Gene expression is known to be highly responsive to the environment and important for adjustment of metabolism but there is also growing evidence that differences in gene regulation contribute to species divergence and differences among locally adapted populations. However, most studies so far investigated populations when divergence had already occurred. Selection acting on expression levels at the onset of adaptation to an environmental change has not been characterized. Understanding the mechanisms is further complicated by the fact that environmental change is often multivariate, meaning that organisms are exposed to multiple stressors simultaneously with potentially interactive effects. Here we use a novel approach by combining fitness and whole-transcriptome data in a large-scale experiment to investigate responses to drought, heat and their combination in Tribolium castaneum . We found that fitness was reduced by both stressors and their combined effect was almost additive. Expression data showed that stressor responses were acting independently and did not interfere physiologically.  Since we measured expression and fitness within the same individuals, we were able to estimate selection on gene expression levels. We found that variation in fitness can be attributed to gene expression variation and that selection pressures were environment dependent and opposite between control and stress conditions. We could further show that plastic responses of expression were largely adaptive, i.e.  in the direction that should increase fitness.

We used a T. castaneum strain (Cro1) which was collected from a wild population in 2010 and adapted to standard control conditions (CTL: 33 °C, 70% relative humidity (r.h.)) since then. To assess the fitness changes caused by stressful environmental conditions, we exposed the beetles to a drought, a heat, and a combined heat-drought treatment (conditions: Dry (=D) : 33°C, 30% r.h.; Hot (=H): 37 °C, 70 % r.h; Hot-Dry (=HD): 37°C, 30% r.h.). Parents of the experimental beetles were reared in control conditions and beetles used for this fitness assay were transferred to treatments at the egg stage. Males and females were separated at the pupal stage. We mated each virgin female with one unrelated male from the same condition in a 15 mL tube with 1 g medium (organic wheat flour mixed with 10% organic baker's yeast). The male was removed after 24 h. Females were removed from the tubes after one week of egg laying, and 9 g medium was added to provide food for the developing offspring. After five weeks the number of offspring was counted. At this time, all offspring had reached the adult stage. Some females did not produce any offspring. Data was collected in 2015.

First column: female ID = female that produced offspring; second column: male ID = mating partner of female, father of offspring; third column: number of adult offspring; fourth column: condition in the climate chamber: CTL=control, D=dry, H=hot, HD=hot-dry