Seasonal, medium-term and daily patterns of tree diameter growth in response to climate

Tree growth is expected to be responsive to climatic drivers across a spectrum of temporal scales, ranging from yearly growth to daily water use and photosynthesis. Automatic dendrometers offer the potential to provide continuous high-resolution measurements of tree radius changes. The signal recorded contains three components: (1) a long-term seasonal growth component, (2) a mid-term component representing swelling after rainfall and subsequent drying and (3) daily cycles of water-uptake related to tree transpiration. For 91 trees at 4 sites (Picea abies: 58, Pinus cembra: 17, Fagus sylvatica: 14, Pinus sylvestris: 2) monitored in Austria between 2012 and 2015, we simultaneously modelled these three processes using a hierarchical nonlinear mixed-effects model represented by two logistic growth curves. The focus was on the mid-term and daily component, and therefore long-term growth that is typically modelled by including tree size, competition or site variables was represented by random effects only. Both mid-term and short-term components were species-specific. In general, P. cembra and F. sylvatica were less sensitive to climate variables than P. abies. For all species, the mid-term component was best represented using a 14-day moving average difference between rainfall and potential evapotranspiration, a 24-h moving average of precipitation and its 1–3 days lags, a 24-h moving average temperature and its 1–3 days lags. The daily cycles of water uptake were best related to hourly humidity and its 3-h lag, and interactions with the 14-day moving average difference between rainfall and potential evapotranspiration accounted for attenuating cycles after rainy events and increasing cycles in dry periods.