The evolutionary history of Drosophila buzzatii. XVII. Double mating and sperm predominance

Sperm predominance in males and double mating in females have been studied in 2 stocks of the cactophilic species Drosophila buzzatii. The relationship between double mating and total productivity of females was also ascertained. Our results show high values of sperm predominance and double mating. Moreover, female productivity is increased with a second mate. These results are discussed in relation to the mating strategy of this species.


INTRODUCTION
Multiple mating is a widespread phenomenon among insect females (Thornhill and Alcock, 1983;Smith, 1984;Ridley, 1988). If the sperm of the first male is not exhausted before female remating, then sperm competition occurs in the storage organs of the female between the sperms of different origin (Parker, 1970(Parker, , 1984.
Sperm predominance, usually that of the last mated male, is the general result of * Correspondence and reprints this competition. When there are genetic differences in the degree of predominance, sperm predominance may result in sexual selection. Prout and Bundgaard (1977) showed theoretically how this type of selection could maintain a population in stable equilibrium for 2 alleles. The mating strategies of many species are determined by the importance of sperm predominance in males and multiple mating in females (Smith, 1984). Several experimental studies have been carried out to ascertain the degree of sperm predominance in Drosophila (Gromko et al, 1984). In the present work, we have studied sperm predominance in the cactophilic species Drosophila buzzatii. In addition, the frequency of double mating and its influence on the total female productivity were determined. D buzzatii belongs to the repleta group of Drosophila and several aspects of its ecology and mating behaviour have been extensively studied in our laboratory (Ruiz et al, 1986, Santos et al, 1988, 1989.

MATERIALS AND METHODS
Two stocks were used in this experiment. One, the wild type stock, was derived from a natural population collected at Carboneras, Almeria (SE Spain), in May 1986. The other stock was homozygous for the sex-linked recessive white mutant which arose spontaneously and was subsequently isolated in our laboratory in April 1983. Since no attempt was made to randomize the genetic background of the 2 stocks, they might differ at many loci and the mutant white was merely a genetic marker.
The experimental procedure was similar to that of Turner and Anderson (1984). The crosses performed are shown in table I. The w/+ females and the w/Y males were the hybrid offspring from the 2 parental stocks. The experiment began with the first cross. Five to 6-d old virgin females were crossed individually with 2 males of the same age. The crosses were carried out in 2 x 8 cm vials with ca 8 cm 3 of food medium. After 24 h the males were discarded. Two d later the second cross was made, also with 2 males per female. After 2 d, the males were discarded and each female was transferred daily for 11 consecutive days without etherization into vials with fresh food. Thereafter, new transfers were made at 2-d intervals.
All the individuals were grown at nearly optimal density (4-5 larvae per cm 3 of medium). A modified formula of David's killed-yeast Drosophila medium (David, 1962) was used as food. The flies were kept at 23°C. The offspring of each female was classified by sex and phenotype. Statistical analysis were conducted with the BMDP Statistical Software which was implemented on the VAX Operating System.

