Collaborative systems combine their capabilities to provide services going beyond their individual capabilities. They are designed for collaboration, and form groups consisting of various concrete systems that interact with and depend on each other during operation. These system instances can be of different system types, possibly developed and provided by different organizations. Functional dependencies between collaborative systems should be defined and analyzed early during development so that the desired interplay can be assured in later development stages. While the functional dependencies are typically specified on the type level for each system type separately, it is important to explore and understand the resulting functional interplay between concrete system instances within possible collaborations, i.e., to investigate candidate high-level architectures. State-of-the-art approaches lack support for analyses that can be applied in early stages where the desired functional interplay between collaborative systems is not yet deeply understood and potential architectures and configurations are explored. This paper contributes a model-based approach for the automated analysis of functional interplay within potential collaborations on the instance level. It builds on feature models as high-level abstractions of system capabilities and their dependencies to other collaborating systems. Our approach supports understanding functional dependencies that might affect the overall collaboration behavior during runtime. We demonstrate the feasibility of our approach by presenting a tool implementation, and illustrate its application to case studies of vehicle platooning and smart factories.