Skip to main content
Dryad

Mate preferences act independently on different elements of visual signals in Heliconius butterflies

Cite this dataset

Smith, Sophie Helen et al. (2024). Mate preferences act independently on different elements of visual signals in Heliconius butterflies [Dataset]. Dryad. https://doi.org/10.5061/dryad.7d7wm383q

Abstract

Mating cues are often comprised of several elements, which can act independently, or in concert to attract a suitable partner. Individual elements may also function in other contexts, such as predator defence or camouflage. In Heliconius butterflies, wing patterns comprise several individual colour pattern elements, which advertise the butterflies’ toxicity to predators. These wing patterns are also mating cues and males predominantly court females that possess the same wing pattern as their own. However, it is not known whether male preference is based on the full wing pattern or only individual pattern elements. We compared preferences of male H. erato lativitta between female models with the full wing pattern and those with some pattern elements removed. We found no differences in preference between the full wing pattern model and a model with pattern elements removed, indicating that the complete composition of all elements is not essential to the mating signal. Wing pattern preferences also contribute to pre-mating isolation between two other Heliconius taxa, H. erato cyrbia and H. himera, therefore, we next compared preferences for the same models in these species. H. erato cyrbia and H. himera strongly differed in preferences for the models, potentially providing a mechanism for how pre-mating isolation acts between these species. These findings suggest that contrasting levels of selective constraint act on elements across the wing pattern.

README: Mate preferences act independently on different elements of visual signals in Heliconius butterflies

https://doi.org/10.5061/dryad.7d7wm383q

Preference for multiple pattern elements: data analysis

The purpose of this analysis was to investigate male preference for multiple pattern elements between species of Heliconius butterflies. Data in this analysis come from behaviourial trials conducted between October 2021 and February 2022, in which courtship behaviours of 3 taxa of male Heliconius butterflies towards 3 types of female models were recorded. The models in the trials were manipulated to remove some elements of the wing pattern, to see whether there was a difference in male behaviour based on different wing pattern elements. Behaviours were recorded per minute of the trial, with only one behaviour per male per model recorded per minute. Counts of the number of behaviour per male per trial are analysed to find if males preferred particular models.

Description of the data and file structure

There are 2 files associated with this analysis: a raw data file (Wing_element_preference_count_data.csv) and an RMarkdown detailing the analysis (Wing_element_preference_analysis.Rmd).

Data file: Wing_element_preference_count_data.csv

The variables in my full dataset are as follows:

  • male_id = identity of the male for which behaviours are recorded
  • male_species = species of the male for which behaviours are recorded (lat = H. erato lativitta, cyr = H. eraot cyrbia, him = H. himera)
  • male_origin = origin of male (Wild = collected from the wild, Reared = laid and reared from an egg at Ikiam, Ismael = laid at insectaries belonging to Ismael Andas, and transported to Ikiam as pupa)
  • trial_id = identifying number of the trial in which behaviours are recorded
  • trial_number = number of trials the male has already been used in
  • weather = categorised description of the weather at the start of the trial, matched to overall light intensity (overcast = complete cloud cover, over50cloud = over 50% cloud cover, but less than 100%, under50cloud = under 50% cloud cover but more than 0, clear = no cloud cover)
  • temperature = temperature at start of trial (°C)
  • cage = identity of the cage in which the trial took place (experimental cage #1 = 1, experimental cage #2 = 2, experimental cage #3 = 3)
  • lat_id = identity of H. erato lativitta male used in trial
  • cyr_id = identity of H. erato cyrbia male used in trial
  • him_id = identity of H. himera male used in the trial
  • model_set = identity of the set of female models used in the trial
  • model_order = position of three models in the trial (3-letter code gives the position of the models in the order 'POSITION1 POSITION2 POSITION3')
  • start_time_min = minutes after 08:00 that the trial began
  • model_type = type of model behaviours are directed towards (A = full H. e. lativitta wing pattern, B = red patches only, C = yellow patches only)
  • approach_all = total number of approaches towards a model per trial, including approaches which resulted in hover behaviours ("hover" + "approach_only")
  • approach_only = total number of approach behaviours towards a model per trial, which did not result in a hover behaviour
  • hover = total number of hovering courtship behaviours towards a model per trial

Data file: Wing_element_preference_analysis.Rmd

This script contains the full workflow of analysis and visualisation of the dataset. This script can be run in R (4.3.0). Comments throughout the script describe packages used, loading of data, data exploration, analysis workflow (Likelihood Ratio Tests and multiple comparisons tests) and visualisation through figures.

Methods

Butterfly collection and rearing.

