Feedback loops mediate warming effects on complexity, stability and predictability of phytoplankton dynamics in global lakes

Understanding the intricate feedback mechanisms between phytoplankton and nutrients is fundamental to predicting aquatic ecosystem responses to global change. While unidirectional nutrient-phytoplankton interactions are well-studied, the reciprocal causality and symmetry of these feedbacks remain poorly explored. Using a global dataset of 134 lakes spanning four continents, we applied convergent cross mapping (CCM) to 593 monthly time series to identify pairwise feedbacks (PairFB) between chlorophyll a (chl a) and macronutrients (ammonium, nitrate/nitrite, phosphate, and silicate). We found that PairFB was highly prevalent (54-75%) and fundamentally symmetrical across global lakes. However, this symmetry is highly sensitive to environmental context. Accelerated warming rates systematically break symmetrical feedbacks, typically driving them toward phytoplankton-dominated control or nutrient-dominant feedback, whereas increasing latitude often promoted the formation of symmetrical feedback. Furthermore, localized factors such as lake water depth, trophic status, and nutrient co-limitation critically mediate the pathways of feedback formation and disruption. Our findings demonstrate that the structure and symmetry of essential phytoplankton-nutrient feedbacks are dynamic, environmentally contingent, and critical for understanding how lakes will respond to escalating anthropogenic pressures, providing a novel framework for diagnosing ecosystem resilience.