Sexual selection was proposed by Darwin (1859) and Darwin (1871), to explain extraordinary sexually dimorphic characters troublesome to his concept of natural selection. Two different kinds of processes could account for the evolution of such traits (Thornhill & Alcock 1983, Alcock & Gwynne 1991). Male-male competition for mates, or intra-sexual selection, is where those males with the most exaggerated trait are supposed to be able to fight and win fights over other males for access to mates. The second process is active choice of individuals of one sex by individuals of the other, usually female choice of mates, namely inter-sexual selection. Again, males with the most exaggerated (Moller & Pomiankowski 1993), most symmetric (Moller 1990), or most attractive traits are assumed to be at a selective advantage because they are more likely to be chosen by a female and they should therefore experience higher mating success. For some species, finding the direct targets of sexual selection can be a difficult task and may frequently lead to misinterpretations. Moreover, the relative role of male-male competition and female choice in sexual selection has been assessed in only few cases, but there is little doubt that while male combat is of overwhelming importance in some species, mate choice predominates in others (Bradbury & Davies 1987).

A good knowledge of sexual selection mechanisms is required for the successful implementation of the sterile insect technique (SIT) developed by Knipling (1955) to control insect pest populations (Burk & Calkins 1983). Sexually selected traits are usually a good reflection of male fitness and are thus a representation of both its genotype and phenotype. Mass-rearing conditions can be controlled to improve male quality in order to promote the occurrence of sexually selected phenotypes in a high frequency. Moreover, directed selection events may be used to select genotypes with high mating success.

The Mediterranean fruit fly, Ceratitis capitata (Wiedemann) has a well-established lek mating system (Prokopy & Hendrichs 1979, Arita & Kaneshiro 1989, Hendrichs & Hendrichs 1990, Whittier et al. 1992). Leks generally take place on the underside of leaves of trees, where males establish territories and release a pheromone to attract females (Féron 1962, Prokopy & Hendrichs 1979, Arita & Kaneshiro 1989). After the arrival of a female, the male begins courtship (Féron 1962) and finally jumps onto the female and attempts to copulate. If unreceptive, the female either leaves before the male jumps or drops from the leaf when the male has already jumped and is trying to copulate (Whittier & Kaneshiro 1995). Females discriminate among potential mates at a lek and reject most courtships (Whittier et al. 1992, Whittier et al. 1994), leading to differential copulatory success within males. The common finding that copulatory success is highly variable among lekking males has led to the predominant view that mate choice is of great importance for sexual selection in such mating systems (Whittier et al. 1992, Whittier et al. 1994, Norry et al. 1999). Despite this disparity of male mating success within leks, male-male interactions may result in the males sorting themselves by territory location or position in a dominance hierarchy as reported for other dipterans (Shelly 1987, Sivinski 1989). As a consequence many traits could be correlated with copulatory success without being mate-choice cues. Male mating activity and aggressive behavior are examples of intra-sexually selected traits in many lek-mating species. According to Whittier et al. (1992) and Whittier et al. (1994)C. capitata males defend their territories very weakly, and there does not appear to be any relationship between territory location and mating success. However, male-male interactions are commonly observed both in the laboratory and in the field (though probably not very well documented).

Mating success in C. capitata males has been associated with nutritional level (Blay & Yuval 1997); decreasing fluctuating asymmetry of male supra-orbital bristles (Hunt et al. 1998); body size (Churchill-Stanland 1986, Orozco & Lopez 1993); eye length and mating activity (Norry et al. 1999). However, as yet there is still no unifying theory. In particular some studies have found an important role of male size (Orozco & Lopez 1993, Blay & Yuval 1997) while others have shown that body size itself is not a direct target of sexual selection (Hunt et al. 1998, Norry et al. 1999). Body size has been associated with male mating success in many studies of sexual selection (Davies & Halliday 1979, Berven 1981), particularly in some Diptera (Ewing 1961, Borgia 1981, Partridge et al. 1987), but this association might be explained by male-male competition (Partridge et al. 1987) and even to differential survival associated with size (Hasson et al. 1993).

Most of the studies on sexual selection of C. capitata have been done under laboratory conditions or with laboratory adapted strains, so the potential role of male-male interactions and female mate choice in the field is still poorly understood. The aim of the present work was to examine sexual selection based on morphometric traits of males of C. capitata from both a genetic sexing strain sterilized with irradiation and a wild population controlled with SIT under field cage conditions.

Materials and Methods

Results

Approximately 50% of the flies mated in each cage. Descriptive statistics for the traits measured are given on Table 1. It can be seen that for all traits Patagonia males were larger than Seib 6-96 males. Moreover, in both strains successful males showed higher average values than unsuccessful males for all traits except FW.

In order to test the observed trends statistically, Spearman’s rank correlations were estimated between 1) morphological traits and strain origin and 2) morphological traits and fitness (Table 2). Significant differences were observed between strains for all traits but FW. This analysis also showed that all traits except FW were positively and significantly correlated with fitness.

The identification of the actual target of sexual selection requires the possible correlation among traits to be removed. This was done by means of a Step-wise multiple regression analysis (Table 3). The three variables that showed a high association (P < 0.01) with fitness were TL, FW and EL (enumerated according to the relevance order), while the effect of WL was non-significant (P = 0.086). According to these results, the apparent lack of effect of FW on fitness in the original Spearman’s correlation might be a consequence of its correlation with the other size related traits.

