Does developmental stress affect female choice?

One of the major aims of this thesis was to investigate whether early environmental conditions affect female choice behaviour. The developmental stress hypothesis suggests that complex male songs evolved as a signal of developmental condition through the action of female choice, but the same stressors that are predicted to regulate male attractiveness have the potential also to act on females. The HVC is thought to play an important role in female song perception (Brenowitz, 1991; Del Negro et al., 1998; Del Negro et al., 2000; Halle et al., 2002), so stress might affect females’ ability to evaluate song at a mechanistic level. Another possibility is that poor quality females could suffer higher costs from choosiness than high quality females and so be less discriminating, or expend less effort in sampling potential mates (Fawcett and

Johnstone, 2003a; Cotton et al., 2006). Four studies have now addressed these questions explicitly (Riebel et al., 2009; Holveck and Riebel, 2010; Woodgate et al., 2010, chapter 2; Woodgate et al., in press, chapter 3).

6.3.1 Perception o f male signals

Perhaps the most important finding to emerge from these four studies is that none have found results consistent with the hypothesis that female song perception is affected by stress. Riebel et al. (2009) found that there was no difference between female zebra finches from large or small broods in either the direction or strength of their preferences when choosing between their tutor’s song and the song of an unfamiliar male. As in previous experiments (Riebel, 2000; Riebel et al., 2002), females were found to prefer songs to which they had been exposed during the sensitive phase for song learning, showing that even females from experimentally enlarged broods retained the ability to recognise their tutor’s song. In this project, I did not test female preferences for familiar songs but, in chapter 3, I found that both females from control and nutritional stress treatments preferred more complex songs, demonstrating that females from the stress treatment were able to discriminate between songs on the basis of their structure. Finally, Holveck and Riebel (2010) found that females showed assortative preferences for males originating from similar broods sizes to themselves, confirming their ability to discriminate between songs, even though the only known structural differences between the two groups of stimulus songs were subtle differences in temporal consistency or the accuracy of syntax learning (Holveck and Riebel, 2010). These results do not prove that there was no detrimental effect of stress treatment on perceptual ability, but they do provide strong evidence that any effect o f early condition on auditory perception is not large enough to prevent females making choices based on naturally occurring levels of variation between male songs. Note that, while the simplified songs I used as stimuli in

chapter 2 were created by digital editing, rather than being naturally occurring songs, they were designed to mimic the levels of song complexity previously found in nutritionally stressed males (as in Spencer et al., 2003).

6.3.2 Direction o f preference

Even in the absence of a direct effect of stress on song perception, developmental conditions could influence the strength or direction of female preferences. Holveck and Riebel (2010) reported a very striking effect of brood size manipulation on the direction of preference, with female zebra finches preferring the songs of males raised in similar sized broods to themselves. Assortative mating by phenotypic quality might be expected to occur if low quality females face greater costs from the search for a high quality mate or are likely be outperformed by higher quality females in competition for mates (Fawcett and Johnstone, 2003a; Hardling and Kokko, 2005), and Holveck and Riebel (2010) suggest this might explain their results. Once again it is possible that differences in the nature or severity of the experimental stressor might underly some of the different results reported by different experiments (see section 6.2, above). Holveck and Riebel (2010) tested the preferences of females from a brood size manipulation experiment for the songs of males from the same experiment. Riebel et al. (2009) also tested the preferences o f females reared in a similar brood manipulation experiment but offered them the choice of songs from unmanipulated males, and found no such evidence for an effect of brood size on female preference, even though the stimulus songs differed in phrase length. This raises the question of whether the females tested by Holveck and Riebel (2010) were expressing preferences for males of similar quality to themselves or for song characteristics that convey information about brood size.

I found no effect of nutritional stress on the direction of female preferences when choosing between song recordings in operant tests (Woodgate et al., in press, see chapter 3). These stimulus songs had been edited to differ in syllable number but were otherwise identical. In the only experiment to date, testing the effects of developmental stress on female preferences for live, unmanipulated males (Woodgate et al., 2010, see chapter 2), I likewise found no evidence that stressed and control females differed in the direction of their preferences. The assortative preferences found by Holveck and Riebel (2010) were not apparent when females were offered a choice between the songs of familiar males, implying that learned preferences are strong enough to override any effect of developmental conditions on song preference.

6.3.3 Choosiness

Females are predicted to be less choosy when the costs of choice are greater, but more choosy when the benefits of choosing the best male are great (Cotton et al., 2006). Both stressed and control females, tested in chapter 3, expressed a preference for more complex songs and there was no difference between the groups in the strength of this preference. Females from small and large broods did not differ in preference strength, either when choosing between familiar or unfamiliar songs recorded from males originating in small or large broods (Holveck and Riebel, 2010), nor when choosing between their tutor’s or an unfamiliar song (Riebel et al., 2009). However, when Riebel et al. (2009) offered females from experimentally enlarged broods a choice between two unfamiliar male songs they expressed significantly weaker preferences than those from smaller broods. In mate choice tests I found that females reared under conditions of nutritional stress were less active and made fewer sampling visits to the stimulus males than controls, indicating reduced investment in mate sampling and reduced choosiness (Woodgate et al., 2010, chapter 2).

The discrepancy between these results might be attributable to differences either in the costs of choice, or in the potential benefits of choosiness, between the studies. Choices for live males in the mate choice apparatus used in chapter 2 are likely to carry greater costs than in operant song preference tests, both in terms of the energetic requirements o f moving between viewing arms and perhaps also the cognitive challenge of relocating the position of a favourite male. If this is the case, low quality females might benefit from reducing choosiness in the live male tests but, in song preference tests, could still expect to maximise their fitness through choosiness. Some evidence in favour of this interpretation comes from the fact that the difference in activity rates between stressed and control females that I observed in mate choice tests was not apparent in later, no-choice tests (Woodgate et al., 2010, chapter 2), although it should be noted that there are a number of potential alternative explanations (see section 2.4).

The benefits of choosiness may also have differed between studies. Stressed females appeared less choosy when offered a choice between unfamiliar, unmanipulated males (Riebel et al., 2009; Woodgate et al., 2010, chapter 2), perhaps indicating that the benefits o f choosiness were relatively small, and did not outweigh the costs, for low quality females. When offered a choice between male songs that were known to differ in structure and might therefore signal differences in male quality, all females expressed strong preferences (Holveck and Riebel, 2010; Woodgate et al., in press, chapter 3), as did they when one o f the stimulus songs was one for which they had a learned preference (Riebel et al., 2009). In these cases the perceived benefits of choice were perhaps great enough that even low quality females could enhance their fitness by expressing strong preferences.

6.2.4 Summary

To summarise, the only studies to date that have addressed the effects of developmental stress on female choice (Riebel et al., 2009; Holveck and Riebel, 2010; Woodgate et al., 2010, chapter 2; Woodgate et al., in press, chapter 3) suggest it is unlikely that developmental condition has any biologically relevant effect on females’ ability to discriminate between male songs. Developmental stress may, however, cause context dependent differences in the strength or even the direction o f song preferences. Even these relatively subtle effects have the potential to have profound consequences for mating decisions in natural populations. For example, differences in the way females sample potential mates could still affect the outcome of mate choice even where females do not differ in preference since females that sample few potential mates will show less bias in their mate choice than those that choose between many (Wagner, 1998).