Testing three climate datasets for dendroclimatological studies of oaks in the South Carpathians.

Three gridded datasets containing interpolated daily and monthly precipitation and temperature values over the past five decades were tested against four tree-ring chronologies of oak (Quercus robur and Q. petraea). The objective of this research was to investigate the climate-growth relationship and whether the Pearson's product-moment correlation coefficients differ significantly if mean monthly precipitation and temperature data from the different climate databases, CRU, E-OBS and ROCADA are used. To this end, we selected two representative oak ecosystems in the South Carpathians, Romania, and analysed earlywood, latewood and tree-ring widths. Climate time series trends for the South Carpathians coldest, warmest days and wettest days were assessed with datasets from E-OBS and ROCADA, which differed in the density of their meteorological station network and their interpolation methods. The observed climatic parameters showed changes towards wetter conditions after the mid-1980s. For 1961-2013, E-OBS underestimated the mean daily temperature and daily precipitation compared with ROCADA. The results showed that higher extreme temperatures from January-March affected earlywood growth. In the investigated study region, latewood formation seemed to be affected by water availability mainly in May. Periods of drought associated with higher temperatures have limiting effects on tree growth, but these events are captured in different ways by each climate database analysed. Similarly, the results showed the discrepancy among datasets for earlywood and climate relationships. The results emphasize the importance of proper selection of climate data for assessing climate-tree growth relationships. For future dendroclimatological and dendroecological studies of oak in Romania, we recommend the ROCADA database, while E-OBS is recommended if an up-to-date climate dataset is needed.