in

Experimental evidence for core-Merge in the vocal communication system of a wild passerine

[adace-ad id="91168"]

Study site and animals

We studied n = 64 flocks of Japanese tits in mixed deciduous-coniferous forests in Nagano and Gumma (36°17-31’N, 138°26-39’E), Japan. Although most of the birds had not been individually colour-ringed, all the experimental trials were conducted at least 400 m apart; previous observations on colour-ringed individuals showed that this distance was enough to ensure the collection of data from different individuals30. In this site, one of the major predators of small birds is the bull-headed shrike, which is often mobbed by small birds including Japanese tits.

Playback stimulus

To test whether Japanese tits recognize an alert-recruitment call sequence as a single unit, we prepared four treatments: (i) one-speaker playback of alert-recruitment call sequences, (ii) two-speaker playback of alert-recruitment call sequences with alert and recruitment calls played from different speakers, (iii) one-speaker playback of recruitment-alert call sequences, (iv) two-speaker playback of recruitment-alert call sequences with recruitment and alert calls played from different speakers (Fig. 3). We created sound files for these treatments using the software program Audacity 2.1.3 (http://www.audacityteam.org). For one-speaker treatments, we composed mono sound files where call sequences were repeated onto a single channel, whereas for two-speaker treatments, we composed stereo sound files where either alert or recruitment calls were repeated onto the right or left channels, respectively. All the files contained an equal number of alert calls (30 calls) and recruitment calls (30 calls) at the same rate (one call every 3 s), resulting in 90-s of stimuli (Fig. 3), which corresponds to the range of the natural calling rate of alert-recruitment sequences during mobbing by Japanese tits10. For all stimuli, within-call-sequence intervals between alert and recruitment calls were constant (0.1 s), which is within the range of intervals of these calls in natural call sequences11,17. In contrast, between-call-sequence intervals varied from 1.50 to 1.81 (median = 1.68) due to the difference in call length, but were constant across playback stimuli within the same “block” where the four treatments were created using the same call exemplars (see below). While alert calls are composed of three distinct note types, recruitment calls are strings of the same note type that vary in repetition number. Since the repetition number can vary depending on predator type10, we conducted predator exposure experiments to Japanese tit flocks (n = 12) and recorded call sequences towards a bull-headed shrike life-like specimen. In response to a shrike specimen, tits produced alert-recruitment call sequences with a recruitment note repetition number ranging from 5 to 15. Since the interquartile range of repetition number was 6.75 to 10, we used recruitment calls with 7–10 notes as playback stimuli in this study. In consideration for the possible influence of sound editing procedure, we created all the stimuli in the same manner; we copied alert and recruitment call parts separately from recording files, and pasted them onto background noise files to produce all four types of stimuli. Playback amplitudes were constant across treatments, 70 dB at 1.0 m measured using a sound level meter (SM-325, AS ONE Corporation). Therefore, the differences between treatments only depend on whether these calls are produced as sequences from the same source and how the calls are ordered.

We carefully designed experiments to control for the possibility that individual-based acoustic features in alert and recruitment calls might influence tits’ responses. First, we prepared 16 unique sets of alert and recruitment calls using either calls from the same bird (n = 8 source individuals, n = 8 unique call sets) or from two different birds (n = 16 source individuals, n = 8 unique call sets). Then, we created the four types of treatments (i.e., alert-recruitment call sequences from the same speaker, from different speakers, and in reversed order from the same speaker and from different speakers) from each of the alert-recruitment call sets, resulting in 16 blocks of playback stimuli (Supplementary Table 3). This allows us to test the possible influence of individual-based acoustic variation on receivers’ responses.

We were also careful to avoid the possible influence of population-level signatures of acoustic features: we only used Japanese tits’ call sequences that had been previously recorded from the same study population. We saved the sound files in .wav format (16-bit accuracy, 48-kHz sampling rate) onto a playback device (iPhone 8, Apple Inc.). We used the default Music app (Apple Inc.) to playback the sound files.

Experiment

We (TNS and YKM) conducted experimental trials from 26 October to 4 December 2020 and during the period of 0800 and 1600 h (Japan Standard Time). We did not conduct trials under wet and windy weather conditions, since these may influence behavioural patterns of forest birds31. First, we searched for and located a flock of Japanese tits. Upon finding a flock, we fixed a taxidermic specimen of bull-headed shrike in a perching posture on the branch at 1.8 ± 0.2 m (mean ± s.d., n = 64) above the ground. Then, we placed either one or two Bluetooth speakers (SoundLink Micro, BOSE) on tree branches at 1.6 ± 0.2 m (mean ± s.d., n = 96) above the ground, and oriented them upwards to control for the possible influence of directionality. We set the distance between the shrike specimen and the speaker(s) at 5 m. For trials with two speakers, we set the distance between speakers at 10 m, mimicking the situation in which two birds are calling (Fig. 3). The shrike specimen was first covered with a black cloth and was exposed by removing the cloth just before each trial.

