Patterns and fitness impacts of phenology shifts in pollination networks

Abstract

Phenology shifts are one of the most prevalent responses to climate change across taxa. Temperature increases have resulted in many species shifting the timing of seasonal behaviours such as bud-burst or nest-building. Interspecific variation in the level of response has given rise to previously interacting species diverging in their phenologies, leading to the possibility of temporal mismatch. One system in which this has been predicted, and in some cases evidenced, is pollination networks. There are several underdeveloped areas of understanding that restrict our ability to anticipate and accurately predict the impacts of phenology shift in pollination networks. There is limited research quantifying the demographic fitness impacts of mismatch, which hinders understanding of the risks of mismatch. The drivers of interspecific variation in phenological sensitivity to temperature are uncertain, limiting capacity to predict which species may undergo divergent rates of shift. Finally, current methods do not account for how phenology shifts may also change how species population are distributed through time, which affects predictions of mismatch. This thesis seeks to examine each of these knowledge gaps. First, a manipulative field experiment was conducted using a generalist plant to quantify how continuous variation in pollinator phenology impacts on seed-set. Pollinator abundance and species composition was found to vary significantly over the field season, resulting in large variation in seed-set depending on flowering phenology. Second, the phenology of a group of key pollinators, Syrphidae (hoverflies), was examined using a UK recording scheme from 1980 onwards. Syrphidae were generally found to be advancing their flight phenologies, although response was asymmetrical over their flight period and first date of flight advanced at a greater rate than peak abundance date or last date of flight. Life history traits that were used 8 to predict phenology response in Lepidoptera were applied to the data set but were not found to be predictive in Syrphidae, potentially indicating that these traits are not general predictors of phenology response in arthropods. Finally, simulations of phenology shift in pollination networks were run using an individual-based model in which the shape of population distributions were also allowed to shift in response to temperature. These predicted large rates of species loss in networks and demonstrated that accounting for the shape of phenology events results in larger rates of species loss, indicating that current predictions may underestimate risk. These findings advance understanding of phenology shifts by showing that they can have severe fitness impacts and that it is necessary to model phenology in a way that takes account of the shape of phenology events.