Plant evolution along the ‘fast–slow’ growth economics spectrum under altered precipitation regimes

Abstract

Plants have evolved different strategies to withstand drought. In general, these strategies can be defined along a plant economics spectrum, which classifies plants depending on whether their growth rate is fast or slow, where fast growth is associated with high mortality, high water use, and high sensitivity to drought. Which strategy along this economy spectrum will be selected under different precipitation regimes is an open question. We address this question with a minimal soil–plant model in which a single plant economy trait related to growth rate characterizes the plant strategy. This generic and dimensionless trait influences both recruitment and mortality, but not background mortality. We explore the evolution of this trait by quantifying its effects on birth, mortality, and transpiration rates. Furthermore, we explore the influence of direct plant density dependence acting on recruitment and mortality, in addition to the indirect density dependence caused by plant feedback on soil water content.We show that: (1) Increasingly fast-growing plants always evolve under increasing background mortality. (2) When soil water only depends on plant density and is independent of precipitation and abiotic water losses, the strategy minimizing soil water content is an evolutionarily stable strategy (ESS); i.e., the evolutionary outcome is a tragedy of the commons (Hardin, 1968). (3) When precipitation, abiotic water losses and trait dependent transpiration determine soil water content, the ESS lies between the strategy maximizing plant density and that minimizing soil water content; i.e., no tragedy of the commons occurs. (4) With a deterministic precipitation model and density dependence acting directly only on recruitment, higher precipitation promotes the evolution of faster plants. The opposite result is found when density dependence is acting directly only on mortality. (5) Similar trends in the economy trait are observed when forcing the model with stochastic precipitation events.