Elevational matches and mismatches of modelled and observed alpine treelines and upper forest limits globally

The highest elevation where trees are phyisologically able to establish and grow constitutes de potential alpine treeline. Treeline elevation has been modelled globally based on growing season length and temperature, leading to good fits between model predictions and training data (Paulsen & Körner 2014). This original model, based on WorldClim climate data, was re-implemented by Karger et al. (2019) based on CHELSA climate data, with a good global fit to independent validation data. However, our geographically diferentiated evaluation of this model with a larger set of validation points (1245 points globally) showed very interesting geographical patterns in the model bias, which await an explanation. Most surprisingly, some treelines were predicted at much lower elevations than the observed forest limits, which should not be possible if the modelled low temperature limit for tree growth is absolute. This was the case in very wet regions and those with very low annual thermal seasonality, especially in Patagonia, the Canadian Coast Mountains and on the Altiplano.

After correcting for these biases, we mapped the vertical distance of current forest limits to the potential treeline, as an indication of disturbances, at treelines globally, at a 1-km² resolution. It is is known that forest cover often ends at some distance below the potential treeline, due to both natural and anthropogenice disturbances and limitations (e.g., landslides, fires, unsuitable substrates, lagged responses to climate change). In the first global analysis of this distance, we found that the vast majority of current upper forest limits are located below the climatic potential treeline elevation. The median is about 500 m at temperate latitudes and between 750 m and 1000 m in the tropics and subtropics. These deviations partly have natural causes, but the largest ones are probably caused by millenia of high-elevation human land-use. Mapping the distances between current forest cover and potential treeline can inform conservation and management decisions for high-levation ecosystems and provides a baseline for interpreting and predicting treeline-ecotone responses to global climate change.