Sensory-motor tuning allows generic features of conspecific acoustic scenes to guide rapid, adaptive, call-timing responses in Túngara frogs
Data files
Aug 15, 2024 version files 1.64 MB
-
code_to_submit.zip
1.63 MB
-
README.md
7.65 KB
Abstract
Male frogs court females from within crowded choruses, selecting for mechanisms allowing them to call at favorable times relative to the calls of rivals and background chorus noise. To accomplish this, males must continuously evaluate the fluctuating acoustic scene generated by their competitors for opportune times to call. Túngara frogs produce highly frequency- and amplitude-modulated calls from within dense choruses. We used similarly frequency- and amplitude-modulated playback tones to investigate the sensory basis of their call-timing decisions. Results revealed that different frequencies present throughout this species’ call differed in their degree of call inhibition, and that lower-amplitude tones were less inhibitory. Call-timing decisions were then driven by fluctuations in inhibition arising from underlying frequency- and amplitude-modulation patterns, with tone transitions that produced steeper decreases in inhibition having higher probabilities of triggering calls. Interactions between the varied behavioral sensitivities to different conspecific call frequencies revealed here, and the stereotyped amplitude- and frequency-modulation patterns present in this species’ calls, can explain previously surprising patterns observed in túngara frog choruses. This highlights the importance of understanding the specific sensory drivers underpinning conspecific signaling interactions, and reveals how sensory systems can mediate the interplay between signal perception and production to facilitate adaptive communication strategies.
README: Sensory-Motor Tuning Allows Generic Features of Conspecific Acoustic Scenes to Guide Rapid, Adaptive, Call-Timing Responses in Túngara Frogs
About the Study:
In this study, we preformed playback experiments to calling male túngara frogs to see how amplitude and frequency modulation in tonal playback stimuli influenced their call-timing decisions. From the results, we gained some clues regarding the functioning of their call timing mechanisms.
Version Information and Packages Used
- RStudio v2021.09.0/351 running R v4.2.2. Packages used:
o tidyverse_v2.0.0
o lme4_v1.1.31
o DHARMa_v0.4.6
o MuMIn_v1.47.5
o sjPlot_v2.8.12
o ggplot2_v3.4.1
o lmerTest_v3.1.3
o car_v3.1.2
o splines_v4.2.2
o afex v1.3-0
o RColorBrewer v1.1-3
o viridis v0.6.4
o cairo v1.6-0
o patchwork v1.2.0
Data Included
- 'gap_detection_dat_final.csv': a csv file containing results of our playback experiments, which provides the data for running our logistic regression model discussed in-text using the code in ‘Main_Text_Logistic_Regression_and_Figs.rmd’. Each row of the dataframe represents a single call produced by the male being tested, and each call included in this dataframe occurred during one of our tonal playback stimuli. Variable description:
o ‘block_ID’: The night on which the male was tested (multiple males were tested each night); ‘male_ID’: A unique identifier for the male being tested.
o ‘stim_ID’: A description of the tonal stimulus during which the call occurred. The first frequency stated is the reference tone frequency, the second is the test tone frequency, and the decimal number is the relative amplitude of the test tone (as a proportion) as compared to the reference tone of that stimulus. E.g., ‘400hz(bg)_850hz(gap)_0.37’ states that the stimulus consisted of a 400Hz reference tone and a 850Hz test tone, with the test tone amplitude being 37% of the reference tone amplitude.
o ‘stim_onset_uncorr’: The onset time of the stimulus during which a call occurred during a playback trial. ‘Uncorr’ refers to this being uncorrected (as opposed to corrected: see next variable).
o ‘stim_onset_corr’: The onset time of the stimulus during which a call occurred, corrected to reduce errors occurring due to the effector delay males exhibit when calling. It takes ~70ms for a triggered call to be produced in this species, so we excluded responses occurring within the first 70ms of stimuli as these were likely triggered prior to stimulus onset and so did not represent an actual response to features of the stimulus. So, this corrected onset is just the uncorrected onset time plus 70ms.
o ‘stim_end’: The time when that stimulus ended during the playback trial.
