TABLE 2.
Comparison of GAMM models with different sets of predictor variables to predict age‐specific hazard (mortality) rates of D. suzukii experimental groups (see Table 1)
| Rank | Predictor variables in each model | ΔAIC | Weight |
|---|---|---|---|
| 1 | Morphotype*Sex | 0 | 1 |
| 2 | Morphotype | 12 | 0 |
| 3 | Morphotype*Reproductive status | 65 | 0 |
| 4 | Morphotype*Sex*Reproductive status | 78 | 0 |
| 5 | Sex | 117 | 0 |
| 6 | Reproductive status | 158 | 0 |
| 7 | Sex* Reproductive status | 184 | 0 |
Candidate models were ranked from best (rank = 1) to worst fit (rank = 7) according to the Akaike information criterion (AIC). Model weights (Weight) reflect the probability that model i is the best model, given the data and the set of candidate models. ΔAIC is the difference in AIC between the focal model and the model with the lowest AIC score. All candidate models included “vial” as random effect. The predictor variables considered for this analysis were the morphotype (WM vs. SM), the sex (female vs. male), the reproductive status (mated vs. virgin and precold mated vs. postcold mated), and all possible interactions between these variables.