Life-history switch points are ubiquitous in nature. Energy is allocated to different physiological functions before and after the switch. Optimal energy-allocation models emphasize the optimal switch strategy depending on physiological and environmental constraints, explaining across-population variability. We propose an Adaptive Dynamics model that also explains within-population variability under disruptive selection.
We model the timing of metamorphosis in a annual semelparous organism. Larvae compete for resources and allocate energy to growth until the switch, when they emerge as adults. This first phase increases size, thus potential fertility. Adults allocate the energy stocked in their body mass to reproduction until they die at the end of the season. This second phase increases realized fertility and reproductive success, thus the number of new larvae in the following season.
Results/Conclusions
The simple trade-off between growth and reproduction in the Adaptive Dynamics setting is able to describe a complex set of scenarios for the evolution of time of metamorphosis. Evolutionary bi-stability depending on the ancestral condition describes across-population variability under the same environmental conditions, while disruptive selection is responsible for within-population variability. The model also predicts stabilizing selection towards an optimal time of metamorphosis depending on environmental conditions, thus also describing across-population variability in agreement with optimization models, as well as selection for a simple life-cycle with no switch point (e.g., direct development). The three mechanisms described in the model (growth, competition, and reproduction) have different effects on time of metamorphosis and its variability among and across populations.
The description of within-population variability is possible thanks to Adaptive Dynamics, that models evolution along an adaptive and density-dependent fitness landscape, where selection can become disruptive and generate diversity in the same environment. This was not possible using optimization approaches, where fitness is static and density-independent.