Determining the proximity of ecosystems to tipping points is a critical yet complex task, heightened by the growing severity of climate change and local anthropogenic stressors on ecosystem integrity. Spatial Early Warning Signals (EWS) have been recognized for their potential in preemptively signaling regime shifts to degraded states, but their performance in natural systems remains uncertain. In this study, we investigated the performance of 'recovery length' - the spatial extent of recovery from a perturbation - and spatial EWS as early warnings of regime shifts in Posidonia oceanica meadows. Our experimental approach involved progressively thinning the P. oceanica canopy, from 0 to 100%, at the edge of a dead-matte area - a structure formed by dead P. oceanica rhizomes and colonized by algal turfs - to promote the propagation of algal turfs. We calculated recovery length as the distance from the dead-matte edge to the point where algal turfs colonized the canopy-thinned region. Our results showed a linear increase in recovery length with canopy thinning, successfully anticipated the degradation of P. oceanica. While spatial skewness decline with increased canopy degradation, other spatial EWS, such as Moran correlation at lag-1, low-frequency spatial spectra, and spatial variance, were ineffective in signaling this degradation. These findings underscore the potential of recovery length as a reliable early warning indicator of regime shifts in marine coastal ecosystems., Competing Interests: Declaration of competing interest The authours declare that they have no conflict that could influence the work reported in this paper., (Copyright © 2024. Published by Elsevier Ltd.)