Puma trophic cascade dynamical simulation

What this research found

After ecologists reported community-wide change following a single mountain lion's return to a small San Francisco Bay Area preserve, K-Dense built a five-node dynamical food-web model to ask whether encouraging apex predators into small suburban preserves is a credible restoration strategy. The model projected 30 years under no pumas, occasional visits, and an established resident, then swept preserve area from 50 to 5,000 hectares. A resident puma produced a 223% gain in woody vegetation and a 320% release of gray foxes, but only in preserves above roughly 400 to 500 hectares — and the recovery was driven almost entirely by fear rather than killing.

• A resident puma flipped the degraded baseline decisively over 30 years: woody vegetation rose 223.0%, gray foxes 319.8%, mesopredator abundance fell 64.8%, and total browsing pressure fell 55.0%.
• Fear rather than predation drives the cascade. Paired counterfactual simulations attributed 95.1% of the resident-puma vegetation response and 96.6% of the gray-fox response to the non-consumptive behavioural pathway, leaving lethal predation about 5% and 3% respectively.
• Occasional visits are not a substitute for residency: vegetation gained only 49.1% and foxes 166.3%, mesopredator reduction was 21.4%, and mean deer abundance actually edged up from 40.12 to 44.79 as transient fear pulses let the browse layer and the deer grow between visits.
• Preserve size sets a hard threshold. Below roughly 400 to 500 hectares the cascade failed under any puma regime; a resident puma turned gray-fox release positive at about 430 hectares and mesopredator suppression at about 494 hectares, but needed about 1,044 hectares to restore foxes halfway and about 1,595 hectares to restore vegetation halfway within 30 years.
• Recovery speeds differ sharply by trophic level. Under residency gray foxes reached half of their ceiling in about 1.5 years and 90% in about 3.3 years, while vegetation reached half in about 2.7 years and never reached 90% inside the 30-year horizon.
• The two least-certain parameters move the tipping point in opposite directions: harsher fox edge mortality pushed the 50% fox threshold from about 790 to about 1,610 hectares, while a stronger deer fear response pulled the 50% vegetation threshold from over 3,500 hectares down to about 740 hectares.

How it was done

The preserve community was modelled as five coupled differential equations for puma, deer, a coyote-and-bobcat mesopredator guild, gray fox, and the woody-vegetation browse layer, using saturating Holling Type II feeding responses. Fear entered as multipliers that reduce deer browsing efficiency and mesopredator activity as puma pressure rises. Parameters were calibrated to California oak-woodland and chaparral so that the predator-free state reproduces the documented degraded signature, confirmed as a single global attractor from eight random starts. Because a locally prey-limited puma cannot self-establish, puma density was prescribed as an external forcing across three regimes, with the stochastic occasional-visit case run over 50 Monte Carlo realisations. Mechanism was decomposed by paired counterfactuals disabling fear or predation under identical schedules, and preserve area entered through edge-core geometry with a 200 m edge buffer, area-scaled carrying capacities, fox edge mortality and an area-dependent chance of residency.

Data sources

• Sonawane et al., Ecology and Evolution 16:e73775 (2026) — mammal community responses to increasing puma activity in a suburban preserve
• Crooks & Soulé, Nature 400:563 (1999) — mesopredator release in fragmented California systems
• Beschta & Ripple, Biological Conservation 141:1249 (2008) — cougar, mule deer and black oak trophic cascade in Yosemite
• Suraci et al., Nature Communications 7:10698 (2016) — experimental demonstration that fear of large carnivores causes a trophic cascade
• Benson et al., Ecological Applications 29:e01868 (2019) — extinction vortex dynamics of urban-isolated top predators