Back

Bounded ecological memory in coral reefs: divergent demographic pathways under recurrent heatwaves and collapse under extreme thermal stress

Varasteh, T.; P. Curran, A.; Mathews, O.; L. Davidson, S.; Warfel, G.; Warsfold, F.; Shen, K.; Backman, V.; A. Marcelino, L.

2026-08-24 ecology
10.64898/2026.08.21.746008 bioRxiv
Show abstract

Anthropogenic warming exposes coral reefs to recurrent marine heatwaves (MHWs), yet responses to comparable thermal stress remain highly variable. A central challenge is determining whether reduced bleaching reflects acclimatization-like persistence of existing colonies or mortality-driven filtering that produces demographically depleted assemblages. Using colony-level surveys across the Florida Reef Tract (2014-2023), we show bleaching sensitivity declined from 2014 through 2022, consistent with ecological memory. A spatial matched-event framework integrating bleaching severity and adult colony density resolved these reduced-bleaching outcomes into two distinct pathways: demographic persistence (stable/increasing density) and mortality-filtered tolerance (declining density). Persistence pathways were maintained under predictable exposure regimes characterized by lower interannual thermal variability. This persistence was driven by weedy life-history taxa (e.g., Porites spp.) replacing historical framework builders, indicating stability reflects ecological reassembly rather than uniform increases in thermal tolerance. Under the unprecedented extremes of the 2023 MHW, this acclimatization-like buffering broke down, causing widespread sensitization and collapse of previously persistent networks. These results support a bounded ecological memory framework; prior exposure reduces bleaching sensitivity under predictable thermal regimes but fails under extreme heat stress. Consequently, modern refugia are best defined by adult standing stock retention representing systems undergoing selective ecological reassembly rather than full functional recovery.

Matching journals

The top 6 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.