Female Bengalese finches recognize their father’s song as sexually attractive

Birdsong is an important communication signal used in mate choice. In some songbirds, only males produce songs while females do not. Female birds are sensitive to inter- and intra-species song variation. Some aspects of female song preference depend on developmental experiences. For example, in Bengalese finches and zebra finches, adult females prefer the song to which they were exposed early in life, such as the father’s song. However, it is unclear whether such song preference in females is sexually motivated. The purpose of our study is to test if female Bengalese finches recognize their father’s song as sexually attractive. We measured copulation solicitation displays during playbacks of the father’s song vs. unfamiliar conspecific songs and found that across individuals, the father’s song elicited more displays than other songs. In addition, we analyzed if a bird’s response to a given song could be predicted by the level of similarity of that song to the father’s song. The results suggest that preference for the father’s song in this species is actually relevant to mate choice. Although more precise control is necessary in future studies to elucidate the process of preference development, our results imply the significance of early-life experience in shaping female song preference.


Introduction 29
Song is an important communication signal used for mate choice in songbirds [1]. In some 30 species of songbirds, only males produce songs. While females do not sing, they can perceive inter-31 and intra-species song variations and change their behavior depending on features of the song that 32 reflect the sex, species, and condition of the signaler [2]. For example, female birds selectively respond 33 to conspecific songs over heterospecific ones [3][4][5]. In some species, this preference seems to be a 34 common tendency across individuals that develops independent of experience after hatching [6,7].

35
However, developmental experience plays a significant role in shaping other aspects of female song 36 preference. A cross-fostering study in the zebra finch (Taeniopygia guttata) demonstrated that 37 recognition of a subspecies based on song depends on auditory experience with the foster father's song

38
[8]. Early-life song exposure in this species is also critical to be able to discriminate song quality [7] 39 or song performance related to social context [9].

40
In addition, developmental experience might be an important source of individual 41 differences in song preference. Previous studies in zebra finches and Bengalese finches (Lonchura 42 striata var. domestica) found that females prefer songs to which they were exposed early in life, such 43 as the father's song [10][11][12][13][14][15]. These studies clearly show that female birds acquire and retain an ability 44 to discriminate their (foster) father's song from other songs, but it is not clear whether such song 45 preference is sexually motivated. Although the possibility of sexual imprinting on the father's song removal.

110
Preparation and presentation of song stimuli 111 In the song preference tests, we presented 5 different conspecific songs to each subject: the 112 father's song and 4 unfamiliar songs. Females had never been exposed to these 4 unfamiliar songs 113 prior to the experiment. We presented the song of one subject's father to 2 or 3 other subjects (that 114 hatched in a different clutch) to exclude the possibility that a particular characteristic of a given song 115 generally attracts females, independent of experience.

116
To prepare stimuli, we recorded songs from adult male Bengalese finches in a soundproof 117 chamber through a microphone (PRO35, Audio-technica) fixed at the top of the cage. The signal from 8 / 24 118 the microphone was amplified (QuadMic, RME) and digitized (Delta66, M-AUDIO) at 16 bits with a 119 44.1 kHz sampling rate. All recordings were undirected songs, as each bird was kept alone in the 120 chamber during recording. For each male singer, we randomly selected 6 renditions of songs that were 121 recorded without any background noise ( Fig. 1(c)). The duration of a single song rendition was 7.04 122 ± 0.64 (mean ± s.d.) seconds. The sound waveform was first band-pass filtered at 0.5 -10 kHz. We 123 normalized the sound amplitude by the standard deviation of the waveform. For each singer, a song 124 was composed of 6 song renditions presented with a 700-ms interval between each rendition (thus 125 approximately 45 seconds in total, Fig. 1(b)).

141
In a single test, 5 songs (each consisting of 6 renditions) were played back in a random order 142 ( Fig. 1(b)). The inter-stimulus-interval was a random value ranging from 40 to 45 seconds. Thus, the 143 duration of one test was approximately 450 seconds. Since some birds did not perform any CSDs in a 144 test, the following criterion was set. We defined a valid test as one in which the subject expressed at 145 least one CSD (in other words, a test was regarded invalid if she did not express any CSDs). We 146 continued testing each female until 5 valid tests were obtained for that individual. One out of the 10 147 birds failed to meet the criterion due to health issues. We ceased testing that bird after 4 valid tests.

148
The total number of tests required to reach the criterion varied between individuals (mean = 6.3 tests 149 range = 5-9 tests). Each bird was tested once every 2 days to avoid habituation to stimuli. Thus, the 150 test period for a subject lasted for 10-18 days, depending on the number of tests required. All the tests 151 were conducted in the morning (7:00AM -12:00PM).

189
The bird's response was binarily coded as 0/1 (not performed/performed), respectively. The type of song stimulus (unfamiliar/father's coded as 0/1) and song presentation order within a test (1-5) were 191 the fixed effects of independent variables. Subject ID (10 levels), stimulus ID (19 levels), and test 192 numbers (9 levels) were also included as random effects. The binomial distribution with a logit link 193 function was specified to represent the probability distribution.

194
We next analyzed the possible effect of song similarity in the prediction of female response

230
We next conducted a post hoc analysis on whether similarity of a song stimulus to the 231 father's song predicted responses in the 7 birds that expressed CSDs to both song types (Fig. 3).

232
Among the independent variables included in the GLMM, presentation order within a test was the 233 strongest predictor of the response ( Table 2). As with the result in Table 1, a song was more likely to 234 elicit CSDs when it was presented earlier in a test (Fig. 3 (c)). Although there was a tendency for 235 unfamiliar songs to elicit more frequent CSDs when these songs had greater similarity to the father's 236 song (higher SAP % similarity, lower tempo difference, Fig. 3(a), (b)), this result was not statistically 237 significant ( Table 2). The overall results imply that female Bengalese finches in our study did 238 recognize their father's song as an attractive courtship signal, but we cannot not conclude whether their mate choice is affected by song features that are shared with their father's song.   CSDs to the father's song than to other unfamiliar conspecific songs (Fig. 2), which suggests that 261 females are sexually motivated when exposed to the father's song.

262
The song preference observed in our study might be regarded as an expression of sexual 263 imprinting on the father's song. This possibility has already been referred to in an early study of captive father's song will be necessary in the future. Moreover, although we found that songs with higher 282 similarity to the father's song elicited more CSDs, this tendency was not statistically significant. In

283
testing the effect of similarity on female response to unfamiliar songs, the small number of trials 284 available for analysis was a limitation (Fig. 3). Manipulation of stimuli based on song similarity or a 285 particular song feature, rather than a post-hoc analysis as in this study, may enable a more detailed 286 analysis on the effect of auditory experience on song perception.