Temperature rising would slow down tropical forest dynamic in the Guiana Shield

Abstract

Increasing evidence shows that the functioning of the tropical forest biome is intimately related to the climate variability with some variables such as annual precipitation, temperature or seasonal water stress identified as key drivers of ecosystem dynamics. How tropical tree communities will respond to the future climate change is hard to predict primarily because several demographic processes act together to shape the forest ecosystem general behavior. To overcome this limitation, we used a joint individual-based model to simulate, over the next century, a tropical forest community experiencing the climate change expected in the Guiana Shield. The model is climate dependent: temperature, precipitation and water stress are used as predictors of the joint growth and mortality rates. We ran simulations for the next century using predictions of the IPCC 5AR, building three different climate scenarios (optimistic RCP2.6, intermediate, pessimistic RCP8.5) and a control (current climate). The basal area, above-ground fresh biomass, quadratic diameter, tree growth and mortality rates were then computed as summary statistics to characterize the resulting forest ecosystem. Whatever the scenario, all ecosystem process and structure variables exhibited decreasing values as compared to the control. A sensitivity analysis identified the temperature as the strongest climate driver of this behavior, highlighting a possible temperature-driven drop of 40% in average forest growth. This conclusion is alarming, as temperature rises have been consensually predicted by all climate scenarios of the IPCC 5AR. Our study highlights the potential slow-down danger that tropical forests will face in the Guiana Shield during the next century.

Figure 1

Evolution of the community-averaged growth and mortality rates for four climate scenarios and the two forest dynamic models. Growth rates (a and c) and mortality rates (b and d) for model 1 on the left (a and b) and model 2 on the right (c and d). Colored areas represent the 95% confidence interval. In model 1, we assumed that the vigor estimator is not impacted by climatic variables that impact the growth, whereas in model 2, we assumed that climatic variables that impact the community growth also impact the vigor and, consequently, the mortality. Scenario A is equivalent to the RCP2.6, B is an intermediary scenario, and C is equivalent to the RCP8.5. BASE is a control scenario that uses the current values of the climatic variables and assumes that they will remain stable over time (Table 2).

Figure 2

Results of the sensitivity analysis. Mean of the 50 Sobol indexes computed for each input and output variable. Inputs: QD: quadratic diameter, AGBF: above ground fresh biomass, growth: average growth rate, morta: mortality rate, BA: basal area. Outputs: A_under: Area over REW and <0.4, Pre: precipitation, TMP: mean temperature, and interactions. Results of model 1 are on the left and model 2 on the right. Almost all outputs are primarily impacted by the temperature changes. Only mortality is strongly impacted by the precipitation changes.