RESULTS
The results are presented in table II. We have estimated some population parameters that characterize sperm predominance in double matings. P 2 is the proportion of second male offspring after remating (Boorman and Parker, 1976). P 2 w is the weighted mean of P 2 , equivalent to the mean of progeny proportions per female weighted by the female's total productivity. P I is the fraction of offspring sired by the first male and P' is the proportion of the first male's sperm that is used by a female before she remates (Gromko et al, 1984). P' was estimated from the control crosses as the weighted proportion of lst-3-day offspring over total female productivity. Both P, and P Z were estimated from the female offspring, since the male offspring did not allow ascertainment of the sperm origin. The estimates of P 2 ' have been corrected for viability differences between the offspring of the 2 male genotypes (detected by the 3-way ANOVA; see below). In all cases, the values of P Z were high. Groups 1b and 2b (2nd male w/Y) showed the lowest values, 0.91 and 0.92 respectively. The other 2 values (groups 1a and 2a) were close to 1. Moreover, only on the 1st d after the 2nd cross did we find offspring of the 1st male. On the other hand, some of these descendants might have been produced prior to the time when the 2nd mating occurred. This would indicate that the actual P 2 values may be larger. The values of P', the fraction of sperm effectively used by the female before she remates, indicate the presence of at least 25-50% of sperm of the 1st male when the 2nd cross occurs. These values are underestimates, since after the 2nd cross was started females may lay eggs before remating. Given the high values of P 2 , the bias of these underestimates is insignificant. The large values of P 2 suggest that the remaining sperm is not used in the following fertilizations. Since the P z 's variances are not equal, we used the Brown-Forsythe test (Dixon, 1985) to compare the P 2 values. The P 2 angular mean was used as the dependent variable. Differences were statistically significant between groups la, 2a and 1b, 2b (change in the order of males, P < 0.001), but not between groups la, 1b and 2a, 2b (change in the female genotype, P = 0.33). So the variation in P 2 is a function of the male genotype. We find, therefore, a high degree of predominance of the last mated male, as well as possible selective differences in this component.
A possible source of error in the estimation of the P 2 values must be now considered. If during the time in which the 2nd cross occurs a female remates more than once, then our P 2 values will be spurious, since they will correspond to P2,3,...,n,! where n is the number of times a female has remated from the begining of the experiment. Patterson and Stone (1952) found that the time between matings was > 135 h in this species. This, however, is not consistent with our results, since the percentage of females that remated was practically 100% (only 3 fertile females did not produce offspring from the 2nd mate). Therefore, the time to remate in our population would be somewhat lower, but we do not know how much lower. Wheeler (1947) found that D buzzatii presents a high degree of expression of the insemination reaction. Patterson and Stone (1952) and Markow (1985) suggest that the insemination reaction works as a mechanism to preclude remating in females. Accordingly, D buzzatii females would present a long refractory period before period remating, which would diminish the chances of a remating during the period of the 2nd cross. In this sense, D buzzatii would be analogous to D mojavensis, which delays additional matings (Markow, 1985). On the other hand, we have estimated the P 2 values in the case of females being remated 2 or 3 times. We have supposed that the proportion between Pi and P i _ 1 is 1:2. This assumption is based on the values of P 2 and P 3 found in D hydei by Markow (1985). For females that remate twice the estimated values of P 2 for each group are: P la , 2 = 0.94, P 16 , 2 = 0.84, P 2a , 2 = 0.98 and P 26 , 2 = 0.94. For 3 times remated females they are: P la , 2 = 0.93, P lb , 2 = 0.81, !,2 ! 0.98 and Pz b ,z = 0.83. These values are still large. So, we think that the P 2 values may be biased upwards, but not so much as to invalidate the conclusion that sperm predominance is high. However, the differences found in P 2 between male's genotypes may be due to differential mating success rather than to selection for sperm predominance.
Additionally, we performed a 3-way analysis of variance to confirm the previous comparisons (table III). The log of the progeny number sired by the first male was used as a dependent variable. The goal of this analysis was to test: a), the effect of the female remating; b), the effect of the female's genotype; and c), the effect of the male's genotype on the offspring of the 1st male. We found significant fixed effects for each factor, but no interaction among them. First, the number of progeny sired by a male is significantly reduced if his partner subsequently remates (F = 9.42, P < 0.01). From this we deduce again that sperm predominance occurs.
Second, the female's genotype also influences productivity (F = 182.32, P < 0.001) very significantly. These differences are probably due to the different genetic background of both stocks and not to the genetic marker itself. From the third factor, we find genetic differences among male genotypes (F = 10.95, P < 0.01), but these cannot be attributed to selective differences in sperm predominance since the mate x male interaction is non-significant (F = 0.50, P = 0.48). The absence of interaction indicates that the variance of this factor is negligible in relation to the variance due to other components, such as viability. It would appear, therefore, that selection for sperm predominance, if it exists, is not as important as the other components in these experiments.
Except for groups 1b and 3b, statistically significant differences were found for the total progeny between the single and double mated females. Similar results have also been obtained in D pseudoobscura (Turner and Anderson, 1983) and D Tnojavensis (Markow, 1982), although contrary results have been reported for D melanogaster (cf Boorman and Parker, 1976;Prout and Bundgaard, 1977;Gromko and Pyle, 1978;Alvarez and Fontdevila, 1981). In our case, it is clear that a single insemination is not sufficient to fertilize all the eggs a female can normally lay. This is against one assumption of the model of Prout and Bungaard (1977), which claims that female output is independent of the number of matings.

DISCUSSION
In the face of sperm competition, the male strategies can lead to 2 different, not necessarily exclusive, ways of adaptation. One which increases P l , the fraction of offspring sired by the first male, either by an increase of P' (ie, through an increase in the time before female remating) or by &dquo;resisting&dquo; the predominance of the second male. The other way is to increase P a . Our results seem to suggest that the male mating strategy of D buzzatii is to maximize P 2 (ie the &dquo;remator&dquo; strategy) as well as P' (see Gwynne, 1984; p 141 for a possible selective explanation), whereas that of the female is multiple mating. In order to confirm this conclusion it would be necessary to perform different experiments with other stocks and varying the time of the second cross, since P 2 could depend both on the stock background and on the remating time (Boormer and Parker, 1976). Gwynne (1984) believes that the males of those species with higher predominance supply food to the female or to his offspring. Ankney (1984, 1988) found in D mojavensis that males transfer nutrients to females. We have indirect evidence that D buzzatii males transfer yeast to females during mating (Starmer et al, 1988). Thus, transmission of nutrients might help to explain the increases in female productivity with the number of matings. A more likely possibility to account for female multiple mating is in relation with the high total productivity found in this species (Barker and Fredline, 1985;Barbadilla, 1986). Ridley (1988) reviews extensive evidence showing that species with higher productivities are less likely to receive sufficient sperms at one mating than species with low productivity, whereby in species with a high productivity a multiple mated female will leave more offspring than a single one.
The strong sperm predominance found in D buzzatii has a considerable technical interest. Natural populations of this species show a moderately high inversion polymorphism in 2 autosomes. The karyotype of wild males can be ascertained by crossing them with virgin females from a laboratory stock homozygous for certain chromosome arrangements, and analyzing the salivary gland chromosomes of a number of larvae from each progeny. On the other hand, we seldom find out the karyotype of wild females, for they are usually already inseminated when collected in the field. One way to overcome this difficulty could be to remove sperm from the females by means of a low temperature treatment prior to their mating with the laboratory stock males (Anderson et al, 1979). This method, however, has proven to be completelly ineffective in D buzzatii (unpublished results). The high values of P 2 suggest a simple solution to this problem. If wild females brought to the laboratory are crossed with males of known karyotypes and then transferred every other day to a new vial with fresh food, we expect that after a few days almost all the progeny will be sired by the laboratory males, allowing the correct identification of the female's karyotype. In a recent study carried out in the population of Carboneras (Ruiz et al, submitted) this prediction has proved to be essentially correct.