Dance or disappear: Strategic sexual signalling in female Peninsular rock agama

Sridharan, Aravind 1 ; Lawrence, Swapna2 ; Isvaran, Kavita1

Published Feb 25, 2025 on Dryad. https://doi.org/10.5061/dryad.4xgxd25mk

Data files

Feb 25, 2025 version files 9.57 KB

Abstract

Sexual signalling has traditionally been studied in the context of male competition and female mate choice, with female signalling considered of limited consequence. However, there is growing evidence of extensive sexual signalling in females in a wide range of taxa. Here, we test how females strategize the use of sexual signals in a socially polygynous lizard, Psammophilus dorsalis. We evaluated key hypotheses for how females should modulate their signalling, focussing on male quality, mate availability and stage of the breeding season. We simulate male quality using artificial male models to individually tagged, and intensely monitored wild female lizards, measuring their strategic investment in sexual signalling. We found that females invest more in signalling towards high quality males and increased their investment towards the later part of the only breeding season. We demonstrate that, contrary to typical expectations in polygynous systems, females invest in costly sexual signals and adjust their use to maximise benefits and minimise costs. We argue that in spatially dispersed species, females may face limited access to high quality mates, favouring costly sexual signalling. Our findings underscore the need to reassess sexual selection frameworks across taxa, recognising female signalling as an active force in shaping mating dynamics.

https://doi.org/10.5061/dryad.4xgxd25mk

Description of the data and file structure

Files and variables

File: behaviour_data_Procb.csv

Description:

Variables

lizard.id: Unique ID of lizard
bead.code: Unique bead code for each lizard (Character that describes the colour of ceramic beads, example WWB mean white,white, blue in the order)
date: Date (%%dd%%mm%%yyyy)
model.id: Two colour states were presented - Orange and Yellow (denoted as ORM and YLM)
prop.redhead: Proportion of time spent in red-head colour state ;Range: 0 to 1 (Time spent in behavioural state/Total observation time)
prop.crouchposture: Proportion of time spent in crouch posture ;Range: 0 to 1(Time spent in behavioural state/Total observation time)
prop.tailraise: Proportion of time spent in tail raise posture ;Range: 0 to 1(Time spent in behavioural state/Total observation time)
prop.darkbody: Proportion of time spent in dark-body colour state ;Range: 0 to 1(Time spent in behavioural state/Total observation time)
nmal: Number of males in the 20 m vicinity within a 1 month period (This data was obtained from regular surveys. In each survey the number of male neighbours were calculated for every individual female and unique males were calculated for each female)
breeding: Categories - Early and late breeding season

File: Behaviourdata_mean_CI.csv

Description:

Variables

Behaviour: The four behaviours - Tail raise, crouch, red-head, dark-body
Model: Orange and yellow model representing colour states
mean: Mean
lower: Lower CI
upper: Upper CI

Code/software

We used R software.

The workflow begins by loading the necessary libraries, ggplot2 for visualization and glmmTMB for mixed-effects modeling. A cleaned behavioral dataset (behaviour_data_Procb.csv) is then imported, containing pre-filtered behaviors of interest. To estimate variability, a bootstrap function is defined and used to calculate 95% confidence intervals (CI) for prop.tailraise as an example, separately for the YLM and ORM models based on breeding status. Mean values of prop.tailraise as an example are also computed for each model. The same is done for each behaviour and saved in a CSV file (Behaviourdata_mean_CI.csv) and reloaded for visualization. A bar plot is created using ggplot2, displaying mean values with error bars representing CIs, styled for clarity. Since beta regression requires values between 0 and 1, any occurrences of 1 in prop.tailraise are replaced with 0.999. Similar graphs were plotted for other behaviours and conditions. A zero-inflated beta regression model is then fitted to prop.tailraise, incorporating model.id, nmal.cat, and breeding as fixed effects, with lizard.id as a random effect. The same was done for prop.darkbody and prop.redhead. Additionally, since prop.crouchposture is relatively rare, it is converted into a binary presence-absence variable, and a binomial mixed-effects model is fitted.

