A report on the analogy between past and future vegetation tipping is provided to Task 8.1.1 (Milestone MS7)

Exploring Tipping Points and Their Impacts Using Earth System Models Milestone MS7 A report on the analogy between past and future vegetation tipping is provided to Task 8.1.1 1 MS7 Report on the analogy between past and future vegetation tipping MILESTONE REPORT Due date of Milestone: 31 December 2025 Actual Submission Date: Submission to the European Commission: 17 December 2025 Milestone MS7 Report on the analogy between past and future vegetation tipping is provided to task 8.1.1 Milestone Lead Commissariat à l’énergie atomique et aux énergies alternatives (CEA) Author(s) Pascale Braconnot (CNRS-IPSL, CEA), pascale.br[email protected]sl.fr Charline Ragon (CNRS-IPSL), charline.r[email protected].fr Andy Wiltshire (METO), [email protected] Eddy Robertson (METO), [email protected] Reviewer Chiara Bearzotti (DMI) Disclaimer: Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Climate, Infrastructure and Environment Executive Agency (CINEA). Neither the European Union nor the granting authority can be held responsible for them. 2 MS7 Report on the analogy between past and future vegetation tipping TABLE OF CONTENTS 1. Means of verification 4 2. About the work done to achieve this milestone 4 2.1 Sahel-Sahara region 4 2.1.1 Greening of the Sahel from a future climate perspective 4 2.1.2 Analogy between Holocene and projected vegetation changes 6 2.1.3 Analogy between AMOC-induced Sahel vegetation tipping point and future tipping point 8 2.2 Boreal forests 9 3. Contribution to the TipESM objectives 11 4. References 12 3 MS7 Report on the analogy between past and future vegetation tipping 1. Means of verification As stated in the Description of the Action, the means to verify that this milestone has been achieved is: “The document available on Zenodo in open access”. 2. About the work done to achieve this milestone The TipESM project extensively investigates tipping in three terrestrial ecosystems: the Amazon rainforest, Sub-Saharan vegetation, and boreal forests in Europe. The objective of this milestone is to investigate analogies between past vegetation tipping points and possible vegetation tipping points in the future. There is thus a need to understand future tipping points as they emerge in climate projections, in particular in the core TIPMIP simulations, therefore in TipESM WP1, and past conditions. Most partners, such as the METO in Section 2.1.1 investigate the differences between pre-industrial and ramp-up/down experiments, whereas CNRS-IPSL considers simulations of the last 6,000 years. This milestone is based on preliminary results obtained with the UKESM1 and IPSLCM6 models, focusing primarily on Sahara-Sahel vegetation and boreal forests. 2.1 Sahel-Sahara region 2.1.1 Greening of the Sahel from a future climate perspective In the future, the Sahara and Sahelian regions could experience more rainfall than today due to climate change. Wetter periods have occurred in the past and are associated with a moist vegetation landscape in contrast to today’s semiarid environment. The period of the Green Sahara in the Early Holocene and its rapid decline between 5 and 3.5 kyrs are illustrative of this (Pausata et al. 2020). These past significant changes raise the question of what may happen under climate change, possibly leading to a tipping point to a ‘Green Sahelian’ region. Figure 1 shows the simulated vegetation cover over the Sahel in the UKESM1.2 TipESM simulations. Vegetation cover generally increases with warming and is associated with an increase in precipitation. However, there is significant variability, particularly in the 4k zero-emission run, which is possibly indicative of threshold behaviour about a changing state. 4 MS7 Report on the analogy between past and future vegetation tipping Figure 1: Results from the UKESM1.2 simulations following the TipESM model protol. 1) shows the extent of vegetation cover over the Sahel region, and b) the annual precipitation. Understanding of past changes in the Sahelian region points to a role of a strong-land atmosphere coupling, where greening the Sahel can make rainfall more sustaining and therefore flip the state of the system. This has led to ideas about GeoEngineering; however, it remains highly uncertain, and the processes are not fully understood. To further investigate, we have used UKESM1.1 AMIP simulations to test the climate impact of the African Great Green Wall. To represent the greening, we increase the Leaf Area Index (LAI) (by 1 m2/m2) in a band south of the Sahara, where precipitation is between 100 and 400 mm/yr. There is a strong latitudinal gradient in soil albedo, so that increases in LAI reduce albedo in the north and increase it in the south, and differences in exactly where the greening occurs will have different impacts on the surface energy budget. In our experiment, we generally observe an increase in absorbed shortwave, leading to small increases in latent heat flux and larger increases in sensible heat flux. There is little coherent impact on temperature or precipitation. The Great Green Wall targets a relatively narrow band of land, including some areas with low soil albedo, and more extensive greening of the Sahara further north would produce a more coherent surface energy budget response that could drive atmospheric feedbacks. Figure 2: Difference between PFT and soil albedo. 