Model-based collaboration is well-established to exchange design information in the Architecture, Engineering, and Construction (AEC) sector. Even though various tools provide comprehensive means to author such models containing discipline-specific elements and semantic information, there is great potential to improve how multiple discipline models are coordinated in an integrated environment. Today, project coordination is carried out through model federation by uploading (discipline) models to Common Data Environments (CDE), which implement the concept of BIM Level 2 according to ISO 19650. This standard, however, handles discipline BIM models primarily as monolithic files and does not foresee any connection among the individual objects the models contain across different disciplines. Subsequently, detecting and negotiating potential impacts on foreign models caused by modifications in a specific discipline model remains a manual task that must be carried out every time a new model version is populated. Incremental version control of BIM models has recently gained increasing research interest. Such approaches aim to exchange only the modified parts of a model. Due to the missing references across discipline models, however, the potential of such methods leaves further potential unexploited. However, enhanced collaboration can be achieved if the actual model increment is exchanged, and its content is used to propose modifications to the other discipline models. This paper presents an approach to establishing additional references across distinct discipline models in the coordination environment to realize this concept. The proposed method helps assess the impact of model modifications across disciplines and allows for implementing automated reactions. The basis of the proposed concept is formed by an Inter-discipline Conjunction Graph, which provides graph-based representations of the different discipline models and several types of conjunction relationships that connect information across the different discipline models. By establishing these interdisciplinary references, a modification potentially affecting other disciplines can be assessed explicitly. At the same time, the accepted best-practice of working with distinct federated discipline models is still respected but significantly improved by a more comprehensive management of the consequences caused by model changes.