| Mate choice and genetic variation in male courtship song in Drosophila montana | ||
|---|---|---|
| Prev | Chapter 3. Results and discussion | Next |
Variation in male song traits was studied in Kemi and Oulanka populations in paper III. The songs of the F2 progenies of the females captured from these populations differed significantly from each other in all song parameters, except in the carrier frequency of the song. Males from the Kemi population had higher pulse and cycle numbers and longer sound pulses, interpulse intervals and pulse trains than the males from the Oulanka population (Table 3 in paper IV). The heritabilities of male song traits were highly variable in both populations, and they also had large standard errors. The only significant heritability, h2 = 0.55, was the heritability of the carrier frequency estimated by sib analysis in the Oulanka population. Heritability of this song trait was even higher in the Kemi population (h2 = 0.59), but due to high standard error it remained nonsignificant. Coefficients of additive variation varied depending on the studied trait, the method used for estimating genetic parameters (regression or sib analysis) and the population, where the flies came from. In the Kemi population, the sib analysis gave higher CVAs for half of the song traits than regression analysis, the results being partly contradictory. In the Oulanka population, the amount of additive variation was very low in most song traits in both analyses. Variation in heritability estimates, obtained with different techniques, has also been observed in earlier studies (Roff & Mousseau 1987). Since almost all of our heritability estimates were nonsignificant, the comparison of the reliability of different techniques is useless.
Falconer (1981) has suggested that the traits closely related to fitness have lower heritabilities than the traits which are more distantly related to fitness. This has been confirmed e.g. by Roff and Mousseau (1987), who reviewed literature for heritabilities of different traits (classified as morphological, behavioural, physiological and life history characters) in Drosophila. However, Houle (1992) has shown that although h2 values are typically smaller for life-history characters than for morphological traits, the relationship is reverse, if the coefficients of additive variation are compared. In D. montana song, pulse characters can be regarded as fitness characters, because the females of this species select their mates according to these characters (Aspi & Hoikkala 1995, Ritchie et al. 1998, paper I), and because the females get indirect benefit from their choice (paper II). In our study, heritabilities were nonsignificant for most song traits in both study populations. The amount of additive variation (measured as a coefficient of additive variation) was at the same level, or even higher, in pulse characters as in pulse train characters, which gives support to Houle’s (1992) suggestion of a higher level of additive variation in fitness than in non-fitness traits.
Heritabilities of song characters have been studied earlier in D. montana Kemi population by Aspi and Hoikkala (1993), using father-son regression between wild-caught fathers and their laboratory-reared sons, and between fathers and sons reared in the laboratory. In their study, the heritability estimates across environments were low and nonsignificant, while for laboratory reared flies they were significant in some song traits (PN, IPI and PL). The difference between the results of the present study and the study by Aspi and Hoikkala (1993) may largely be due to different measuring techniques. Aspi and Hoikkala calculated means of song characters over five pulse trains of each male, while we analysed only one pulse train per male. We chose to use the latter technique, because we wanted to reveal the whole amount of residual variation in male songs, including variation within the males. Heritability has usually been estimated by minimising environmental variance, because high residual variation easily leads to nonsignificant heritability. This kind of study technique systematically inflates h2.
The amount of additive variation, measured as a coefficient of variation, was very low in most song traits of the males from the Oulanka population. Sib analysis revealed additive variation only in cycle number and carrier frequency in the Oulanka population, while it revealed additive variation in all song traits in the Kemi population. The difference in the amount of additive variation in the two populations may be due to different selection pressures or to differences in mechanisms maintaining genetic variation in populations. In Oulanka, the mating season of the flies is a few weeks later than in the Kemi population. The two populations also differ in their species constitution. In Kemi, the fly population on our study site consists of four D. virilis group species: D. littoralis, D. montana, D. lummei and D. ezoana, D. littoralis being the most abundant species (Aspi et al. 1993). In Oulanka, the riverside population consists of mainly two D. virilis group species, D. montana and D. ezoana, with about equal frequencies (Lumme et al. 1979). Among the Finnish D. virilis group species, male courtship song is an essential part of the courtship in D. montana and D. ezoana. Mate choice in D. montana has so far been studied only in the Kemi population, where the females have been found to prefer males producing short and dense (i.e. high frequency) sound pulses (Aspi & Hoikkala 1995). In this population, interspecific courtship has been found to usually break off, when the male begins to produce courtship song (Liimatainen & Hoikkala 1998). Differences between the songs of the males from the Kemi and Oulanka populations are most evident in pulse train characters, which may be more important in species’ recognition than in sexual selection. These characters vary distinctly between the species found in both of our study areas.
Phenotypic plasticity is determined as the expressed phenotype of a genotype as a function of the environment (e.g. Scheiner 1993). It is usually restricted to the change that happens during the development of an individual. We applied the term plasticity to the changes in phenotype caused by the environmental factors during adult life.
In paper II, we exposed the males from the Oulanka population to an artificial winter (4°C for six months), and studied whether the songs of the males had changed during this treatment. Only the pulse number and pulse train length were repeatable over the recordings. The carrier frequency of the song differed between the recordings of the same males before and after cold treatment. Also, the mean frequency of male songs had increased during the treatment. A two-way ANOVA on half-sister progenies revealed a significant cold treatment effect in this trait. Interaction between the genotype and the treatment appeared to be significant only in cycle number, i.e. the progenies having the highest cycle number before the cold treatment were different from those having the highest cycle number after the treatment. CVAs of all song traits, except interpulse interval, increased significantly and the CVRs of the same traits decreased during the cold treatment.
D. montana flies overwinter as adults and have a mating season in spring after the cold period (Aspi et al. 1993). During this season, the females exercise mate choice on pulse characters of the male song (Aspi & Hoikkala 1995). In study I, the females were also found to exercise the same kind of mate choice in the laboratory, among the cold-treated males. In study III, environmental factors, such as cold-treatment, were found to increase phenotypic variation in sexually selected song traits, revealing the additive component of variation. These findings suggest that the preferred male song traits are condition dependent and that they also vary genetically. There is a great deal of empirical support for the condition dependence of sexually selected male traits also in other species (Andersson 1994).
Can our findings be applied to the situation in wild populations? First of all, the conditions in the wild in winter are harder and more variable than they were in our cold-treatment experiment. This may lead to even higher CVR and nonsignificant heritabilities in male song traits in the wild (Aspi & Hoikkala 1993). Secondly, the carrier frequency of the male song is very sensitive to temperature changes, raising about 10 Hz per 1°C (Hoikkala 1985a). As the temperature can vary during the mating season of the flies about 10 degrees, the females have to be able to compensate for changes in male song frequency. An alternative is that the females simply choose the best singing male (Hoikkala & Aspi 1993).