All experiments took place in insectaries at Universidad Regional Amazónica Ikiam, Tena, Ecuador (Figure 1, red diamond) between November 2021 and January 2022. H. erato lativitta were collected in the southeast of Napo province, Ecuador, whereas H. erato cyrbia and H. himera came from insectary populations established from individuals collected in southwest Ecuador (Figure 1). Butterfly stocks were maintained in separate 2 x 2 x 2m cages for each sex and species, with 10-50 individuals per cage. Each cage was supplied with nectar-bearing flowers, and ~20% sugar solution with supplementary pollen for additional nutrition (Gilbert 1972), and Passiflora punctata shoots for oviposition in female cages. Caterpillars from these stocks were reared indoors and fed ad libitum on Passiflora punctata.

Mounted female wing models.

To make the female wing models, wings from sacrificed H. erato lativitta females were removed and subsequently washed for 5 minutes in dichloromethane to remove pheromones (Darragh et al. 2017). The wings were then manipulated by coloring over a patch of color on both sides of the wing with a black marker pen (COPIC Ciao Black 100 (Too Marker Products Inc.)) to generate one of three distinct wing patterns: ‘intact’, ‘red-only’ or ‘yellow-only’ as shown in Figure 2A-C. The remaining unmanipulated part of the wing was colored with a colorless marker pen with the same solvent (COPIC Ciao Colorless Blender 0 (Too Marker Products Inc.). Wings were then glued to white card, using a thin paper strip to allow flapping motion, mounted onto a black foam strip that acted as the butterfly’s body and allowed to dry for at least 2 hours before use.

Male preference experiments.

Our experimental design involved assessing the preference of male H. erato lativitta, H. erato cyrbia and H. himera between models of H. erato lativitta females with three types of wing pattern manipulation. Trials took place between 08:00 and 17:00, and males were tested at least 5 days after eclosion. One male each of H. erato lativitta, H. erato cyrbia and H. himera were tested together in the same trial, and the combination of three males tested together was randomized between trials. No interactions between males that would impede a male’s ability to interact with the models were observed. Males were released into experimental cages between 1 and 24 hours before the trial to acclimatize, during which they were provided with nectar-bearing flowers and sugar water solution, as above, but these were removed during the trial.

During 30-minute trials, males were presented with a set of three female models (intact, red-only, and yellow-only) mounted on a circular frame, 90cm in diameter, suspended 1m from the ground (Figure 1B). The models were randomly placed at one of three positions on the frame, equally spaced 60cm away from each other. A string attached to the center of the frame was pulled continuously throughout the trials to simulate wing movement (Klein & de Araújo 2010). Four different sets of models were used to account for individual variation between models. At the beginning of each trial, we recorded the date, time, weather, (recorded as ‘clear’, ‘under 50% cloud cover’, ‘over 50% cloud cover’ or ‘overcast’), model set, and temperature.

Two types of behavioral interactions between the males and the models were recorded during the trial, using the Field Notes app (Neukadye). These were ‘approach’ and ‘courtship’. We defined ‘approach’ as a male directing its flight towards and coming within 10cm of the model. ‘Courtship’ was defined as ‘approach’ but additionally included hovering above the model for at least 1 second. Accordingly, if a male courted the model, this was also recorded as an ‘approach’. Both behaviors were scored independently once per minute, for each model type and male, and counts of each behavior per trial were used in the subsequent analysis.

Preference trials for each male were repeated 5 times for each model set to account for individual variation and differences in male activity levels between different day times and weather conditions. Each male was tested on multiple model sets when possible, to account for variation in female model quality, though this was not possible for cases in which males died between tests on different sets, therefore most males (60%) were tested with only one set. Each model set was used for testing at least two different males of each species. Males were tested a maximum of 2 times per day and remained in experimental cages between the 5 repeated trials per model set, in the presence of other males.

Statistical analysis.

To determine whether males responded differently to the three model types, we fitted generalized linear mixed models (GLMM) with Poisson error distribution, implemented in the lme4 package (Bates et al. 2015) in R (R Core Team 2021). The dependent variable was either ‘approach’ or ‘courtship’ encoded as counts of minutes per trial in which a male did the behavior towards a model. Saturated models included male species, model type and their interaction as dependent variables. Weather was also included as a fixed factor with four levels: ‘100% overcast’, ‘>50% overcast’, '<50% overcast’ and ‘clear’ which correlated strongly with temperature (see supplementary materials). Male ID and trial nested within model set were included as random variables in all models to account for repeated measures. For some models tested, we implemented model optimizer BOBYQA (Powell 2009) to overcome convergence warnings associated with the complex structure of the models. When necessary, we subsequently checked for deviation between log likelihoods of different optimizers using the command allFit from the lme4 package (SM Table 1). Significance of fixed factors was determined using likelihood ratio tests (LRTs) by comparing models with and without individual terms. Effect sizes were calculated as the estimated marginal means (EMM) of the minimum adequate model using the emmeans package (Lenth 2021). We generated ternary plots of the raw data to visualize differences in ‘approach’ and ‘courtship’ by each male type, towards the three models. All analysis scripts are included in the supplementary materials.

Funding

European Research Council, Award: 851040