Self-referential mate choice in swordtails

• Mary and Ryan sent me data on 163 dichotomous female choice trials using X. nigrensis
• In each, a female chose between a small and large female
• the data are:
  • the time with each male
  • the standard lengths of the large male, small male, and female
  • the number of glides (n = 159 trials)
  • the number of up-downs (n = 49 trials)

• Just as an internal check, let's make sure we get the same result that Ryan got in his Animal Behavior paper: Larger females have a greater preference for large males.
• Throughout this presentation, I'll use both a female's preference score (time spent with large male relative to time spent in front of both males) and difference in association time spent near the large male relative to the small male.

# gross association time
nigrensis$prefLargeMale <- nigrensis$timeLargeMale - nigrensis$timeSmallMale
# preference score
nigrensis$prefScore <- nigrensis$timeLargeMale / (nigrensis$timeLargeMale+nigrensis$timeSmallMale)

• So we're able to replicate Ryan's previous results.

• Now we want to explore the possibility of a self-referential mate choice rule. Under this senario, there might be some optimal male size relative to a female's size at which preference peaks.
• First, I calculate this ratio for all trials and plot that data using a loess curve.

One thing we might want to know is whether there is a significant negative correlation between this ratio and female preference.

                            Estimate Std. Error t value  Pr(>|t|)
(Intercept)                   0.9041    0.09463   9.554 2.054e-17
nigrensis$maleToFemaleRatio  -0.1856    0.07558  -2.455 1.514e-02

The residuals are ugly though, and log-transforming the ratio doesn't help:

Instead, we might want to do a a Spearman's rank corrlation test, which doesn't make any assumptions about the residuals.

    Spearman's rank correlation rho

data:  nigrensis$prefScore and nigrensis$maleToFemaleRatio
S = 936889, p-value = 0.0001114
alternative hypothesis: true rho is not equal to 0
sample estimates:
    rho 
-0.2981 

This significant results holds whether we use preference score (shown in previous slide), or difference in association time.

cor.test(nigrensis$prefLargeMale,nigrensis$maleToFemaleRatio,method="spearman")

    Spearman's rank correlation rho

data:  nigrensis$prefLargeMale and nigrensis$maleToFemaleRatio
S = 901940, p-value = 0.001311
alternative hypothesis: true rho is not equal to 0
sample estimates:
    rho 
-0.2496 

Next, I repeat all the analyses I did previous, but using the difference in number of glides a female displays towards the large and small males.

There are also no significant relationships between female size or the male:female size ratio and the number of glides.

For fun, I also wanted to see whether there was evidence that a female had a stronger preference towards the larger male when the ratio between the large and the small male was greater.