Supplementary material for article "Climate change will cause non-analogue vegetation states in Africa and commit vegetation to long-term change"

Published: 15-10-2020| Version 1 | DOI: 10.17632/yx8wj84bd2.1
Mirjam Pfeiffer,
Dushyant Kumar,
Carola Martens,
Simon Scheiter


This archive contains time series maps (videos) that show decadal maps of simulated biome distribution in Africa between 1970 and 2099, relationship between same decade pairing of transient and equilibrium scenarios, relationship between closest decade pairing of transient and equilibrium scenarios, and maps of single state variables. Simulations were conducted using the aDGVM (Scheiter & Higgins (2009), GCB, doi: 10.1111/j.1365-2486.2008.01838.x). It also contains the decadally averaged aDGVM model output data analyzed in the study, as well as the scripts used to conduct data analysis and to create the Figures shown in the manuscript and its supplementary material. Abstract: Vegetation responses to changes in environmental drivers can be subject to temporal lags. This implies that vegetation is committed to future changes once environmental drivers stabilize. Understanding the trajectories of such committed changes is important as they affect future carbon storage, vegetation structure and community composition and therefore need consideration in conservation management. In this study, we investigate whether transient vegetation states can be represented by a time-shifted trajectory of equilibrium vegetation states, or if they are vegetation states without analogue in conceivable equilibrium states. We use a dynamic vegetation model, the aDGVM, to assess deviations between simulated transient and equilibrium vegetation states in Africa between 1970 and 2099 for the RCP4.5 and 8.5 scenarios. Euclidean distance between simulated transient and equilibrium vegetation states based on selected state variables was used to determine lag times and similarity of vegetation states. We found that transient vegetation states over time increasingly deviated from equilibrium states in both RCP scenarios, but that deviation was more pronounced in RCP8.5 during the second half of the 21st century. Trajectories of transient vegetation change did not follow a "virtual trajectory'' of equilibrium states, but represented non-analogue composite states resulting from multiple lags with respect to vegetation processes and composition. Lag times between transient and most similar equilibrium vegetation states increased over time and were most pronounced in savanna and woodland areas, where disequilibrium in savanna tree cover frequently acted as main driver for dissimilarities. Fire additionally enhanced lag times and Euclidean distance between transient and equilibrium vegetation states due to its restraining effect on vegetation succession. Long lag times can be indicative of high rates of change in environmental drivers, of meta-stability and non-analogue vegetation states, and of augmented risk for future tipping points. For long-term planning, conservation managers should therefore strongly focus on areas where such long lag times and high residual Euclidean distance between most similar transient and equilibrium vegetation states have been simulated.


Steps to reproduce

Modeled with aDGVM and averaged on decadal basis for nine key model output variables