o ‘reference_Hz’: The frequency of the reference tone of that stimulus.
o ‘test_Hz’: The frequency of the test tone of that stimulus.
o ‘test_amp_reduc’: The relative amplitude of the test tone as compared to the reference tone for that stimulus (as a proportion).
o ‘call_t’: The onset time of that call during the playback trial.
o ‘prior_IOI’: The inter-onset interval preceding the current call (i.e. the time elapsing between the onset of the current call by the focal male and the onset of his most recent call immediately preceding it).
o ‘in_test.1.yes’: Indicates whether that call occurred during the reference tone or test tone of that stimulus. 1 states that it occurred during the test tone, 0 that it occurred during the reference tone.
o ‘test_t’: For calls that occurred during test tones, this is the onset time of the test they occurred during.
o ‘test_ord’: For calls that occurred during test tones, this is the ordinal test tone segment the call occurred in within the stimulus, e.g. 1 indicates it was the 1st test tone segment of the stimulus, 2 the 2nd, etc.
o ‘male_ID’: A unique identifier for the male being tested.
o ‘mass’: Male mass in grams.
o ‘svl’: Male SVL in mm.
o ‘temp’: Temperature of the water the male called in during playback (C).
o ‘total_calls’: The total number of calls he produced during playback. Not all males called through the whole session.
- ‘binary_dat_final.csv’: This is a csv used to do the analysis mentioned in section SI2 using the code in ‘SupplementalSI2_Logistic_Regression.rmd’. All variables with identical names to those in ‘Gap_Detection_Data.csv’ above have identical descriptions.
o ‘called.1.yes.’: A binary male response to each individual stimulus he encountered while calling during a playback trial. 1 indicates that he called at least once during that stimulus, and 0 indicates that he refrained from calling at all during the entire stimulus.
- 'IOIs_combined.csv’: For plotting figure 4B in-text. This figure compares inter-onset intervals observed in the current study to those observed in a previous study on this same species in which males called in live choruses.
o ‘IOI_curr’: Inter-onset intervals (s) observed during the current study.
o ‘IOI_AN’: Inter-onset intervals (s) observed in the previous study (Larter and Ryan (2024), American Naturalist).
- ‘Inhibition_sensitivity_ranks.csv’: This contains the inhibition ranks of the different frequencies we used in out experiments as revealed by our in-text logistic regression model, as well as the sensitivities to these different frequencies as reported in several previous neurophysiology studies. Used for creating figure 5.
o ‘hz’: the frequency for which a rank is displayed per row.
o ‘inhibition_ranks’: Rankings of how inhibitory the different frequencies used in our experiment were as indicated by our in-test logistic regression model. 1 indicates the least inhibitory frequency, and 6 the most inhibitory.
o ‘ponnath’: Sensitivity rankings for the same frequencies extracted from Ponnath et al. (2022), Plos One. For all sensitivity ranks, 1 indicates the least sensitive frequency, and 6 the most sensitive.
o ‘ryan: Sensitivity rankings for the same frequencies extracted from Ryan et al. (1990), Nature.
o ‘taylor: Sensitivity rankings for the same frequencies extracted from Taylor et al. (2019), Current Zoology.
o ‘wilczynski: Sensitivity rankings for the same frequencies extracted from Wilczynskiet al. (2001), Brain, Behavior, and Evolution
Code for Analysis and Figures Included
- ‘Main_Text_Logistic_Regression_and_Figs.rmd’: An R markdown file which processes and analyses the data used for our in-text logistic regression model, and which uses the above CSVs to plot all within-text figures. Some figures were edited slightly after being exported from R. Fairly extensive comments within this code should guide the reader through the different steps involved in all of these tasks.
- ‘SupplementalSI2_Logistic_Regression.rmd’: An R markdown file which processes and analyses the data used for the supplementary analysis described in the supplementary information SI2. Fairly extensive comments within this code should guide the reader through the different steps involved in all of these tasks.
Methods
In this study, we preformed playback experiments to calling male túngara frogs to see how amplitude and frequency modulation in tonal playback stimuli influenced their call-timing decisions. From the results, we gained some clues regarding the functioning of their call timing mechanisms.