David Fastovich, Stephen Meyers, Erin E Saupe, John Warren Williams , Maria Dornelas, Elizabeth M. Dowding, Seth Finnegan, Huai-Hsuan M. Huang, Lukas Jonkers, Wolfgang Kiessling, Ádám T Kocsis, Qijian Li, Lee Hsiang Liow, Lin Na, Amelia M Penny, Kate Pippenger, Johan Renaudie, Marina C Rillo, Jansen Smith, Manuel J. Steinbauer, Mauro Sugawara, Adam Tomašovỳch, Moriaki Yasuhara, Pincelli Hull

Climate and ecosystems are dynamic across timescales, but existing research into climate-driven vegetation dynamics usually focuses on singular timescales. We develop a novel spectral analysis-based approach that provides unprecedentedly detailed estimates of the timescales at which vegetation tracks climate change, from 10¹ to 10⁵ years. We report dynamic similarity of vegetation and climate at centennial frequencies (146^-1 to 17,505^-1 years^-1). A breakpoint in vegetation turnover (759^-1 years^-1) matches a breakpoint between stochastic and autocorrelated climate processes, suggesting ecological dynamics are governed by climate across these regimes. However, heightened vegetation turnover at millennial frequencies (4,650^-1 years^-1) highlights the risk of abrupt ecological responses to climate change, while vegetation-climate decoupling at frequencies shorter than 146^-1 years^-1 suggests that plant assemblages may continue changing for centuries in response to anthropogenic climate change, with potentially lasting consequences for ecosystem function and biodiversity.