Complex effects of elevated temperatures on consequences of phenological shifts
Our results show that the effects of phenological shifts varied with warming, but to what extent and why differed between hosts and parasitoids. While growth rates of both host species increased with warming, the declines in parasitism rates and increased host survival across phenological shifts did not scale with differences in growth rates. Thus, other factors, besides variation in growth rates among temperatures, community contexts, and species contribute to differences in effects of phenological shifts across treatments. Our dynamic model revealed that host-parasitoid interactions were less likely to persist in predicted global warming temperatures and identified that some host and parasitoid species were completely unable to support interactions under warming, even when parasitoid survival, host fecundity, and phenological relationships were ideal (Fig. 3; Fig S7, Fig. S9, Fig. S10). This highlights that physiological changes in elevated temperatures, beyond shifts in development rates, can considerably modify the effect of phenological shifts on species interactions. For example, regardless of the developmental relationship, warming may compromise immune systems or the ability to heal from injuries (i.e., oviposition punctures) (Catalán et al. 2012; Abram et al.2019), though other studies have found that hosts are more able to successfully defend themselves against immature parasitoids in warmer temperatures (Thomas & Blanford 2003; Wojda 2017). In addition, parasitoids usually show lower thermal tolerances relative to their hosts (Karban 1998; Fellowes et al. 1999; Zamani et al. 2007; Agostaet al. 2018; Thierry et al. 2019), both of which support our current findings. These discoveries have important implications for persistence of key trophic interactions and top-down control, as such interruptions to top down processes has been identified as key consequences of climate change in terrestrial and aquatic systems (Shurin et al. 2012; Velthuis et al. 2017; Derocleset al. 2018).