Optimal phenotypic plasticity in a stochastic environment minimizes the cost/benefit ratio

TL;DR

This paper investigates how organisms can optimize phenotypic plasticity in fluctuating environments to minimize energy costs while maximizing fitness, revealing that a certain level of genetic diversity is maintained even near optimal plasticity.

Contribution

It introduces a combined numerical and analytical model to determine optimal phenotypic plasticity considering energy constraints and environmental variability.

Findings

Optimal plasticity balances energy costs and fitness benefits.

Genetic diversity persists even near optimal plasticity levels.

Species in stable environments tend to be less plastic.

Abstract

This paper addresses the question of optimal phenotypic plasticity as a response to environmental fluctuations while optimizing the cost/benefit ratio, where the cost is energetic expense of plasticity, and benefit is fitness. The dispersion matrix \Sigma of the genes' response (H = ln|\Sigma|) is used: (i) in a numerical model as a metric of the phenotypic variance reduction in the course of fitness optimization, then (ii) in an analytical model, in order to optimize parameters under the constraint of limited energy availability. Results lead to speculate that such optimized organisms should maximize their exergy and thus the direct/indirect work they exert on the habitat. It is shown that the optimal cost/benefit ratio belongs to an interval in which differences between individuals should not substantially modify their fitness. Consequently, even in the case of an ideal population, close to the optimal plasticity, a certain level of genetic diversity should be long conserved, and a part, still to be determined, of intra-populations genetic diversity probably stem from environment fluctuations. Species confronted to monotonous factors should be less plastic than vicariant species experiencing heterogeneous environments. Analogies with the MaxEnt algorithm of E.T. Jaynes (1957) are discussed, leading to the conjecture that this method may be applied even in case of multivariate but non multinormal distributions of the responses.