1. The resilience literature often assumes that social–ecological reorganization will result in either the removal of deficient system elements (components, interactions) or social learning. Major perturbations are expected to lead to either adaptation or, if accompanied by a regime shift, transformation. This has led to a conflation of the concepts of resilience and adaptation, which has in turn made it difficult to quantitatively distinguish between cases in which a system returned to a previous state, and adaptation or learning occurred, and cases in which the system was resilient but adaptation or learning did not occur. 2. We used a network analysis of nine years of ostrich movement data to explore the social–ecological resilience of the Western Cape ostrich industry, which nearly collapsed following an outbreak of highly pathogenic avian influenza in 2011 and has gradually rebuilt. 3. The system that emerged following the outbreak contained fewer farms but was more connected than at any period prior to the outbreak. As system reorganization proceeded, network traits began to fluctuate seasonally and to approach values similar to those observed prior to the outbreak. It was estimated that it would take 4–5 full seasonal cycles for the system to return to a similar state to that prior to the disease outbreak. In other words, although the system reorganized following the system collapse, it remained within the same regime and showed no obvious evidence of adaptation or learning. 4. Policy implications. The majority of previous work on studying system response to disturbance has focused on outcome-based adaptation and learning. This study highlights the need to understand systems that respond to disturbance without learning or adaptation. Network analysis offers a useful quantitative tool for exploring social–ecological resilience and tracking changes in vulnerability. However, the development of better ways of incorporating additional data from multiple scales into network analysis remains an important priority for improving the predictive power and policy relevance of network approaches to analysing resilience.