The datasets comprise nearly 19’000 trees of European beech (Fagus sylvatica L.) from unmanaged forests in Switzerland, Germany / Lower Saxony and Ukraine. Tree death was modelled as a function of size and growth, i.e., stem diameter (DBH) and relative basal area increment (relBAI). To explain the spatial and temporal variability in mortality patterns, we considered a large set of environmental and stand characteristics.

Inventory data

The strict forest reserves in Switzerland and Germany had been established in the period of 1961-1975 and 1971-1974, respectively. Every reserve included up to 10 permanent plots ranging from 0.09 to 1.8 ha in size, with slightly irregular re-measurement intervals. Permanent plots with pure or mixed beech stands were selected from the reserves of both networks. Reserves with considerable wind disturbance during the monitored intervals were excluded from the analysis. In addition to data from the Swiss and German reserves, data from a 10 ha plot in the primeval beech forest Uholka in Western Ukraine including three remeasurements were used. The inventory data provide diameter measurements at breast height (dbh) for revisited trees with a diameter of more than 4, 7 and 6 cm for Switzerland, Germany and Ukraine, respectively.

Mortality predictors

A set of three consecutive inventories was used to generate records for the calibration of mortality models based on trees that were alive in the first and second inventory and either dead or alive in the third inventory. As an explanatory variable, the annual relative basal area increment (relBAI) was calculated based on the first and the second dbh measurement as the compound annual growth rate of the trees basal area. Tree dbh in the second inventory was used in addition to relBAI to model tree status (alive or dead) of the third inventory.

To increase the generality of the mortality models, we selected environmental variables that are known to have a considerable influence on growth and mortality of beech. We emphasized the effects of water availability using a large set of drought characteristics that were calculated based on the local site water balance. We also related beech mortality to soil pH, temperature, precipitation and growing degree-days. Additionally, we considered stand characteristics that reflect the development stage, competition and structure of the forests.

Further information

For further information, refer to Hülsmann et al. (2016) Does one model fit all? patterns of beech mortality in natural forests of three European regions. Ecological Applications.