Results
Maximum and minimum critical thermal limits were differentially affected by the thermal environment experienced by flies.CTmax was significantly higher in flies from parental generation reared at variable thermal environment (V) than those reared at constant temperature (C), whileCTmin was not different between parental thermal environments (Figures 2A, 2B and Supplementary Figure S1). Interestingly, whereas CTmax was significantly higher in females than males, CTmin did not vary between sex (Table 1, Supplementary Figure S1). With regards to the F1, the thermal environment experienced by parental generation affected the critical thermal limits of their offspring (Table 1, Figure 3). More specifically, F1 flies raised under variable conditions whose parents were also maintained in this environment (i.e., VV) exhibited a significantly higher CTmax than all other F1 groups (including flies from CV, see Figure 2 and Supplementary Figure S1). Also, CTmax of F1 flies reared under constant environment, whose parents were reared in the alternative treatment (i.e., VC) had significant lower CTmaxthan F1 flies from a constant environment whose parents were maintained in a constant environment (CC). Besides, cold tolerance of F1 flies reared in a variable thermal environment (i.e., VV and CV) were lower than those reared at C (i.e., CC and VC).
Changing now to demographic descriptors of fitness, both survival and fecundity per female was lower in P flies reared at variable environments (Figure 3, detailed analyses in Supplementary Table S2), resulting in substantially lower \(R_{0}\) and \(T_{g}\) in C in comparison to V (Figure 2, Supplementary Table S3). In contrast, the variation in survival and fecundity was substantially reduced in the F1 regardless of the thermal treatment, which suggests some sort of compensation across generations (Figures 2 and 3). In fact, flies from the second generation exposed to a variable thermal environment showed a significant increase of \(R_{0}\) and \(T_{g}\) values compared with their parents (Figure 2)\(.\) Effects of the parental environment were still evident, however, with both \(R_{0}\) and \(T_{g}\ \)being on average lower in F1 lines derived from parents subjected to a variable environment (i.e., VC and VV groups), indicating that the apparent compensatory response to a stressful parental environment was only partial (Figure 2). Interestingly, flies whose parents were maintained at a constant temperature (CC and CV) did not show differences in\(R_{0}\) and \(T_{g}\) between them or their parents regardless of the thermal environment in which they were raised (Figure 2).
Results from our GAM analyses in conjunction with a model comparison approach support the observations listed above (Table 2, Supplementary Table S3). Thermal treatment had a major effect in these fitness components, explaining up to 53.6% and 34.9% of the variance in\(R_{0}\) and \(T_{g}\), respectively, after controlling for density (density effects on demographic variables are represented in Supplementary Figure S2). Furthermore, the models with the lowest AIC in both cases included the treatment factor (see Methods), which encapsulates the thermal environment of the parents (TP ), the offspring (TF1 ) and their interaction, and resulted in a model with a better fit than those where these factors were included separately (Table 1).
Interestingly \(R_{0}\) and CTmax were negative and significantly correlated (r=-0.8, T=-2.69, df=4, P=0.027 ) but, we did not find significant correlations among\(T_{g}\)and CTmax (r=-0.55, T=-1.32, df=4, P=0.12 ).