Radial stem growth dynamics at seasonal resolution are essential to understand how forests respond to climate change. We studied daily radial growth of 160 individuals of seven temperate tree species at 47 sites across Switzerland over eight years. Growth of all species peaked in the early part of the growth season and commenced shortly before the summer solstice, but with species-specific seasonal patterns. Day length set a window of opportunity for radial growth. Within this window, the probability of daily growth was constrained particularly by air and soil moisture, resulting in intermittent growth to occur only on 29 to 77 days (30 to 80 %) within the growth period. The number of days with growth largely determined annual growth, whereas the growth period length contributed less. We call for accounting these non-linear intra-annual and species-specific growth dynamics in tree and forest models to reduce uncertainties in predictions under climate change.

Data originate from the TreeNet network (treenet.info), measuring stem radius changes and meteorological data in 10-min resolution at 47 sites in Switzerland.

Dendrometer data: Stem radius changes were derived from point dendrometers (ZN11-T-IP and ZN11-T-WP, Natkon, Switzerland), mounted perpendicular to the stem in a slope-parallel orientation, 1.3 m above the ground. Trees equipped with dendrometers were mature, dominant trees. Details about the installation of sensors in the field, data acquisition and processing, as well as extraction of the growth component and associated uncertainties are described in Zweifel et al. (2021a). The growth component from the stem radius changes was extracted based on the “zero growth” concept. Hereby, the accumulated growth (GRO in µm) appears as stepped line, which increases in periods when the stem radius exceeds a former maximum. The Ecology Letters manauscript has a detailed methods section that describes data collection, processing and methodology.

Meteorological data: Meteorological data were obtained either from weather stations from nearby MeteoSwiss stations (www.meteoswiss.admin.ch), or were recorded at the site (www.treenet.info, www.lwf.ch, www.empa.ch/web/s503/nabel), and aggregated to daily values. Hourly precipitation data were derived from the CombiPrecip model by MeteoSwiss, in which rain-gauge measurements and radar estimates are combined and interpolated at a 1-km² grid. Air temperature (Temp, °C) and relative humidity (RelH, %) were measured at 2 m height within the forest stands, and global radiation (RAD, W m^-2) was measured above the canopy or at a nearby open site, all at a 10-min resolution. In case of missing data, gaps were filled with data derived from the nearest MeteoSwiss station, which were corrected according to established linear regressions for periods with available data of both data sets. Vapor pressure deficit (VPD, kPa) was calculated from Temp and RelH with the R package ‘plantecophys’. Soil water potential (SWP, MPS-2/MPS-6 sensors, Decagon Devices, Pullman, US) was measured at 10-20 cm soil depth at each site and corrected for soil temperature. For each site the day length (hours of daylight between sunrise and sunset) was obtained as a function of latitude with the R package ‘insol’.