Presets
Playback
Parameters
Layers
Model check
The science
Deer and wolves (Lotka-Volterra)
dx/dt = αx - βxy, dy/dt = δxy - γy. The equilibrium is (x*, y*) = (γ/δ, α/β). Near it the populations oscillate with period T = 2π/sqrt(αγ). The quantity V = δx - γ ln x + βy - α ln y is conserved, so the orbit is closed (the fourth graph shows the numerical drift of V). With a carrying capacity K the deer term becomes αx(1 - x/K), the orbit spirals into the equilibrium, and V is no longer conserved.
Grass, salmon, evolution
Grass follows logistic growth, closed form B(t) = K / (1 + (K/B0 - 1) e-rt) with K = 0.38 kg/m². The salmon run follows a Ricker map Sn+1 = a Sn e-bSn with b = ln(a)/40, so the fixed point is S* = 40; for a above about 7.4 the fixed point is unstable and the run cycles or turns chaotic. The deer-speed trait follows the breeder's equation R = h² S with truncation selection on slow animals.
Checks
(1) V drift over 200 time units. (2) Measured small-oscillation period against 2π/sqrt(αγ) and Ricker convergence to ln(a)/b. (3) The selection differential of the top 20 % against φ(z)/p, and Kleiber's law against the textbook value for 70 kg.
Simplifications
Two-species predator-prey with constant rates (no age classes, no seasons, no spatial movement); the real 3D valley adds grazing, bears, birds and weather. Real populations are discrete and noisy, so the closed orbits here are an idealisation. The model code is the same module as in the 3D simulation.
What to try
- Set a carrying capacity of 150 and watch the closed orbit become a spiral.
- Raise the Ricker parameter a past 7.4 and look at the salmon graph.
- Drag heritability up and see the third graph climb faster.