Environmental vs demographic variability in stochastic lattice predator-prey models

STATPHYS 25, Seoul, 25 July 2013

Ulrich Dobramysl & Uwe C. Täuber

Department of Physics, Virginia Tech Blacksburg, VA, USA

Mauro Mobilia, School of Mathematics, University of Leeds, U.K. Ivan T. Georgiev, JP Morgan, London, U.K.

Mark J. Washenberger, Nebula Inc., San Francisco Bay Area, CaliforniaQian He, Citigroup, New York City, New York

Royce K.P. Zia, Iowa State University and Virginia Tech

Lotka-Volterra predator-prey interaction

• predators: A → 0 death, rate μ

• prey: B → B+B birth, rate σ

• predation: A+B → A+A, rate λ

mean-field rate equations for assumed homogeneous species densities:

da(t) / dt = - µ a(t) + λ a(t) b(t)

db(t) / dt = σ b(t) – λ a(t) b(t)

coexistence: a* = σ / λ , b* = μ / λ

K(t) = λ (a + b) - σ ln a – μ ln b = const.

→ neutral cycles, population oscillations

A.J. Lotka (1920); V. Volterra (1926)

not robust against model variations

Predator-prey coexistence: focal fixed point

Resonantly amplified stochastic population oscillations near focus A.J. McKane,T.J. Newman, PRL (2005)stochastic spatial model features robust

“individual-based” stochastic lattice Monte Carlo simulations: M. Mobilia, I.T. Georgiev, U.C.T., J. Stat. Phys. 128 (2007) 447

finite system should reach absorbing state, but survival times huge for large N

Quantitative analysisM.J. Washenberger, M. Mobilia, U.C.T., J. Phys. Cond. Mat.19 (2007) 065139

Correlation functions: oscillation frequency is strongly renormalized by internal reaction noise and spatial correlations

employ Doi-Peliti field theory to compute fluctuation corrections: U.C.T., J. Phys. A 45, 405002 (2012)

Environmental variability: spatially varying rates U. Dobramysl, U.C.T., Phys. Rev. Lett. 101, 258102 (2008)

predation probabilities λ: quenched random variables (fixed in each run), drawn from Gaussian distribution, truncated to interval [0,1], mean 0.5; study as function of variance σ

• stationary predator and prey densities increase with σ• amplitude of initial oscillations larger• Fourier peak associated with oscillations broadens, faster relaxation• origin: more localized activity patches, causing enhanced fluctuations• spreading population front speed enhanced

λ

λ

Demographic variability: rates attached to individuals U. Dobramysl, U.C.T., Phys. Rev. Lett. 110, 048105 (2013) each individual is assigned a predation / evasion efficacy η / η : drawn from evolving distribution, initial values 0.5; effective predation rate: λ = (η + η ) / 2

A

A

B

B

leave σ = 0.5 = μ fixed

offspring efficacies η : drawn from Gaussian around parent’s rate, mutation variance w

stochastic dynamics over many generations:

→ evolutionary optimization

P

O

Quasi-species mean-field approach

↓

→ inherent selection bias for prey population

Well-mixed (zero-dimensional) Monte Carlo simulations• random sequential updates, average over 1000 realizations• initially sharp distribution peak at η = 0.5 for predators and prey • study dependence on variability / mutation variance wP

no fixation: nonlinear predation process, mutation variance remains fixedmean-field theory naturally over-estimates cooperative optimization effects

Lattice (two-dimensional) Monte Carlo simulations• 128 x 128 square lattice, periodic boundary conditions, • data averaged over 10000 realizations• initially sharp distribution peak at η = 0.5 for predators and prey • study dependence on variability / mutation variance wP

spatial correlations enhance deviations from mean-field model results

Environmental vs demographic variability U. Dobramysl, U.C.T., Phys. Rev. Lett. 110, 048105 (2013)

smooth tuning parameter 0 ≤ ς ≤ 1: λ = ς η + (1 – ς) (η + η ) / 2

stationary predator population: correlation length / relaxation time:

density enhancement optimalfor equal variances w = w , and minimal for ς ≈ 0.3

→ spatial variability effects dominate

extinction time in small 10 x 10 lattice:variability increases population robustness

P S

S A B

Summary and conclusions

• predator-prey models with spatial structure and stochastic noise: invalidates Lotka-Volterra mean-field neutral population cycles

• stochastic models yield long-lived erratic population oscillations; resonant amplification mechanism for density fluctuations

• spatial stochastic predator-prey systems: complex spatio-temporal structures; spreading activity fronts induce persistent correlations; stochastic spatial scenario robust with respect to model modifications

• fluctuations strongly renormalize oscillation properties; fluctuation corrections captured perturbatively through Doi-Peliti field theory

• spatial variability in the predation rate results in more localized activity patches; population fluctuations in rare favorable regions cause marked increase in the population densities / fitness of both predators and prey

• variable rates attached to individuals with inheritance and mutation: intriguing dynamical evolution of rate distributions, no fixation

• quasi-species mean-field model allows semi-quantitative description