The root system is essential but essentially invisible. Plants are anchored to the soil through their root system; their adaptation and survival abilities are highly dependent on their ability to learn to take advantage of the space that surrounds them. Certain trees are nevertheless some of the largest and oldest living species on the planet- evidence of their ability to adapt to changes in their environment. Within managed anthropogenic constraints, particular species are vulnerable. It is clear that a wide range of parameters are likely to influence the root system and its operation, which offers many entry points to improve our understanding of a root's capacity for expansion, its dynamics, its role within the plant itself and within the soil plant atmosphere continuum. Available data on the extent and dynamics of plant roots includes several thousand references. However, the vast majority of these observations were made within the first meter of the soil profile. Data acquisition for the fine and/or deep roots is currently limited by constraints of time and financial resources. To overcome this lack of information, while trying to assess root dynamics under different environmental conditions, many models have been developed. However, it is still difficult to describe the complexity of root development in the community and to integrate its "plasticity".To understand such a complex environment, we must work toward establishing a definition of objectives and the tools necessary to develop and implement them. The work developed in the first part of this thesis is the subject of two articles and focuses on a literature review about deep roots. The vital role of the root system for the plant is well highlighted, as is its impact as a link within the atmosphere. We discuss the role of deep roots in carbon storage, which is critical and often underestimated. The different techniques for accessing the root systems in situ are also considered, and we present our access technique for observing the roots down to depths of five meters. We have also developed tools for shooting through a flatbed scanner and image analysis ( IJ_Rizo ) now available online.In the second part of the thesis, we focus on the study of root systems of adult rubber trees (Hevea brasiliensis L. RRIM 600) in the northeast of Thailand. Thailand has greatly expanded its acreage to extend its operations beyond its natural climate zone, into areas of high water stress. In this context, we studied the seasonality and dynamics of fine roots for three years and their contribution to the carbon cycle. We were able to highlight the continuity of root dynamics during independent periods of defoliation, yet link these to the seasonality of rainfall. We were also able to characterize the low differentiation of root dynamics at 0 to 4.5 m of depth in this context.The third part of the thesis concerns the study of teak particular to the region of Luang Prabang, Laos. By applying the rain exclusion technique for 2 years on a plantation of twenty-year-old trees, we observed the influence of precipitation patterns. During the period of water stress, an almost total cessation of root growth, at the surface and significant depths, has been highlighted. We conclude that in the case of species with a high dependence on water resources there is a direct impact on the physiological state and stagnated trunk growth. Some individuals have demonstrated an ability to adapt by changing their foliation/defoliation cycle, accompanied by a resumption of trunk growth during the second year of imposed drought. We also demonstrated the importance of rooting as a means of carbon storage, which in this context represents more than 45% of total carbon- roughly double the amount published in other literature .The data obtained for the species studied in this paper can be used for modelling scenarios simulating climate change and changing land use.