Frog Aggression

You are behavioural ecologist interested in territoriality and aggression in frogs. The brilliant-thighed poison frog Allobates femoralis relies on the presence of small pools of water for their reproduction. Males of this species engage in territorial competition to secure areas that contain these pools of water. It is unknown, however, at which point in the frogs’ development the territorial behaviour emerges. To test this, you traveled to a field site in eastern Ecuador and captured 50 A. femoralis of various developmental stages. You recorded each frog’s sex, measured each frog’s Snout-Vent Length (SVL; in mm), then subjected the frogs to a mirror test. In the mirror test, you recorded whether they expressed aggressive behaviours (e.g. lunges, bites, or grappling) towards their reflections on the mirror.

Note: This scenario involves a more advanced statistical analysis known as ‘Logistic Regression’, not covered in either Descriptive or Inferential Statistics. The general premise of Logistic Regression is to determine whether a predictor variable has an effect on the outcome of a categorical  response variable. In this scenario our categorical outcome is binary (no aggression, 0, versus aggression, 1). This makes this scenario more specifically a ‘Binomial Regression’, which is a subset of Logistic Regression. For a primer on how to conduct Binomial Regression in R, see:

https://bookdown.org/ndphillips/YaRrr/logistic-regression-with-glmfamily-binomial.html

1. Load the data. Print the SVL range for males, and the SVL range for females

2. Using the glm() command in R, determine whether there is an association between either sex or body size with aggressive behaviour in these frogs. Do not fit an interaction between body size and sex.

3. For both sex and body size, write an inferential statement based on the results of your binomial regression model that included both sex and body size.

4. Create a binomial regression model using only SVL as a predictor variable, and save it into an object called ‘mod’.

5. Using Base R graphics, plot the predicted values for a vector of theoretical snout-vent lengths of 5mm to 30mm, using the SVL Binomial Regression model
   • For an example of how to plot predicted values of logistic regression, see:
     https://www.geeksforgeeks.org/how-to-plot-a-logistic-regression-curve-in-r/

Files to download:

To download, right-click and press “Save File As” or “Download Linked File”

1. FrogAggro.csv

Laboratory and Institution or PI

Behavioural and Sensory Ecology Lab, Dr. James B. Barnett, Department of Zoology, Trinity College Dublin, Ireland

References and Further Reading

Chaloupka, S., Peignier, M., Stückler, S., Araya-Ajoy, Y., Walsh, P., Ringler, M., & Ringler, E. (2022). Repeatable territorial aggression in a Neotropical poison frog. Frontiers in ecology and evolution, 10, 881387.

Rodríguez, C., Fusani, L., Raboisson, G., Hödl, W., Ringler, E., & Canoine, V. (2022). Androgen responsiveness to simulated territorial intrusions in Allobates femoralis males: evidence supporting the challenge hypothesis in a territorial frog. General and comparative endocrinology, 326, 114046.