Context and aim: plant phenological traits often display clinal genetic variation along climatic gradient, suggesting that these traits mediate local adaptation. However, various clinal patterns are observed, e.g. co-gradient vs counter-gradient patterns, suggesting complex evolutionary processes involved. Multiple physiological processes can drive selection on plant phenological traits, with possible interactions among them. Moreover, the viability and fecundity components of selection are particularly difficult to distinguish in plants, where vegetative and reproductive phenologies are closely synchronized. Finally, interference with sexual selection may arise from assortative mating, that is the positive correlation between mates for flowering time. Our objective here is to disentangle the components of selection on spring phenology in a major tree species, the European beech. Material & methods: we used phenological survey to estimate the timing of budburst (TBB) in respectively 147 and 192 adult trees at low and high elevation along a steep altitudinal gradient spanning over less than 1km (Mont Ventoux, SE France). Male and female individual fecundities were estimated using respectively paternity analyses of 1414 seedlings and parentage analyses of 473 seedlings in a spatially explicit mating model approach. Fecundity and sexual selection were investigated by regressing fecundity against TBB and mating opportunities, estimated from phenological mismatch within the mating neighbourhood. Viability selection was estimated through the effect of TBB on frost damages, in a common garden experiment where 20 maternal progenies from each plot were grown. Results: First, assortative mating was significant only at the low-elevation plot, where spring phenology was more spread out. Secondly, phenological mismatch with neighbours reduced male but not female fecundities at both plots. Thirdly, female fecundities revealed directional selection for earlier TBB at both plots, with a stronger selection gradient at the cooler, upper plot. Finally, seedlings with later TBB showed less frost damages in the common garden, but only in families from the high-elevation plot. Conclusion: our results suggest that selection on spring phenology combines (1) stabilizing selection through the male reproduction function to maximize mating opportunities, (2) directional selection for earlier TBB through the female reproductive function to maximize the timing of fruit maturation (3) directional selection for later TBB to reduce damages due to late frost. Such intertwining of sexual, fecundity and viability selection calls for an integrative approach to predict the evolution of spring phenology under changing climate.