Bet hedging in desert winter annual plants: Optimal germination strategies in a variable environment

Background/Question/Methods

Biological bet hedging refers to strategies that sacrifice mean fitness in order to reduce variance in fitness, thus maximizing fitness in a variable environment.  Delayed germination is a classic example of bet hedging, in which a fraction of a plant’s seeds remain dormant as a hedge against the risk of low survival or fecundity of germinating seeds.  The value of this strategy depends on post-germination conditions, such as weather and competition as well as the probability of surviving while in the seed bank.  Yet, many studies lack a detailed understanding of seed mortality and density-dependence of vital rates, and even fewer provide rigorous tests for bet hedging.  Here, we asked whether delayed germination functions as a bet hedging strategy using 20 years of detailed demographic data and soil seedbank studies in a desert annual community.  Using both density-independent and dependent models, we estimated fitness for each species across a range of germination fractions and identified optimal germination fractions from each model.  We then compared predicted optimal germination fractions to observed germination fractions to determine the precision of bet hedging and how it varies among species.   

Results/Conclusions

Our analyses indicate that germination fractions below 1 have lower arithmetic mean fitness and variance in fitness, but have higher geometric mean fitness, thus meeting the criteria to function as a bet hedging strategy.  Predicted germination fractions from the density-dependent model corresponded better with observed values, indicating that delayed germination provides an escape from competition as well as environmental stochasticity.   These patterns of variation in germination among species have been linked to mechanisms that contribute to coexistence, which has implications for understanding community dynamics and patterns of biodiversity in variable environments.