Reversal from historical gains to future losses of tropical ecosystem resilience under climate changeShow others and affiliations
Number of Authors: 102026 (English)In: Journal of Cleaner Production, ISSN 0959-6526, E-ISSN 1879-1786, Vol. 583, article id 149811Article in journal (Refereed) Published
Abstract [en]
Understanding the dynamic evolution and driving mechanisms of tropical terrestrial ecosystem resilience represents a critical scientific priority for safeguarding global biodiversity and ecological security. This study systematically integrated remote sensing observations, TRENDY dynamic vegetation model ensembles, and Coupled Model Intercomparison Project Phase 6 (CMIP6) climate projections, using the lag-1 autocorrelation method to quantitatively assess the spatio-temporal characteristics of the resilience of tropical ecosystems in history (1982-2018) and future (2015-2100). Random forest and PCMCI causal analysis were used to explore the response mechanism of resilience to environmental changes and its time-lag effect. The results showed that: (1) Tropical ecosystem resilience exhibited a significant strengthening trend during the historical period (1982-2018; p < 0.01), with shrublands (representing unmanaged grass-shrub mosaics and natural grasslands) displaying the most rapid resilience gain among the four vegetation categories, followed by forests, croplands, and pastures. (2) Attribution analysis identifies climate change (44.0%) as the dominant driver over CO2 (34.9%) and LUCC (21.1%). Random-forest and SHAP analysis identified temperature, radiation, and precipitation as the dominant hydrothermal controls, accounting for 22.86%, 43.42%, and 17.04% of the tropical area, respectively, while root-zone and surface soil moisture exerted broadly positive effects, particularly in Africa. (3) PCMCI causal analysis demonstrates both precipitation and root zone moisture have a positive time-lag effect on resilience, while vapor pressure shows a negative lag effect. (4) Future projections (2015-2100) reveal that tropical ecosystem resilience shows a significant declining trend (p < 0.01) under SSP126, SSP245, and SSP585 scenarios—reversing the historical strengthening pattern and indicating increased vulnerability to critical transitions—while SSP370 diverges from this pattern with a slight resilience gain. Collectively, this study establishes a transferable framework for integrating observations, process-based models, climate projections, machine learning, and causal discovery to diagnose and project ecosystem resilience across the global tropics. The results provide a scientific basis for incorporating resilience-based indicators into global ecosystem monitoring, early-warning systems, climate adaptation, and biodiversity conservation planning.
Place, publisher, year, edition, pages
2026. Vol. 583, article id 149811
Keywords [en]
Autocorrelation, Climate variability, CMIP6, Resilience, TRENDY multi-model, Tropical region
National Category
Ecology Physical Geography
Identifiers
URN: urn:nbn:se:su:diva-261431DOI: 10.1016/j.jclepro.2026.149811Scopus ID: 2-s2.0-105054373984OAI: oai:DiVA.org:su-261431DiVA, id: diva2:2114888
2026-11-032026-11-032026-11-03Bibliographically approved