Interactions between temperature and nutrients determine the population dynamics of primary producers

Global change is rapidly and fundamentally altering many of the processes regulating the flux of energy throughout ecosystems, and although researchers now understand the effect of temperature on key rates (such as aquatic primary productivity), the theoretical foundation needed to generate forecasts of biomass dynamics and extinction risk remains underdeveloped. We develop new theory that describes the interconnected effects of nutrients and temperature on phytoplankton populations and show that the thermal response of equilibrium biomass (i.e. carrying capacity) always peaks at a lower temperature than for productivity (i.e. growth rate). This mismatch is driven by differences in the thermal responses of growth, death, and per-capita impact on the nutrient pool, making our results highly general and applicable to widely used population models beyond phytoplankton. We further show that non-equilibrium dynamics depend on the pace of environmental change relative to underlying vital rates and that populations respond to variable environments differently at high versus low temperatures due to thermal asymmetries. Here, we integrate the thermal responses of intracellular processes regulating phytoplankton growth to understand how temperature and nutrients collectively affect population-level responses. We develop an analytical representation of how the thermal response for phytoplankton carrying capacity (i.e. population equilibrium biomass) depends on nutrient uptake. Notably, a population's carrying capacity is thermally dependent such that it is always optimized at temperatures lower than that for the population's productivity (i.e. from a thermal performance curve), addressing a longstanding puzzle regarding the utility of thermal performance curves for population forecasting. Importantly, the underlying mechanisms driving differential thermal responses for performance and biomass ought to be prevalent across taxa, making this result applicable to more general population models.image