Table 4.

Reproductive success as a function of sperm and wing length in A. gambiae males (Parameter estimates)

model 2: p.insem ~B0 + B1(z.sperm) + B2(z.wing)
param	estimate	s.e.	Mean fitness (p.insem)	% Δ
B0	0.95	0.297	exp(B0)/(1+exp(B0)) = 0.72
B1	-0.84	0.330	exp(B0 + B1)/(1+exp(B0 + B1)) = 0.53	-27.0
B2	0.38	0.328	exp(B0 + B2)/(1+exp(B0 + B2)) = 0.79	9.6
model 7: p.motile ~B0 + B1(z.sperm) + B2(z.wing)
param	estimate	s.e.	Mean fitness (p.motile)	% Δ
B0	-0.41	0.265	exp(B0)/(1+exp(B0)) = 0.40
B1	-0.69	0.325	exp(B0 + B1)/(1+exp(B0 + B1)) = 0.25	-37.3
B2	0.39	0.337	exp(B0 + B2)/(1+exp(B0 + B2)) = 0.49	24.0
model 12: p.ovip ~B0 + B1(z.sperm) + B2(z.wing)
param	estimate	s.e.	Mean fitness (p.ovip)	% Δ
B0	-0.34	0.265	exp(B0)/(1+exp(B0)) = 0.42
B1	-0.73	0.329	exp(B0 + B1)/(1+exp(B0 + B1)) = 0.25	-38.8
B2	0.52	0.343	exp(B0 + B2)/(1+exp(B0 + B2)) = 0.54	30.9

The parameters (B₀, B₁, and B₂) and their estimates and standard errors (s.e.) from GLM models 2, 7 and 12 in Table 3 after standardizing the sire family's mean sperm length (z.sperm) and mean wing length (z.wing) to z-scores. The logit-link function was used to back-calculate the mean fitness (p.insem, p.motile, p.ovip) when the average sire family (z.sperm = 0, z.wing = 0) increased its mean sperm length by one standard deviation while mean wing length was held constant (z.sperm = 1, z.wing = 0), and vice versa (z.sperm = 0, z.wing = 1). The percent change in fitness (% Δ) was calculated relative to the mean fitness of the average sire family.