Plant trait networks reveal the ecological strategies of <i>Arabidopsis thaliana</i> along ontogeny

Abstract Plant trait networks (PTNs) quantitatively describe the trait correlation patterns, which constrain the dynamic responses and individual fitness of plants under climate change. Despite renewed attention directed toward intraspecific trait correlation within trait‐based ecology in the last decade, thorough descriptions of how trait correlation patterns within species respond to changing environments ontogenetically and influence ecological strategies are still lacking. In this study, a controlled experiment was conducted on Arabidopsis thaliana wild‐type (Columbia), with three treatment groups (control/drought/heat) over three stages of the life cycle (vegetative/inflorescence/reproductive stage). Thirty traits (aboveground biomass and seed biomass were not used in trait correlation analyses) were obtained to investigate how trait correlation patterns in particular life stages (three T‐PTNs constructed based on trait variation across treatments within life stages) and during whole‐life cycles (three L‐PTNs constructed based on trait variation across life stages in particular treatments) respond to stress. The results showed that traits were correlated along an axis within life stages, reflecting an economic trade‐off between acquisitive strategies (in control plants) and conservative strategies (in drought/heat plants). With ontogeny, the higher sensitivity of central traits (highly connected with other plant traits in a network) in T‐PTNs to stress resulted in more significant separation among treatment groups along the axis. Second, based on life cycles, stress decoupled trait correlations in L‐PTNs and decreased the value of central traits (related to carbon assimilation/accumulation) compared with the ambient control, resulting in lower fitness. These results suggested that central traits in PTNs drove the shifts in ecological strategies across changing environments, and the network topology (the connection frequency or clustering degree among traits in a network, describing the correlation patterns) constrained the resource utilization efficiency in plants; both affect plant fitness collectively. These insights will facilitate more accurate and broader applications (e.g., the assessment of plant fitness or sensitivity to stress) of PTNs in trait‐based ecology.