Understanding the present and past climate-fire-vegetation dynamics of southern South America (40 - 45°S)

Abstract

The forest-steppe boundary that runs north-to-south along the eastern Andes is particularly dynamic over millennial time scales. Yet the relative role of long-term climate change and fire is poorly understood. In this study, I analyze the potential in using a process-based model in predicting species distribution, and the role fire and climate played in shaping the vegetation and treeline dynamics of Northern Patagonia (lat. 40 - 45 ° S). Paleoecological data, e.g., pollen, has been extensively used to study the relationship between climate and vegetation but has a low spatial resolution to distinguished between climate-fire-vegetation dynamics. Process-based model thus offers a transparent and robust method of incorporating a varying degree of complexity to understand fire behavior and fire-vegetation dynamics. Recently, LPJ-GUESS was parameterized to simulate major regional plant functional type (PFTs) and tree species distributions in this region. The model is able to predict regional species distribution across spatial scales by coupling establishment, growth, and mortality processes. Predicting spatial and temporal scale species distribution cannot be achieved without having the right climate and soil data, the climate data used was downscaled from 50 km to 1 km resolution using Worldclim climate data ( ~ 1 km) as the reference data. LPJ-GUESS model produced regional species distribution with fair to very good agreement with observation. The optimization of bioclimatic parameters and drought tolerance that is related to root depth, adaptability of plant to seasonal drought, and movement of nutrients consistently improved the accuracy of regional prediction of the species range. The model predicted that the vegetation distribution of present-day is mainly determined by climate and CO 2 rather than fire., while forest productivity responds strongly to elevated CO 2. However, based on the employed statistical methods of Canonical Correspondence Analysis (CCA) and Random Forest machine learning, combined with simulation results using paleoclimate. Results show that an increase in winter temperature drives the postglacial species distribution while changes in precipitation control radial growth and seedling establishment in the upper and lower treeline. These findings emphasize the importance of combining paleoecological methods with modeling to disentangle coarse-scale climate drivers from local influences.