Most forest gap models have been calibrated so that trees perform well in the centre of their altitudinal or latitudinal distribution. Model curves for growth response to climatic factors, such as temperature, are therefore not always optimized to realistically simulate the effect of limiting resources (e.g. cold or drought extremes).
ForClim (Bugmann 1994) and its derived model TreeMig (Lischke et al. in prep.) use the traditional parabolic representation of growth response to degree-day sum. This curve has been improved for the warm boundary of species' ranges in more recent versions of ForClim (Bugmann & Solomon 2000), by replacing it with an asymptotic function. In this paper, the focus is on the other extreme, namely growth in alpine or boreal tree-line conditions, with a view to improving the growth response to degree-day sum at its lower limit in the TreeMig model. This is achieved by comparing measured tree-ring width with degree-day sum, calculated from daily minimum and maximum temperature measurements, at various locations and for different species in the Swiss Alps.