In order to obtain completely uncorrelated (orthogonal) variables a principal component analysis (PCA) was performed with the morphometric traits analyzed. Three PCs were obtained from the factor analyses, which account for 90% of the total variance, and are explained respectively by TL and WL (PC1), FW (PC2), and EL (PC3). Factor loadings and eigen values are detailed in Table 4.

Since in the current experiment fitness was a dichotomic variable, a logistic (instead of a linear) regression analysis of fitness on the three axes from the PCA was performed. All the obtained axes showed a highly significant association with fitness (probability of mating in Table 4). It should be noted that the logistic regression coefficient for FW was negative, suggesting again that those males with smaller traits have the highest mating success.

Discussion

In the present study head traits, eye length (EL) and face width (FW), and male size, thorax length (TL), have been revealed as targets of sexual selection. EL had previously been reported as a predictor of mating success under laboratory conditions with two different strains and in outdoors field cage tests with a long established laboratory strain (Norry et al. 1999). These authors have shown the occurrence of sexual selection on head morphology even in the absence of male competition, suggesting female choice. Results from the present study performed under a different testing protocol (wild females and outdoors field cages) are highly consistent. Also, EL appeared to be associated with mating success for both wild and Seib 6-96 males when each strain was analyzed separately (data not shown). These results can be interpreted as further evidence of the importance of mate choice over male-male competition in the sexual selection process acting on EL, and probably other head associated traits as well. Discrimination among potential mates on the basis of the male head morphology probably takes place when the female approaches the wing-fanning and head rocking male (Féron 1962, Calcagno et al. 1999) face to face (Norry et al. 1999). Hunt et al. (1998) have reported an association between male supra-orbital bristle fluctuating asymmetry and mating success, another example of head traits also identified as targets of sexual selection.

The finding that TL was associated with mating success suggests that male size may be another trait subjected to sexual selection. Pupal weight, as an adult size related trait, has been reported as a determinant of mating success in field cage studies of sexual selection with laboratory irradiated and wild flies (Orozco & Lopez 1993). Additionally pupal weight has been proved as a key determinant of both male and female mating activity (Churchill-Stanland et al. 1986), though these results account more for assortative mating rather than sexual selection towards male size. Blay & Yuval (1997) using wing length as a determinant of male size have found that for matings taking place within the first 1.5 h after release of females, variation in copulatory success could be explained by variation in wing length for both protein-fed and protein deprived males. When whole day observation data was pooled this effect disappeared for protein-fed males but remained for protein deprived males. Wing length can only be environmentally influenced at the larval stage and not after imaginal molt, and the experimental design followed by these authors determined the difference among protein-fed and protein-deprived males at the adult stage. It is thus tempting to conclude that the variation in mating success of these males is probably a consequence of differential mating activity, and not necessarily due to any possible effect of male size. Male size might be indirectly associated with mating activity as a consequence of nutritional reserves passed from the larvae to the adult, bigger adults being best nourished at the larval stage than the smaller ones.

In the case of the present study TL seems to be associated directly with copulatory success. This view is supported by the step-wise regression analysis (Table 3) and the significant correlation of PC1, which accounted for body size only, with probability of mating (Table 4). The non-significant effect of wing length (WL) in the step-wise regression (Table 3) indicates that wing size even when included on PC1 (Table 4) is a trait associated with mating success only as a side effect of its high correlation with TL. This is a clear example of the power of the combination of step-wise regression plus factor analysis coupled with regression analysis (either logistic or linear). These approximations succeed in removing WL as a target under selective pressure contrary to the first exploratory Spearman’s rank correlation analysis (Table 2).

On the contrary, Whittier et al. (1994) found no correlation between male body weight and copulatory success, but they did detect a high correlation for number of attempted copulations, i.e., mating activity. Similar results have been found by Norry et al. (1999) in relation to WL and TL and by Hunt et al. (1998) for WL. These studies have been done under laboratory conditions, and the field cage test results reported by Norry et al. (1999) where no male size advantage was detected was carried out with a laboratory strain (approximately 30 generations under lab conditions). To summarize, the work by Orozco & Lopez (1993) was the only evidence to date of sexual selection favoring male size of lab males mated to wild females under field cage conditions. No evidence had been detected when females belonged to a lab strain even when tests were performed in outdoor field cages (Norry et al. in press), or when flies were tested under laboratory conditions (Hunt et al. 1998, Blay & Yuval 1997, for the case of pooled data, Norry et al. 1999). A close look at this apparent contradiction reveals that for the indoor and lab strain tests percentages of mating were very high (ranging from 69 to 86%) compared with percentages of mating found for wild females (50%, the present work). Such percentages imply that highly receptive yet low discriminating lab females may be responsible for hiding the effect of male size or size related traits on sexual selection. Rearing conditions may also represent a different environment where lek formation might not be so important as in nature.

Another possible cause for the lack of detectable size effects on sexual selection is that size can account for nutritional level only in those cases where food quality has been the same and any size variation is only a consequence of lesser consumption. Arita & Kaneshiro (1988) have found that flies emerging from coffee beans copulated more frequently even being significantly smaller than flies emerging from cherry beans. Orozco & Lopez (1993) also reported smaller size for flies emerging from coffee berries but no effect on mating activity was detected. In the current study, the advantage of bigger males was observed in both strains (the interaction fitness × origin was non significant), and the wild males, which were shown to have higher success (Cayol et al. 1999), were on average bigger than their mass-reared counterparts.