The growing prevalence of self-adaptive systems (SASs) in cyber-physical environments, and their close interactions with human operators and end-users increase the need for understanding and explaining their system behavior. However, explainability requirements for SASs are scarcely discussed in the literature. To identify aspects that need to be considered when designing explanations for SASs, we collected insights from literature and analyzed explainability requirements of several SAS exemplars from different domains. We identified the necessity, content, and characteristics of explanations across various stakeholder groups by utilizing five guiding questions for a systematic analysis. The results show that many existing exemplars would most likely profit from integrating context- and stakeholder specific explanations in order to increase their understanding for developers and end-users.