Model analysis of temperature impact on the Norway spruce provenance specific bud burst and associated risk of frost damage

The annual growth cycle of boreal trees is synchronized with seasonal changes in photoperiod and temperature. A warmer climate can lead to an earlier bud burst and increased risk of frost damage caused by temperature backlashes. In this study we analysed site- and provenance specific responses to interannual variation in temperature, using data from 18 Swedish and East-European provenances of Norway spruce (Picea abies), grown in three different sites in southern Sweden. The temperature sum requirements for bud burst, estimated from the provenance trials, were correlated with the provenance specific place of origin, in terms of latitudinal and longitudinal gradients. Frost damage had a significant effect on tree height development. Earlier timing of bud burst was linked to a higher risk of frost damage, with one of the sites being more prone to spring frost than the other two. The estimated provenance specific temperature sum requirements for bud burst were used to parametrize a temperature sum model of bud burst timing, which was then used together with the ensemble of gridded climate model data (RCP8.5) to assess the climate change impact on bud burst and associated risk of frost damage. In this respect, the simulated timing of bud burst and occurrence of frost events for the periods 2021–2050 and 2071–2100 were compared with 1989–2018. In response to a warmer climate, the total number of frost events in southern Sweden will decrease, while the number of frost events after bud burst will increase due to earlier bud burst timing. The provenance specific assessments of frost risk under climate change can be used for a selection of seed sources in Swedish forestry. In terms of selecting suitable provenances, knowledge on local climate conditions is of importance, as the gridded climate data may differ from local temperature conditions. A comparison with temperature logger data from ten different sites indicated that the gridded temperature data were a good proxy for the daily mean temperatures, but the gridded daily minimum temperatures tended to underestimate the local risk of frost events, in particular at the measurements 0.5 m above ground representing the height of newly established seedlings.