The effect of tree diversity on forest productivity and resilience has been the subject of numerous research programs in the past decade. Large research projects like the BEF-China experiment, and networks like TreeDivNet and EuMIXFOR are evidence for the large investments into deciphering diversity-productivity relationships (DPR) in mixed forests around the globe (Zhang et al., 2012). For example, EuMIXFOR established a network of hundreds of research plots with a triplet design of a mixed Fagus sylvatica -Pinus sylvestris stands compared to pure stands of the two species (Ruiz-Peinado et al., 2018). A global meta-analysis showed that forest productivity increases with species richness and trait variation (Zhang et al., 2012). Mixed forests are, on average, 24% more productive than monoculture forests, with large variability among studies. Indeed, cases where mixtures are less productive than monocultures also exist (Forrester, 2014). In the BEF-China tree diversity experiment in a subtropical forest, tree growth increased with neighborhood species richness, leading to a positive DPR at the community scale (Fichtner et al., 2018). In a tropical

Root exudates are part of the rhizodeposition process, which is the major source of soil organic carbon (C) released by plant roots. This flux of C is believed to have profound effects on C and nutrient cycling in ecosystems. The quantity of root exudates depends on the plant species, the period throughout the year, and external biotic and abiotic factors. Since root exudates of mature trees are difficult to collect in field conditions, very little is known about their flux, especially in water-limited ecosystems, such as the seasonally hot and dry Mediterranean maquis. Here, we collected exudates from DNA-identified roots in the forest from the gymnosperm Cupressus sempervirens and the evergreen angiosperm Pistacia lentiscus by 48-hour incubations on a monthly temporal resolution throughout the year. We examined relationships of the root exudate C flux to abiotic parameters of the soil (water content, water potential, temperature) and atmosphere (vapor pressure deficit, temperature). We also studied relationships toC fluxes through the leaves as indicators of tree C balance. Root exudation rates varied significantly along the year, increasing from 6 μg carbon cm−2 root day−1 in both species in the wet season, to 4-fold and 11-fold rates in Pistacia and Cupressus, respectively, in the dry season. A stepwise linear mixed-effects model showed that the three soil parameters were the most influential on exudation rates. Among biotic factors, there was a significant negative correlation of exudation rate with leaf assimilation in Cupressus; and a significant negative correlation with leaf respiration in Pistacia. Our observation of enhanced exudation flux during the dry season indicates that exudation dynamics in the field are less sensitive to the low tree carbon availability in the dry season. The two key Mediterranean forest species seem to respond to seasonal changes in the rhizosphere such as drying and warming, and therefore invest C in the rhizosphere under seasonal drought.

<p>It has been assumed that mixing of species with high physiological diversity reduces competition over water and light resources, compared to single-species forests. Although several mechanisms to explain this observation have been proposed, empiric evidence is lacking. Here we studied water-use dynamics at a monthly resolution for two years in five key tree species in a mature, mixed, evergreen, Mediterranean forest. Root distribution was measured with DNA barcoding and soil cores. Measurements at the tree-scale were up-scaled using an ecosystem model of coupled water, carbon and energy fluxes (Regional Hydro Ecologic Simulation System, RHESSys). Tree species showed contrasting water-use patterns, with year-round activity in angiosperms, and mostly wet season-activity in gymnosperms. Water-use patterns matched the rooting patterns, with the deep- and shallow-rooted Ceratonia and Cupressus, showing year-round and seasonal behaviors, respectively. RHESSys simulations captured well the species-specific behaviors in the mixed forest, and were further applied to simulate monocultures of each of the species, which proved less productive than the mixed forest. Our results provide evidence for niche partitioning of the soil water resource among co-habiting tree species. This partitioning is driven by spatiotemporal species differences in rooting depth and eco-physiology, and facilitates the higher productivity of the mixed forest.</p>