The insurance hypothesis predicts that forests with tree species mixtures may resist better to stressful environmental conditions than forests composed of only one tree species. Most of the currently available literature tested this hypothesis for aboveground productivity and its related response variables, but less is known about belowground processes. In my PhD thesis, I studied the drivers of belowground productivity and decomposition across climatic gradients and how they are affected by tree mixtures. I hypothesized that mixing of tree species with contrasting rooting patterns and fine root morphologies, would result in a release of competitive pressure belowground, and translate into higher fine root standing biomass and increased fine root productivity. Moreover, I hypothesized that roots with contrasting chemical and morphological characteristics in mixed stands would decompose faster, which may be particularly important under nutrient-limited conditions. Under water-limiting conditions, such as during extreme summer drought, I hypothesized overall slower decomposition but an attenuating effect of tree mixtures on decomposition due to improved micro-environmental conditions, in particular for leaves, since roots decompose in a more buffered soil environment. To test these hypotheses I examined the variation in tree root functional traits (across- and within-species), and its consequences for fluxes of C, N and P at the ecosystem scale. I addressed three main objectives and associated research questions to quantify the interactive effect of tree mixtures and climate on: 1) vertical root segregation and fine root standing biomass, 2) fine root dynamics and their associated nutrient fluxes and 3) fine root- and leaf litter decomposition. I could benefit from two different field experiments for my work, one with a 10-year-old tree-plantation experiment with birch and pine close to Bordeaux (ORPHEE experiment), the second along a latitudinal gradient of mature beech forests in the French Alps (BIOPROFOR experiment).I observed that roots from the birch and pine tree-plantation showed similar vertical distribution and similar belowground root standing biomass in tree mixtures compared to monocultures, contrary to my first hypothesis. However, the greater allocation of pine but not of birch to root growth within the top soil horizons under less water-limiting conditions suggests locally favourable conditions that may lead to soil depth-specific asymmetric competition. In the same experiment, fine root production and decomposition were similar in mixtures and in monocultures, in contradiction with my second hypothesis. Moreover, I did not observe any interactive effects of tree mixtures with stand density or water availability. Interestingly though, birch roots, but not pine roots released P during root decomposition, which suggests an important role of birch in the P-cycle and for P nutrition of trees on these P-limited sandy soils. In line with my third hypothesis, I observed a slower decomposition of leaf litter and fine roots in response to reinforced and prolonged summer drought, irrespective of the position along the latitudinal gradient in the Alps. However, this slower decomposition under drought was not attenuated in forest stands with mixed tree species compared to single species stands. Compared to leaf litter, fine roots decomposed slower and released less C. Interestingly, I found a net N release in decomposing fine roots but not in decomposing leaf litter, which suggests a distinct role of fine roots in the N cycle. In conclusion, I found that mixing tree species did not attenuate negative effects of climate change. However, this thesis demonstrates that promoting mixtures can still be beneficial for at least one of the admixed tree species, through species addition (i.e., complementing one tree species with another tree species), as one tree species may facilitate another via belowground fluxes of N and P.