Disturbances can wipe out populations of marine species, and often recovery does not follow population growth models. This is challenging for species management and the communities and ecosystems they facilitate. We viewed a shellfish die off in New Zealand as an opportunity to test hypotheses about the causes of recovery lags (e.g., loss of facilitation feedbacks causing hysteresis vs. misalignment of biophysical conditions). We also explore how these population shifts ripple out across the community and ecosystem. We excluded adult intertidal cockles in 9m2 plots at 23 intertidal soft sediment sites across 4 estuaries in Aotearoa New Zealand that reflect differing levels of exposure, mud content, and natural cockle density. Over the next year we measured shifts in several cockle population parameters and the macrofauna community composition when adult cockles were absent indicating the potential for broken feedbacks. There were also biophysical drivers of these shifts indicating potentially multiple drivers of deviations from population growth models. We use these multiple lines of enquiry to delve into the possible causes for recovery lags in ecosystems that have experienced mass mortality of a key species. Our project addresses the globally outstanding challenge of translating resilience and disturbance-recovery theory into real-world ecology.