The timing of different life history events are often correlated, and selection might only rarely be exerted independently on the timing of a single event. In plants, phenotypic selection has often been shown to favour earlier flowering. However, little is known about to what extent this selection acts directly vs. indirectly via vegetative phenology, and if selection on the two traits is correlational. We estimated direct, indirect and correlational phenotypic selection on vegetative and reproductive phenology over three years for the perennial herb Lathyrus vernus. Direct selection favoured earlier flowering and shorter timespans between leaf-out and flowering in all years. However, early flowering was associated with early leaf-out, and the direction of selection on leaf-out day varied among years. As a result, selection on leaf-out weakened selection for early flowering in one of the study years. We found no evidence of correlational selection. Our results highlight the importance of including temporally correlated traits when exploring selection on the phenology of seasonal events.

Phenotypic plasticity can mask population genetic differentiation, reducing the predictability of trait-environment relationships. In short-lived plants, reproductive traits may be more genetically determined due to their direct impact on fitness, whereas vegetative traits may show higher plasticity to buffer short-term perturbations. Combining a multi-treatment greenhouse experiment with global field observations for the short-lived Plantago lanceolata, we 1) disentangled the genetic and plastic responses of functional traits to a set of environmental drivers and 2) assessed the utility of trait-environment relationshisps inferred from observational data for predicting genetic differentiation. Reproductive traits showed distinct genetic differentiation that was highly predictable from observational data, but only when correcting traits for differences in their (labile) biomass component. Vegetative traits showed higher plasticity and contrasting genetic and plastic responses, leading to unpredictable trait patterns. Our study suggests that genetic differentiation may be inferred from observational data only for the traits most closely related with fitness.