Testing a general theory for flowering time shift as a function of growing season length

Understanding the effects of climate change on the evolution of phenological timing, such as start of flowering, is of major interest because phenology is critical for fitness of populations, and underpins many ecological dynamics. Much recent research has focused on the correlation between phenological timing and the arrival of spring. However, the evolutionarily optimal seasonal timing should depend also on the duration of the growing season, within which entire annual life cycles must unfold. Optimal energy allocation theory can explicitly address life-history scheduling in a seasonal environment, and be used to predict how scheduling should adaptively respond to seasonality shifts. Here we extend a seminal theoretical framework for perennial plant scheduling by Iwasa & Cohen (1989), and predict a specific nonlinear relationship between growing season length and optimal flowering time expressed as number of days after the start of the growing season. We tested and found strong support for this a priori prediction in two independent common-garden experiments with purple loosestrife (Lythrum salicaria) and European goldenrod (Solidago virgaurea) populations sampled along latitudinal gradients in Sweden. Climate warming is commonly associated with changes in both the start and the duration of the growing season. Considering both effects, our findings suggest that as springs start earlier and growing seasons lengthen, shifts in optimal flowering time expressed as calendar date may initially stall before accelerating, potentially explaining observed variation in phenological shifts across systems. More broadly, we show how mechanistic life history theory can advance understanding of phenological change beyond correlative conclusions. SIGNIFICANCE STATEMENTThe optimal timing for a plant to flower depends both on when spring begins and on the total length of the growing season--both shifting with climate change. However, the adaptation of flowering time to changes in growing season length is much less explored than to an advancing spring. Building on existing theory, we predict a nonlinear relationship between growing season length and optimal flowering time. Data on two species (purple loosestrife and European goldenrod) grown in common gardens strongly support this prediction. Our results demonstrate why species can vary in their observed phenological responses to climate change. More broadly, we highlight the power of mechanistic life history theory for explaining and predicting phenological shifts.