations of the seaweed fly, Coelopa frigida. Hereditas, 99, 135—145. to natural selection at high population densities, DAY,T. H., MILES, S., PILKINGTON, M. I). ANI) BUTLIN, R. K. but at low densities being subject to adverse sexual (1987).Differential mating success in populations of sea- selection. weed flies (C'oelopa frigida). Heredity, 58, 203—212. GENETIC DIFFERENCES IN FEMALE MATE CHOICE 131

DAY, T. H., FOSTER. S. ANt) ENGELHARD, 0. (1988). Mating MAJERSJS, M. E. N., O'DONALD, P., KEARNS, P. W. E. AND behaviour in seaweed flies (Coelopafrigida). Submitted to IRELAND, H. (1986). Genetics and evolution of female J. Insect Behav. choice. Nature, 321, 164-167. FISHER, R. A. (1930).TheGenetical Theory of NaturalSelection. O'DONALD, p(1980).Genetic Models of Sexual Selection. Clarendon Press, Oxford. Cambridge University Press, Cambridge. FISHER, R. A. 1934). Statistical Methods for Research Workers O'DONALD, P. AND MAJERUS, M. E. N. (1985). Sexual selection (Section 21.1). Oliver and Boyd, Edinburgh. and the evolution of preferential mating in ladybirds. HEISLER, I. L. (1984). Inheritance of female mating propensities I. Selection for high and low lines of female preference. for yellow locus genotypes in Drosophila rnelanogaster. Heredity, 55401-412. Genet. Res., Camb., 44, 133-149. PARTRIDGE, L. (1980). Mate choice increases a component of KELSEY, M. C. (1969). An investigation into the alcohol dehydro- offspring fitness in fruit flies. Nature,283, 290-291. genase polymorphism in Coelopa frigida (Fabricius). B.Sc. READ, A. F. (1986). Female mate choice in pied flycatchers: an Thesis, University of Edinburgh, U.K. answer and a problem. Trends Ecol. Evol., 1, 85. KIRKPATRICK, M. (1982). Sexual selection and the evolution RYAN, M. .i. (1983). Sexual selection and communication in a of female choice. Evolution, 36, 1—12. neotropical frog, Physalaemus pustulosus. Evolution, 37, LANDE, R. (1981). Models of speciation by sexual selection on 261—272. polygenic traits. Proc. Natl. Acad. Sci. USA, 78, 3721-3725. SEMLER, D. E. (1971). Some aspects of adaptation in a poly. LEWONTIN, R. C. AND FELSENSTEIN, J. (1965). The robustness morphism for breeding colours in the three-spine stickle. of homogeneity tests in 2 x N tables. Biometrics, 21, 19—33. back (Gasterosteus aculeatus). J. Zool. Lond., 165, 291-302. MAJERUS,M. E. N. (1986). The genetics and evolution of female THORNHILL, R. AND ALCOCK, j.(1983).The Evolution of Insect choice. Trends Ecol, Evol., 1, 1—7. Mating Systems. Harvard University Press, Cambridge. MAJERUS, M. E. N., O'DONALD, P. AND WEIR, j.(1982). Female mating preference is genetic. Nature,300,521—523.