Extending dynamic vegetation models to simulate range shifts

Sammanfattning: In response to rapid climate change and increasing human pressure, a wide variety of taxa have shifted their distribution in the past decades (range shift), with important consequences for ecosystem services and human health and economy. However, it is not yet clear whether most species will be able to track their favourable habitats or lag behind the climate signal (migration lag). Studying the paleo-vegetation response to past climatic fluctuations may help to understand the ecological processes underlying range shift dynamics. This thesis aims to implement an efficient model to hind- and fore-cast the range shift of the vegetation at large temporal and spatial scales. To this end, we used the dynamic global vegetation model (DGVM) LPJ-GM, which couples a migration module to the widely-used DGVM LPJ-GUESS, thus allowing plant species to migrate while interacting with each other. First, we assessed and calibrated the migration module of LPJ-GM 1.0 by using estimates of migration rates derived from pollen records of major European tree species, resulting in the parametrized model LPJ-GM 1.1. In agreement with previous modelling studies and ecological theory (Reid’s paradox), long-distance dispersal events were found to be crucial in realizing the high spreading rates of migrating trees during the last deglaciation. Next, we upgraded the existing migration routine of LPJ-GM 1.1 (henceforth the SEEDISP option) and implemented a second migration routine (henceforth the FIXSPEED option) to decrease the computational cost of range shift simulations at larger scales. These implementations resulted in the model LPJ-GM 2.0. We then applied LPJ-GM 2.0 to simulate the paleo-vegetation dynamics during the last (inter-)stadial cycles (Europe after the Last Glacial Maximum and the Northern Hemisphere, NH, for the last 50 ky) under two dispersal modes, where plant establishment was determined by: (1) the standard LPJ-GUESS routine (free dispersal), or (2) additionally constrained by “seed” availability (dispersal limitation of LPJ-GM 2.0; SEEDISP for Europe and FIXSPEED for the NH). Applying migration constraints to vegetation dynamics altered the paleo-vegetation distribution at points of rapid climate change. A number of tree taxa and forested biomes experienced multi-millennia migration lags during the sudden warming events following the cold spells of the Older Dryas (around 14.5 ka) and the Younger Dryas (around 11.5 ka). We found the highest migration lags (>3 millennia) in the boreal forests colonizing Central Siberia across the Holocene, and in the post-glacial expansion of European temperate forests. The magnitude of migration lags of forested biomes and tree taxa depended on the presence of dispersal barriers, distance to glacial refugia, the configuration of the migrant population, thermal requirements for establishment and species-specific dispersal ability (and subsequent competition). The performance of both SEEDISP and FIXPSEED was higher than “free dispersal” simulations when compared with pollen reconstructions. Altogether, our results suggest that accounting for migration processes in vegetation models will increase our confidence in future projections of plant range shifts and thus, of ecosystem services and climate-vegetation feedback.