5 MS7 Report on the analogy between past and future vegetation tipping Figure 3: Change in surface energy fluxes and near-surface air temperature due to increased LAI in the Great Green Wall region. 2.1.2 Analogy between Holocene and projected vegetation changes The team at CNRS-IPSL is investigating whether analogies of tipping points found in simulations of the last 6000 years (mid-Holocene to pre-industrial) share some analogies with emerging tipping in future projections. For this, we performed a transient simulation from the mid-Holocene to 2100 with the IPSL model including dynamical vegetation. The model version corresponds to version 4 described in Braconnot et al. (2025). The simulation only accounts for natural vegetation, computed online. Land use was not included. Aerosols are prescribed to be those of 1850, as in the CMIP6 preindustrial simulation (Eyring et al. 2016). The external forcing is thus due to the slow changes of the Earth’s orbit and changes in trace gases. 6 MS7 Report on the analogy between past and future vegetation tipping The simulation has been extended to the historical period and to 2100, using the SSP4.5 scenario. All the simulations are concentration-driven. The emission-driven version of the IPSL model used in WP1 for the TIPMIP simulations is not yet compatible with the dynamical vegetation. A key feature of the last 6,000 years is the rapid desertification that happened during the Holocene. We characterise aridification by considering the evolution of the desert area, defined as the grid cells in which the bare-soil fraction exceeds that of all other grouped Plant Functional Types (PFTs) (tropical trees, temperate trees, boreal trees, and grasslands). Figure 4 shows that the simulated aridification is gradual, as expected from the evolution of the insolation, but punctuated by abrupt events. By analysing the spatial pattern (Fig. 5), we observe that the desert first extends southward in the eastern part, before the western part is affected, as already observed in the literature (Watrin et al. 2009, Lézine et al. 2011a, Dallmeyer et al. 2020, Hopcroft & Valdes 2021). The expansion is associated with a reduction of both the number of PFTs per cell and of the LAI. The 150-year projection from pre-industrial shows a reduction in the bare-soil fraction in the south-east, while the number of PFTs per cell and LAI increase. It indicates that the greening of the Sahara is a response to increased atmospheric CO2. Also, the greening is fast compared to the time scales identified over the last 6,000 years. Interestingly, the greening of the Sahel/Sahara in the projection follows the reverse pattern compared to the Holocene desertification. A preliminary result of this study is that the vegetation in the Eastern part of the Sahel would provide better constraints on the possibility of tipping points than the rest of the region. A preliminary evaluation of the vegetation between East and West seems to be in rather good agreement with what has been inferred from pollen and lake reconstructions (Lezine et al. 2011b), even though these reconstructions do not have temporal resolutions allowing for the detection of a tipping point. Figure 4: Evolution of the desert extent in North Africa, as estimated from the bare soil fraction evolution, in a transient simulation with the IPSL model. The slow tendency is computed using wavelets. 7 MS7 Report on the analogy between past and future vegetation tipping Figure 5: Bare soil fraction over north Africa (row 1), number of PFT per cell (row 2) and LAI (row 3) in mid-Holocene (-6 ka, column 1), in pre-industrial (1850, column 2, plotted as the difference with column 1), and in projection (2100, column 3, plotted as the difference with column 2). On row 2, dotted areas correspond to desert areas following our criteria, and hatched areas to places where the bare soil fraction equals zero. All the quantities shown correspond to means over 31-year periods. 2.1.3 Analogy between AMOC-induced Sahel vegetation tipping point and future tipping point We forced a tipping of the AMOC in a snapshot simulation with the IPSL model (Fig. 6) to evaluate its impact on vegetation systems in the Sahara/Sahel region (Fig. 7). As a response, the vegetation system becomes even more desertic in the eastern part compared to the effect of the insolation, confirming the higher vulnerability and more rapid changes of the vegetation in this region, in the IPSL simulation. 8 MS7 Report on the analogy between past and future vegetation tipping Figure 6: North Atlantic Deep Water formation [Sv = 106 m3/s] of transient simulations with (grey) and without (blue) dynamical vegetation, and of a snapshot simulation (green) with dynamical vegetation and forced AMOC tipping. The green simulation is used to investigate the effect of AMOC tipping on vegetation dynamics, and is initialised by combining atmospheric and oceanic conditions from the blue curve to the vegetation maps of the grey one. Figure 7: Bare soil fraction over north Africa (row 1), number of PFT per cell (row 2) and LAI (row 3) before (column 1) and after (column 2, plotted as the difference with column 1) a tipping of the AMOC. On row 2, dotted areas correspond to desert areas following our criteria, and hatched areas to places where the bare soil fraction equals zero. All the quantities shown correspond to means over 31-year periods. 2.2 Boreal forests Using the Holocene transient simulation described in section 2.2, we also focused on boreal forests in northern Europe. The extent of the boreal forest is defined as the grid cells where the cover fraction of 9