Ecological networks of interacting species are higher dimensional complex systems involving nonlinear coupling. Variations in the coupling strength can weaken the resilience of the network and result in tipping or critical transitions. Such networks are continuously subject to environmental perturbations, climate change and other anthropogenic activities, which can produce uncertainty in the system parameters, such as the growth rate of the species, further impacting tipping. Despite some studies successfully quantifying uncertainty in such networks, efforts to reduce this uncertainty are limited. Here, we advance by quantifying and minimizing uncertainty in tipping thresholds across a class of important ecological networks-mutualistic networks. This is done by deploying Bayesian inference on random ordinary differential equations, leveraging a universal dimension reduction technique to transform the high-dimensional system into a one-dimensional framework. Through Bayesian inference, we demonstrate how uncertainty in the occurrence of tipping points can be narrowed, providing estimates of tipping bounds. Further, we find the giant component of interaction matrices by considering near-neighbour interactions in mutualistic networks of different dimensions, which accurately captures the uncertainty present in the original networks. Our work is one of the first to mitigate uncertainty in tipping points in networks with applicability to higher dimensional systems spanning different domains. This article is part of the theme issue 'Critical transitions and intelligent control in complex systems'.
Deb et al. (Thu,) studied this question.