Predicting population- and species-level responses to future environments is imperative under global climate change. This challenge is exacerbated by simultaneous changes in multiple climate drivers that are projected to produce conditions that have no contemporary analogue. To assess and overcome this barrier to population forecasting, we imposed a four-year factorial climate manipulation of the timing of snowmelt (advanced vs. control) and ambient temperature (warmed by open-top chamber vs. control) on individuals of a long-lived dioecious plant species, Valeriana edulis. By measuring the sex- and size-specific demographic responses of plants to these single and joint changes in climate, we were able to (i) measure the responses of individual fitness components across the lifecycle and (ii) parameterize an Integral Projection Model to simulate how multivariate climate change will affect population composition (sex and size structure) and dynamics (per-capita population growth rate).
Results/Conclusions
Our climate manipulations produced conditions that approximated the projected climate of Colorado’s Rocky Mountains in the year 2100: we advanced snowmelt by 13.1 ± 2.4 days and increased ambient air temperature by 1.9 ± 0.4°C relative to controls. Plant fitness components and population dynamics responded strongly to these climatic changes. Advancing snowmelt alone modestly increased the survivorship of both sexes, male flowering, and female seed production. Warming chambers alone increased male and decreased female survivorship, reducing the female survival advantage under control conditions. However, when imposed together the climate treatments had an additive effect on female seed production, a sub-additive effect on male flowering probability, and a super-additive effect on the survival of both sexes. These demographic responses drove changes in the equilibrium population sex ratio and per-capita population growth rate (λ). Equilibrium male frequency was reduced by advancing snowmelt (–3.8%) but was increased by warming chambers (+4.9%). The joint impact of the treatments on the equilibrium sex ratio was super-additive (+7.4%). In contrast, both individual climate drivers increased λ, with a sub-additive interaction (snowmelt alone: +1.7-fold; warming chambers alone: +2.7-fold; both: +2.6-fold). Retrospectively, our results from the joint climate manipulation explain the observed patterns of sex ratio variation across Valeriana’s elevation (=climatic) range and in response to climate change since the late 1970s. Prospectively, however, the profusion of interactions between climate drivers—from antagonistic to synergistic—on plant responses—from individuals to populations—suggests that a reliance on biological responses to past or contemporary climate will be insufficient to accurately forecast responses to non-analogue future climates.