Understanding the response of forest net ecosystem productivity (NEP) to environmental drivers under climate change is highly relevant for predictions of annual forest carbon (C) flux budgets. Modeling annual forest NEP with soil-vegetation-atmosphere transfer models (SVATs), however, remains challenging due to unknown responses of 5 forests to weather of the previous year. In this study, we addressed the influence of previous year's weather on the inter-annual variability of NEP for a

more »

... spruce forest in Switzerland. Analysis of long-term (1997-2011) eddy covariance measurements showed that the Norway spruce forest Davos Seehornwald was a consistent sink for atmospheric CO 2 , sequestering 210 ± 88 g C m −2 per year on average. Previous year's 10 weather strongly affected inter-annual variability of NEP, increasing the explained variance in linear models to 53 % compared to 20 % without previous year's weather. Thus, our results highlight the need to consider previous year's weather in modeling annual C budgets of forests. Furthermore, soil temperature in the current year's spring played a major role controlling annual NEP, mainly by influencing gross primary productivity 15 early in the year, with spring NEP accounting for 56 % of annual NEP. Consequently, we expect an increase in net CO 2 uptake with future climate warming, as long as no other resources become limiting. relevant variable using soil-vegetation-atmosphere transfer models (SVATs) difficult 15588