Navigation mit Access Keys




The development, extension and application of TreeMig is a core project of the group dynamic macroecology.


Spatio-temporal vegetation dynamics, including plant species migration, plays an important role for range dynamics, carbon sequestration, climate-biosphere-feedbacks and biodiversity in the context of climate change. To assess future vegetation dynamics, large scale, spatio-temporal vegetation models simulating plant migration are indispensable.

TreeMig approach

Detailed model description

We developed the TreeMig model, a spatially explicit and linked forest landscape model originally based on a forest gap model (Lischke et al., 1998), which takes additionally into account tree species migration (Lischke et al., 2006). In each cell (sidelength from 25m to 1km) of a rectangular grid, forest dynamics is simulated at the species level, including environment dependent reproduction, growth, competition, and mortality, and between-cell seed dispersal which allows the simulation of migration (Lischke et al., 2006). Within-cell vertical and horizontal structure is depicted by frequency distributions of tree density in height classes and therefore light attenuation. TreeMig is furthermore parallellized by simulating stripes of the simulation area by different processors which need to communicate only by the seeds which are dispersed through their boundaries. This approach is very computing efficient and thus allows for large extent/high resolution simulations.


TreeMig: modelling spatio-temporal forest dynamics from stand to continental scale


A forest landscape model including spatial interactions allows to simulate forest dynamics and migration for various applications


Further model developments

  •  Coupling with land abandonment (Rickebusch et al., 2007)...
  •  ... and avalanche dynamics in high alpine regions (Zurbriggen et al., submitted)
  • Upscaling by
    - deriving migration speeds from TreeMig simulations on a transect under various climatic, competitive and fragmentation conditions (Meier et al., 2012)
    - simulating only representative grid cells for certain processes (Nabel and Lischke, 2013; Nabel, 2015) (MUSCATELLA)
  • Test of the effect of the formulation of climate input variability (MUSCATELLA
  • Coupling of TreeMig with the country scale hydrology model PREVAH, via phenology, leaf area and drought stress 

Planned developments:

  • Inclusion of trait variability
  • Simulation of seed dispersal by fast fourier transform
  • Fine-parametrization and validation
  • Metamodelling by fitting to a Lotka-Volterra model

TreeMig Applications

First regional scale (50*50 cells a 25m*25m) tests, initialized with seeds only in the center cell show first a wavelike spread of the pioneer species, and for species with low seed production strong effects of stochastic seed dispersal. Competition plays a major role in shaping the pattern (->animation).
Several simulation studies with TreeMig on the local to regional scale demonstrated that climate change and species migration can drastically affect composition and distribution of forests in past and in the future  (Lischke, 2004, 2005). 

For example, TreeMig simulations in the boreal/arctic zone of Siberia with climate change indicate that migration of tree species lags the northward shift of their potential niches by several centuries (Epstein et al., 2007; Goetz et al., 2011), which has consequences for feedbacks to the climate system by changed albedo and carbon sequestration.


Simulations for the Holocene in the Valais region show a vivid dynamics of species composition, determined by succession and migration, which are triggered by immigration events and climate (Lischke, 2004, 2005; Lischke et al., 2006).  In contrast, TreeMig simulations indicate that the afforestation (associated with a change in albedo and surface roughness) after a drastic temperature rice during the late-glacial is mostly driven by climate and positive feedbacks via nutrient dynamics and albedo for most species, initiating a successional pattern, which is later affected by the lagged immigration of Pinus (Lischke et al., 2013). 

TreeMig has also been applied to simulate climate change driven forest composition in the US (Minnesota) and to interprete the spread of Quercus ilex in Western France during the last century (Delzon et al., 2013). Current applications comprise ensemble simulations of climate change and disturbances effects in the alpine valley of Davos and in North-Eastern China, both in comparison with other forest landscape models.

We ran first Swiss-wide simulations with a 1km resolution, driven by a single realization of a climate change scenario (A1b), and not taking into account land use. The simulations showed a gradual decrease of biomass in the lower altitudes until 2050, followed by a drastic breakdown around 2075 and a slow part-recovery thereafter (Zappa, 2010; Schattan, 2013).

In an ongoing study contributing to the Swiss-wide climate change impact assessment CH2014, TreeMig Swiss-wide simulations with 200m resolution driven by one CC- and several scenarios for land-use change and migration point to strong climate and locally strong and overall moderate effects of land-use, competition and dispersal(-> TreeMig-CH). 

Code availability

The code of TreeMig can be requested from Heike Lischke.