Abstract: The goal of this article is to contribute towards the conceptual and quantitative understanding of the evolutionary benefits for (microbial) populations to maintain a seed bank consisting of dormant individuals when facing fluctuating environmental conditions. To this end, we discuss a class of '2-type' branching processes describing populations of individuals that may switch between 'active' and 'dormant' states in a random environment oscillating between a 'healthy' and a 'harsh' state. We incorporate different switching strategies and suggest a method of 'fair comparison' to incorporate potentially varying reproductive costs. We then use this concept to compare the fitness of the different strategies in terms of maximal Lyapunov exponents. This gives rise to a 'fitness map' depicting the environmental regimes where certain switching strategies are uniquely supercritical.
Introduction
In biology, dormancy is often viewed as a "bet-hedging" strategy: by keeping some individuals inactive, a population can survive "harsh" periods (like an antibiotic treatment or a cold winter) that would otherwise wipe out all active members. However, maintaining a seed bank is costly—resources spent on dormant "seeds" are resources not spent on reproduction.
This paper addresses the central challenge: Under what specific environmental conditions and switching strategies does the benefit of dormancy outweigh its reproductive cost? While previous models existed, they were often deterministic or lacked a "fair comparison" framework—a way to compare populations that use the same total amount of resources but distribute them differently between active and dormant offspring.
Methods and Model Structure
The researchers employ 2-type Bienaymé-Galton-Watson processes in random environments (BGWPRE). The population consists of "active" (white) and "dormant" (grey) types. The environment fluctuates between "healthy" (high reproduction) and "harsh" (low reproduction) states based on a Markov chain.
As shown in Figure 1, active individuals can die, reproduce into two active offspring, or transition into a dormant state. Dormant individuals can die, resuscitate into an active state, or remain dormant. The fitness of these strategies is measured using the maximal Lyapunov exponent ($\phi$), representing the long-term exponential growth rate. A population is "super-critical" and likely to survive if $\phi > 0$.
Switching Strategies
The model compares four primary strategies:
- Responsive: Switching triggered by environmental cues.
- Stochastic (Spontaneous): Switching at random intervals, regardless of the environment.
- Prescient: A counter-intuitive strategy where individuals invest in dormancy during good times in anticipation of a crash.
- Mixed: A combination of the above.
To ensure results are due to strategy rather than raw birth rates, the authors use a "fair comparison" method, normalizing models so all strategies have the same expected total offspring production per unit of resource.
Key Insights and Results
The Frequency of Change Matters
There is no single "best" strategy. The optimal approach depends on environmental dynamics:
- Responsive switching is most effective when the environment changes slowly.
- Stochastic switching (bet-hedging) is superior in moderately fluctuating environments.
- Prescient switching—investing in seeds during the "healthy" phase—is optimal when the environment flips almost every generation, similar to annual plants facing predictable seasons.
Dormancy as a Life-Saver
The model demonstrates "strong fitness advantages," where a population without a seed bank goes extinct ($\phi < 0$), but a population with the right dormancy strategy becomes super-critical ($\phi > 0$) in the exact same environment. Furthermore, in certain moderate regimes, only a mixed strategy can prevent extinction, suggesting a form of "second-level bet-hedging."
Mathematical Rigor
The paper provides exact analytical solutions for the Lyapunov exponents of "rank-1" switching strategies (where outcomes depend only on the environment). For more complex "rank-2" strategies (where parentage matters), the authors established rigorous upper and lower bounds for fitness, providing a toolkit for estimating survival probabilities in complex systems.
Limitations and Future Work
The current model is limited by its binary environment (only "healthy" and "harsh") and the assumption of stationarity. Moving toward non-stationary models to account for climate change and incorporating competition between different dormancy strategies (adaptive dynamics) are identified as key areas for future research. Additionally, translating the discrete-time model to continuous-time birth-death processes would allow for modeling extremely rapid environmental fluctuations.
Summary updated