Back

Heat Waves Promote, but Cyclones Reverse, Coral-to-Seaweed Regime Shifts by Reshaping Reef Resilience

Schmitt, R. J.; Kopecky, K. L.; Brooks, A. J.; Holbrook, S. J.

2026-09-28 ecology
10.64898/2026.09.27.754837 bioRxiv
Show abstract

Disturbances alter ecosystem resilience by killing organisms and leaving behind material legacies. We tested how material legacies from marine heat waves and cyclones govern transitions between coral- and seaweed-rich states on tropical reefs. Heat waves kill coral tissue but leave their skeletons to erode gradually, whereas cyclones scour away emergent biota. Field experiments revealed that dead skeletons protect seaweeds from herbivory, catalyzing a transition to a seaweed-rich state that becomes self-sustaining after skeletons disappear. After four years, 73% of skeleton-added (heat wave) plots transitioned to seaweeds compared to 0% of scoured (cyclone) plots that lacked skeletons, while 87% of seaweed-rich plots scoured to simulate a cyclone stayed seaweed-free. Reef trajectories after natural disturbances were similar: four years after heat wave-induced coral mortality, 63% of time-series plots had transitioned to seaweeds with none recovering to coral; four years after a cyclone removed skeletons from a prior disturbance, none had transitioned to seaweeds and 60% had returned to coral. A disturbance not only can displace an ecosystem state, it can rapidly reshape the resilience landscape through the material legacies it creates or removes. Our findings also show that successive disturbances need not have cumulative negative effects: after a reef is driven into a seaweed-rich state, a cyclone can act as a rescue disturbance by removing established seaweeds and, when structural refugia are also removed, shrinking the seaweed stability domain. By reshaping resilience through contrasting material legacies, disturbance type and sequence can be as important as frequency and severity in determining ecosystem trajectories.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.