•Site conditions exert larger influence over growth than competition.•Secondary growth in Pinus pinea is mainly controlled by water stress.•Effect of competition on growth is alleviated on extreme dry years.•Under future climate scenarios a significant decrease of production is expected. Climate, competition and site conditions are the main drivers controlling annual secondary growth in tree species. These factors do no act independently on tree growth, but by means of interactions, resulting in mediated interactive effects. For example, the stress gradient hypothesis postulates alleviated interspecific competition under limiting spatial (site) or temporal (climate) resources. According to this, models predicting annual growth and yield for a given forest should consider these issues in their formulation. In this study, we present a modelling approach based on using data from permanent plots and dendrochronological analysis in order to describe annual tree growth in pure, even-aged stands of Pinus pinea L. in the Spanish Northern Plateau, a highly limiting environment due to its Mediterranean continental climate. Our method is based on identifying the different sources of variability by means of a multilevel linear mixed model, and thereby identifying the potential covariates explaining observed variability at the different spatiotemporal scales. Our results indicate that site related factors such as site index or dominant height exert a greater influence on annual secondary growth than size-symmetric competition. In addition, we found that the controlling influence of water stress is greater than that of temperatures on tree growth. Furthermore, our results allow evidence to be identified for the stress gradient hypothesis in temporal intraspecific interactions, since trees exposed to a higher degree of competition tend to grow more than expected in dry periods. In contrast, the effect of competition on growth, on average, tends to be aggravated at very poor sites. Finally, our modelling approach allows us to conduct growth and yield simulations under different climate scenarios at different spatial scales, providing results which point to significant decreases in timber and cone production under the more severe scenarios, which can be alleviated through more intensive silviculture.