We began playbacks when at least two Japanese tits were present within 15 m from the shrike specimen. During 90-s of playbacks, we recorded (i) whether birds approached within 2-m of the shrike specimen during the playback and (ii) whether birds exhibited wing flicking displays12,13. We counted the number of individuals within 15 m from the shrike during 90-s of playbacks and considered it as flock size. During trials, we sat on the ground at ca. 10 m from the shrike specimen to decrease the influence of the observers’ presence on bird behaviour. To account for the inter-observer reliability32, we calculated intra-class correlation coefficient (ICC; icc function in the R package irr) between us. The lowest ICC was 0.998, indicating high degree of inter-observer reliability for the two behavioural measurements. We also video-recorded the responses of tits using a digital video camera (FDR-AX60, SONY). After completion of each trial, we checked the video recording and made an on-the-spot confirmation of the exact location at which each bird made the closest approach to the shrike specimen during the 90-s of playbacks. Then, using a tape measure, we recorded the minimum approach distance of birds to the shrike specimen. Thus, our final data set consisted of the most reliable observations confirmed by two experimenters and video evidence.

The order of trials was randomized within each block (n = 16 blocks), each of which is composed of a unique alert-recruitment call set but includes four treatments differing in the number of speakers and call order. Therefore, responses to all four treatments were observed under largely similar conditions. In a few trials, the first bird to approach the shrike specimen was from a heterospecific species, such as a varied tit (n = 1) or a long-tailed tit (n = 1). To account for the possibility that these birds evoke mobbing behaviour in Japanese tits, we only used the data from instances where the first individual to approach the shrike was a Japanese tit. Otherwise, we repeated the same treatment at a different site.

We used 64 unique playbacks created from 16 unique sets of alert-recruitment calls for 64 trials in order to avoid pseudoreplication33. We prepared two specimens of male bull-headed shrikes and used each of them for the equal number of trials. We did not use specimens of female shrikes since females migrate from the study site in late summer and only males were observed during the study period.

Statistical analysis

We analyzed the effect of playback treatments on the mobbing behaviours of Japanese tits using generalized linear mixed models in R34,35. We used the proportions of Japanese tits in flocks that (i) approached within 2-m of the shrike specimen and (ii) exhibited wing flicking displays. For the analysis of predator approach, we prepared two vectors (i.e., the number of Japanese tits that approached the shrike specimen and the number of Japanese tits that did not approach the shrike specimen). Then, we created a single response variable by binding together these two vectors using cbind function. Similarly, for the analysis of wing flicking displays, we created a single response variable by binding two vectors (i.e., the number of tits that exhibit wing flicking and the number of tits that did not exhibit wing flicking). We fitted playback treatments as a fixed term, and flock size (maximum number of Japanese tits observed during 90-s of playback) and the way of creating playback stimuli (whether the two call types were recorded from a single individual or two individuals) as covariates. We also included identity of alert-recruitment call sets that were used for creating playback stimuli (i.e., call sets from either one or two source individuals) and identity of shrike specimens as random terms. We used a binomial error distribution and logit-link function (glmer in the R package lme4) for these models. Statistical significance was calculated by log-likelihood ratio tests using anova in the R package stats. We further conducted post-hoc pairwise comparisons between treatments by using estimated marginal means (emmeans in the R package emmeans). When making pairwise comparisons, we adjusted p-values by applying a false discovery rate control for multiple testing36. All tests were two-sided and the significance level was set at α = 0.05. Exact p-values are reported when p ≥ 0.0001.

Ethics statement

All protocols were approved by the ethics committee of Kyoto University, the Ministry of the Environment, and the Forestry Agency of Japan, and adhered to Guidelines for the Use of Animals of the Association for the Study of Animal Behaviour/Animal Behavior Society37.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.


Source: Ecology - nature.com

Dynamics of microbial community and enzyme activities during preparation of Agaricus bisporus compost substrate

Environmental conditions experienced upon first breeding modulate costs of early breeding but not age-specific reproductive output in peregrine falcons