The landscape is a grid of habitat cells. A slice of it is designated "protected" (raised green blocks); the rest is degraded matrix (brown). The species-area relationship (SAR) predicts how many species a patch of area A can support, and splitting the same total protected area into more, smaller patches lowers the predicted total — because the power-law curve is concave (z < 1). This is the ecological logic behind the "SLOSS" (single large or several small) reserve-design debate.
SAR (per patch): S = c · A^z (c=5, z=0.25)
Fragmented total: S_total = n · c · (A_total / n)^z
→ S_total decreases as n (patch count) grows,
for the same A_total, since z < 1.
Independently, 8 animal species are simulated as agents wandering the grid. Inside protected or corridor cells they reproduce (birth chance per second, capped by carrying capacity); on degraded cells they suffer elevated mortality. A species goes locally extinct in the simulation when its population reaches zero.
- Protected coverage — % of the grid designated as reserve; more area supported per the SAR curve.
- Fragmentation — splits the same protected area into more disconnected patches, lowering SAR-predicted richness and isolating agent populations.
- Corridors — links patch centres with safe travel strips, raising the connectivity index and letting agents recolonize neighbouring patches, cutting extinction risk.
- Rewild / Reset — reshuffles patch placement and restores all 8 species to their starting populations.
Real reserve networks (e.g. Natura 2000, national park systems) use exactly this trade-off: a few large, well-connected reserves usually outperform many small, isolated ones for maintaining biodiversity.