Access information

Other publicly accessible locations of the data:

Data was derived from the following sources:

From data we collected in field

Data collection

To assess hypotheses for variation in sexual signalling in females, a field experiment was conducted. Male models of different colour states, representing two levels of male quality, were presented to tagged females on their territories. Regular monitoring of the study population provided information on mate availability.

Model presentation experiment

Female signalling responses to potential mates were assessed through field experiments on individually tagged females. 3D printed models mimicking male lizards were painted in two different colour states chosen to represent male quality: yellow (low), and bright orange (high). Female lizards were presented with these models at different times during their breeding lifespan. During each trial, the behaviour of the female before and after the presentation of a model was recorded on a video camera mounted on a tripod. The video camera was placed at a distance of at least 15 metres from the focal lizard to avoid any disturbance to the focal lizard. The behaviour of the focal lizard was recorded for 15 minutes. Next, a male model was carried towards the focal individual and placed on a perch at a distance less than 5 metres from the focal lizard. After placing the model on a perch where male lizards usually display, the observer returned back to the video camera. We began the trial only when the focal lizard was in a position such that the model was in direct line of sight of the lizard. Observations of the focal lizard were carefully recorded on the video camera for a period of up to 30 minutes. Any adult conspecific lizard appearing within 5 m of the focal lizard during the experiment was noted. If the model or focal lizard was disturbed by other lizards in the area, then the trial was aborted. The order in which the models (orange, yellow) were presented to a tagged lizard was randomised. A gap of a minimum of 4 days was given between different model trials to avoid habituation to models by the focal lizard. There were at least two individual models printed and painted of each type to offset the effect of the model.

Assessing mate availability

To assess mate availability, lizards were followed and observed through regular surveys. The surveys involve the observer walking along a fixed path through the sheet rocks and making note of all the lizards in the study area. Tagged individuals can be identified with images shot from a camera without disturbing the lizard. The lizards were noted down onto a map, which was later geo-referenced to get the exact location of the lizard. Colour status of males and gravidity status of females was noted for each lizard. Each area was surveyed twice within the same session to maximise sightings, however only previously unrecorded tagged lizards were noted the second time to avoid duplicating records. The survey data on lizard distributions let us estimate mate availability.

Data Analysis

Model presentation trials were transcribed using the software BORIS (Friard and Gamba, 2016).

An elaborate ethogram of all behaviours was created. We identified sexual signalling behaviours as those associated with courtship interactions from previous studies (Deodhar and Isvaran, 2018; R. S. Radder et al., 2006; Ranade, 2020), and ad-libitum observations of courtship behaviour in the field. The behaviours "crouch" and "tail-raise," along with the colour states "red-head" and "dark-body," (Fig:2, supplementary video) were further analysed as they were associated with sexual signalling. The behaviours are described in greater detail in the supplementary section. The analysis was carried out in R (Version 4.1.3)(Team, 2021).

Measuring social condition of lizards

The spatial information from regular surveys was geo-referenced from maps with the help of Qgis software. The tagged female lizards we sampled through assays were followed through regular surveys. We considered survey data 30 days prior to the day of a trial to measure the provisional mate availability for the focal female for that trial. Females were usually recorded in multiple surveys over the 30 day window. To each spatially marked sighting of a focal female, we used a 20 m radius buffer (using gbuffer function in R, package rgeos) and recorded the number of males that were observed for each unique sighting of the focal female. The 20 m was considered based on home range sizes from earlier studies. We measured the number of unique males each female was exposed to, in the 30-day period before the assay. For each sighting of the female, in the surveys, we were able to assign a total number of males and since the males were tagged, we were able to identify the number of unique males a female was exposed to, and this was considered as a proxy for mate availability.

Breeding season

The breeding season lasts from May till October. Trials were predominantly carried out between June and October. June and July were considered as early breeding season and August and September were considered late breeding season.

Dance or disappear: Strategic sexual signalling in female Peninsular rock agama

Data files

Abstract

README: Dance or disappear: Strategic sexual signalling in female Peninsular rock agama

Description of the data and file structure

Files and variables

File: behaviour_data_Procb.csv

Variables

File: Behaviourdata_mean_CI.csv

Variables

Code/software

Access information

Methods