Table 1.
Trait | Linear effects | Nonlinear effects | |||
---|---|---|---|---|---|
P | C | P × P | C × C | P × C | |
(A) Males | |||||
3H2B | |||||
Coefficienta ± SE | −0.243 ± 0.054 | 0.539 ± 0.054 | 0.024 ± 0.049 | −0.172 ± 0.027 | −0.131 ± 0.104 |
t 239 | 4.546 | 10.068 | 0.482 | 6.279 | 1.265 |
P | 0.0001 | 0.0001 | 0.630 | 0.0001 | 0.207 |
2MT | |||||
Coefficient ± SE | −0.227 ± 0.053 | 0.557 ± 0.053 | 0.030 ± 0.049 | −0.170 ± 0.027 | −0.188 ± 0.102 |
t 239 | 4.278 | 10.484 | 0.608 | 6.289 | 1.834 |
P | 0.0001 | 0.0001 | 0.544 | 0.0001 | 0.068 |
4E2M | |||||
Coefficient ± SE | −0.131 ± 0.061 | 0.495 ± 0.061 | 0.011 ± 0.058 | −0.149 ± 0.032 | −0.254 ± 0.121 |
t 239 | 2.163 | 8.161 | 0.197 | 4.670 | 2.097 |
P | 0.032 | 0.0001 | 0.844 | 0.0001 | 0.037 |
(B) Females | |||||
Clutch size | |||||
Coefficient ± SE | 0.105 ± 0.070 | 0.351 ± 0.070 | −0.079 ± 0.073 | −0.142 ± 0.044 | −0.356 ± 0.113 |
t 226 | 1.489 | 5.004 | 1.082 | 3.263 | 3.151 |
P | 0.138 | 0.0001 | 0.280 | 0.001 | 0.002 |
Gestation timeb | |||||
Coefficient ± SE | 0.022 ± 0.072 | 0.227 ± 0.072 | 0.034 ± 0.078 | −0.039 ± 0.046 | 0.067 ± 0.120 |
t 226 | 0.308 | 3.144 | 0.436 | 0.842 | 0.560 |
P | 0.759 | 0.002 | 0.663 | 0.401 | 0.576 |
The linear regression coefficients (i.e., P and C) describe the linear slope (given by β i) of the relationship between nutrient intake and the response variable, whereas the quadratic regression coefficients (i.e., P × P and C × C) describes the curvature (given by γ ii) of this relationship, with a negative γ ii indicating a convex relationship (i.e., a peak on the response surface) and a positive γ ii indicating a concave relationship (i.e., a trough on the response surface). The correlational regression coefficient (i.e., P × C) describes how the covariance between the two nutrients (γ ij) influences the response variable, with a negative γ ij indicating that a negative covariance between nutrients increases the response variable and a positive γ ij indicating that a positive covariance between nutrients increases the response variable. Full details of this approach are provided in Lande and Arnold (1983).
The sign of standardized gestation time was reversed for analysis to make this response variable directly comparable to other traits (i.e., a short gestation time is good for reproductive potential).