model id

model

res df

res dev

AIC


1

p.insem ~ sperm * wing

12

16.59

44.38

2

p.insem ~ sperm + wing

13

16.60

42.40

3

p.insem ~ wing

14

23.84

47.63

4

p.insem ~ sperm

14

17.97

41.77

5

p.insem ~ 1

15

23.88

45.68

effect

comparison

Δdf

Δdev

p

sperm*wing

1 vs. 2

1

0.02

0.903

sperm

2 vs. 3

1

7.23

0.007

wing

2 vs. 4

1

1.37

0.242

model id

model

res df

res dev

AIC

6

p.motile ~ sperm * wing

12

15.38

45.83

7

p.motile ~ sperm + wing

13

15.96

44.41

8

p.motile ~ wing

14

21.02

47.47

9

p.motile ~ sperm

14

17.34

43.79

10

p.motile ~ 1

15

21.16

45.61

effect

comparison

Δdf

Δdev

p

sperm*wing

6 vs. 7

1

0.58

0.447

sperm

7 vs. 8

1

5.06

0.025

wing

7 vs. 9

1

1.38

0.240

model id

model

res df

res dev

AIC

11

p.ovip ~ sperm * wing

12

14.42

44.90

12

p.ovip ~ sperm + wing

13

15.56

44.05

13

p.ovip ~ wing

14

21.24

47.72

14

p.ovip ~ sperm

14

17.99

44.47

15

p.ovip ~ 1

15

21.73

46.21

effect

comparison

Δdf

Δdev

p

sperm*wing

11 vs. 12

1

1.15

0.284

sperm

12 vs. 13

1

5.68

0.017

wing

12 vs. 14

1

2.43

0.119

 Experiment 3: GLM with binomial errors was used to model the proportion of inseminated females (p.insem), the proportion of females with at least one motile sperm in their spermathecae (p.motile), and the proportion of ovipositing females (p.ovip) as a function of the sire family's mean sperm length (sperm), the mean wing length (wing), and their interaction (sperm*wing). Shown for each model are the residual degrees of freedom (res df), the residual deviance (res dev), and the Akaike information criterion (AIC). Shown for the log likelihood ratio tests of the effects are the change in the degrees of freedom (Δdf), the change in the residual deviance (Δdev), and the statistical significance of the effect (p). The best model according to AIC and the statistically significant effect sizes are shown in bold.