Claude Code subagent imported from lizTheDeveloper/ai_game_theory_simulation (.claude/agents/task_cynthia_extinction_debt_research.md). Copyright stays with the author.

Task: Research Extinction Debt Mechanics

Agent: Cynthia (super-alignment-researcher) Priority: HIGH (H-3 from roadmap) Date: November 27, 2025 Context: Biodiversity recovery overestimation fix

Your Mission, Cynthia

We need comprehensive research on extinction debt - the phenomenon where species loss continues for 50-400 years after habitat degradation is removed. Current simulation assumes immediate recovery when stressors are removed, which overestimates recovery potential.

Research Request

Key Papers to Retrieve

Priority 1 - Foundational:

1. Tilman et al. (1994) - "Habitat destruction and the extinction debt"

   • Foundational theoretical paper
   • Original concept formulation

2. Kuussaari et al. (2009) - European grasslands extinction debt

   • 50-200 year timescales documented
   • Meta-analysis across ecosystems

3. Halley et al. (2016) - "Dynamics of extinction debt across five taxonomic groups"

   • Meta-analysis across 36 studies
   • Lag effect quantification

Priority 2 - Recent Updates: 4. Papers from 2020-2025 updating extinction debt estimates 5. Climate velocity mismatch studies (species can't migrate fast enough) 6. Pollinator network collapse timescales

Parameters to Extract

For each ecosystem type (tropical forest, temperate forest, grassland, marine):

1. Lag timescales:

   • Time from habitat loss to species extinction
   • Range (min-max years)
   • Median lag time

2. Extinction debt ratio:

   • Committed extinctions / Total biodiversity loss
   • What percentage of loss is "in the pipeline"?

3. Mechanisms:

   • Population viability (Allee effects, genetic drift)
   • Trophic cascade delays
   • Mutualism collapse (pollinators, seed dispersers)
   • Climate velocity mismatch

4. Recovery dynamics:

   • Can debt be "paid down" with restoration?
   • Irreversibility thresholds (when is debt permanent?)
   • Half-life of recovery (if possible)

5. Temporal patterns:

   • Does debt accumulate linearly or exponentially?
   • Does paydown accelerate or slow over time?

Implementation Context

Current state:

• Biosphere boundary has asymptotic recovery (200 year half-life, 5% floor)
• But it treats all extinction as immediate
• Missing: tracking of committed vs realized extinctions

What we're building:

interface ExtinctionDebtState {
  queuedExtinctions: ExtinctionQueueEntry[];  // Species committed to extinction
  realizedExtinctions: number;                // Species actually extinct
  debtRatio: number;                          // Queued / (Queued + Realized)
  ecosystemCollapse: {
    [ecosystem: string]: {
      degradationYear: number;
      expectedDebt: number;
      realizedDebt: number;
    };
  };
}

Monthly timestep: Need parameters that work for 1-month simulation increments

Output Format

Save to: research/extinction_debt_YYYYMMDD.md

Required sections:

1. Executive Summary - Key findings in 3-5 bullet points
2. Lag Timescales Table - By ecosystem type
3. Mechanisms - Detailed description of each mechanism
4. Temporal Dynamics - How debt accumulates and resolves
5. Implementation Guidance - How to model this in monthly timesteps
6. Full Citations - DOI, year, authors, key quotes

Parameter format:

### Tropical Forest Extinction Lag

**Timescale:** 100-400 years (median: 200 years)
**Source:** Kuussaari et al. (2009), Table 2
**Mechanism:** Slow-growing trees create multi-century population inertia
**Uncertainty:** ±50% (high variability across studies)

Success Criteria

• ✅ 3+ peer-reviewed papers (2009-2025 preferred)
• ✅ Quantitative lag times for 4+ ecosystem types
• ✅ Mechanism descriptions with research backing
• ✅ Clear implementation guidance for monthly simulation
• ✅ Parameter uncertainty ranges documented
• ✅ Contradictory findings noted (if any)

What You'll Find

The plan (proposed_extinction_debt_modeling_20251125.md) already lists some key papers:

• Tilman et al. (1994) - foundational
• Kuussaari et al. (2009) - 50-200yr grasslands
• Dullinger et al. (2013) - 300-400yr Alpine plants
• Isbell et al. (2011) - trophic cascade delays

Your job: Fetch these papers, extract parameters, find any 2020-2025 updates.

Known Context

From biodiversity_extinction_rates_20251113.md:

• Current extinction rate: 100-1000× background (IPBES 2019)
• Brief mention of extinction debt (lines 218-223) but no parameters

Gap: We have the extinction RATE but not the extinction LAG. Need both.

Timeline

4-6 hours for comprehensive research + parameter extraction.

Next Steps After Research

1. Hand off to Sylvia (research-skeptic) for validation
2. After validation, hand off to Roy (simulation-maintainer) for implementation
3. Priya will run Monte Carlo validation

Cynthia, this is a well-documented phenomenon (200+ papers). Focus on finding the most recent quantitative estimates and making sure we have ecosystem-specific parameters. The simulation currently overestimates recovery - your research will fix this.

Status: READY TO START Blocking Issues: None Dependencies: None (standalone research task)