nucleation

LIT-726: Catastrophic regime shifts in ecosystems: linking theory to observation

2 min read

Properties5
StatusActive — read 2026-10-05 (NOTE-571): in full, from a UC Berkeley course page. Worth reading for its standard of evidence for alternative stable states. Field patterns (jumps in time series, multimodal distributions, dual relationships to a driver) are only hints. Experiments are what demonstrate: different initial states lead to different end states, a temporary disturbance causes a lasting shift, or the response to forward and backward change shows hysteresis. Positive feedback is not enough unless strong. Long return times can mimic an alternative state.
Tagsecology, complex-systems
ExtendsLIT-720 — Catastrophic shifts in ecosystems
Cited as “paper” byNOTE-571 — Catastrophic regime shifts in ecosystems: linking theory to observation
Cited as “source” byTHEORY-136 — A persistent or abrupt change is not evidence that a system has alternative stable states: field patterns such as jumps, multimodality and dual relationships also arise from a stepwise driver or a simple threshold, and only hysteresis, initial-state dependence or a lasting shift after a temporary disturbance, with slow return excluded, comes close to showing one

Graph View

Marten Scheffer & Stephen R. Carpenter (2003), Trends in Ecology & Evolution 18(12):648–656 — DOI-10.1016/j.tree.2003.09.002

Key takeaways

  • Theory (Box 2). In a stability landscape, a change in conditions can shrink a basin until it disappears at a bifurcation, and the system jumps to another basin. To return, conditions must be reversed past a second bifurcation: hysteresis. Basin size is resilience, and it can shrink while the state looks unchanged, so the system becomes “brittle”.
  • Caveats (§2). Stable states are really dynamic regimes. Shifts have multiple causes, and slow erosion of resilience is neglected in favour of the obvious trigger. Patterns depend on scale. The control condition must be external to the system or much slower than it.
  • Field hints, never conclusive (§3). Jumps in time series, multimodal distributions and dual relationships to a control factor can each also come from a stepwise driver or from a threshold that is not a bifurcation. Massive colonisation does not count unless the new state regenerates itself.
  • Experimental evidence (§4, Figure 4). Three kinds can come close to proof. Different initial states lead to different final states. A temporary disturbance leads to a lasting new state. Slow forward and backward change in a control factor shows hysteresis. Each has a trap. Recovery must not be blocked by isolation. Long return times mimic a new state: biomanipulated Lake Zwemlust stayed clear for six years, then slipped back. Hysteresis-like loops appear whenever the system is slow relative to the driver.
  • Models (§5). Alternative attractors require positive feedback, but “positive feedbacks only lead to alternative attractors if they are sufficiently strong”, so qualitative reasoning from feedback loops is not enough.
  • Coral reefs. The Caribbean shift to algae is told as nutrient loading masked by fish, then by Diadema, until a pathogen removed the urchins (§1).

Standing in the record

Filed on 2026-10-05 at the owner’s request, standing in for the unread Nature paper (LIT-720). It is the evidential standard for the essay’s “protective lock-in” (§5.4). A regime that persists is not thereby a self-maintaining alternative state. That needs hysteresis, or a lasting shift after a temporary disturbance, against the confound of slow return. Hughes et al. 2007 (LIT-711) is in the record as a reef case where the algal state vanished within 30 days of reopening the fish-exclusion cages. The account is THEORY-136.

Backlinks