Groundwater is essential for human activities and is sometimes referred to as a non-renewable resource in the same way as oil and gas. During the past century, precious groundwater reserves worldwide have been threatened by anthropogenic release of chemical compounds. Among these, chlorinated ethenes (CEs) such as tetrachloroethene (PCE) and trichloroethene (TCE) belong to the most common class of groundwater contaminants. In oxic conditions, PCE and TCE are recalcitrant to any form of degradation and constitute a long-term source of groundwater contamination. It has been reported that CEs can be degraded biologically in anoxic conditions, serving as electron acceptors of an anaerobic respiratory process called organohalide respiration (OHR). In this process, PCE is sequentially reduced to TCE, cis-dichloroethene (cDCE), vinyl chloride (VC), and finally to harmless ethene (Eth). Engineering strategies based on reductive dechlorination, such as monitored natural attenuation (MNA), have been designed for the bioremediation of aquifers contaminated with CEs. However, stalling of the sequence of dechlorination has often been observed, resulting from incomplete or impeded biodegradation of the highly toxic daughter molecules (cDCE and VC) and leading to their accumulation in situ. It was shown in laboratory experiments that OHR of CEs is more efficient in mixed cultures and that organohalide-respiring bacteria (OHRB) live in association with other microorganisms in microbial consortia. In the aquifer ecosystem, OHRB rely on complex interactions between members of bacterial communities for their electron donors and carbon supplies. On the other hand, they are in competition for these resources with other terminal electron-accepting processes (TEAPs). These complex interactions together with the intrinsic structural heterogeneity of aquifers make it difficult to understand the reasons for lower CEs accumulation, to predict the fate of OHR, and to design bioremediation strategies. This thesis aimed at elucidating the reasons for lower CEs accumulation in situ. An ecological approach considering the aquifer ecosystem in its whole complexity was developed and applied to the specific cases of two contaminated sites showing accumulation of lower CEs. The proposed methodology relies on the precise depiction of both aquifer habitat and microbial communities interacting therein. However, the potential impact of the first technical step, namely the impact of the pumping parameters on groundwater samples used for the description of the bacterial communities, was not known at the start of this thesis work. Results of investigations addressing this topic showed that parameters related to the tubing characteristics did not impact the apparent bacterial community structures (BCS) in a laboratory experiment. However, the study revealed a significant impact of the pumping flow rate on apparent BCS extracted from the aquifer. Terminal-restriction fragment length polymorphism (T-RFLP), coupled with an appropriate groundwater sampling strategy, enabled an accurate profiling of the BCS. However no indication was provided concerning the identities of the community members. A bioinformatics tool called PyroTRF-ID was developed to overcome this obstacle. The software enabled affiliating terminal-restriction fragments (T-RFs) to precise phylotypes by coupling T-RFLP and pyrosequencing data. An additional function enables in silico assessment of restriction enzymes for designing appropriate T-RFLP procedures. The developed methodology was applied all along the thesis work. Optimized tools and procedures were used to investigate the reasons for the accumulation of VC, and to a lesser extent cDCE, in the first aquifer showing relatively homogeneous lithological composition and slow groundwater fluxes. Statistical findings indicated that VC reduction was outcompeted by Fe(III) reduction in some sections of the aquifer. T-RFs showing significant correlations with VC reduction variables were identified by sequencing and with PyroTRF-ID. Results showed sequences closely affiliated to uncultured bacteria of the "Lahn Cluster" (LC) within the class Dehalococcoidetes, previously reported as PCE-to-cDCE dechlorinating microorganisms, and to the genus "Dehalococcoides". According to present knowledge, only members of this genus are able to reduce cDCE and VC. A major T-RF negatively correlated with the LC T-RF affiliated to the genus Rhodoferax, containing iron-reducing bacteria (IRB). Furthermore, indications were obtained of a mutualistic interaction between IRB and sulfate-reducing bacteria, potentially playing an important role by reducing the Fe(III) contents locally. The importance of the aquifer hydrogeological structure and functioning was exemplified in the study of the second aquifer ecosystem, where discrepancies in the fate of OHR were observed. In the immediate vicinity of the source zone, complete OHR was occurring, with production of Eth, whereas cDCE was accumulating further downstream, in apparent similar redox conditions. The profiling and metagenomic techniques revealed drastically different bacterial communities in the zones displaying different OHR fate. The source zone displayed bacterial populations typically found in highly reduced environments and involved in OHR, such as Dehalococcoides sp. In contrast, the downstream zone showing an accumulation of cDCE displayed bacterial populations typically found in more oxidized environments, such as aerobic bacteria, nitrifiers, and denitrifiers. In this zone, the conditions were probably slightly oxidized periodically by the combination of aquifer recharge, a specific lithological configuration, and lower organic matter content. Finally, and based on these considerations, a methodology for the investigation of the reasons for lower CEs accumulation in contaminated aquifers is proposed and discussed. This methodology follows on a multidisciplinary ecological approach for the description of the functioning of the aquifer ecosystem, and enables formulating scenarii of the reasons for lower CEs accumulation.