DYNAMICS IN 3-SPECIES PREDATOR-PREY MODELS WITH TIME DELAYS. Wei Feng

Similar documents
GLOBAL ATTRACTIVITY IN A CLASS OF NONMONOTONE REACTION-DIFFUSION EQUATIONS WITH TIME DELAY

DYNAMICS OF A PREDATOR-PREY INTERACTION IN CHEMOSTAT WITH VARIABLE YIELD

AN EXTENDED ROSENZWEIG-MACARTHUR MODEL OF A TRITROPHIC FOOD CHAIN. Nicole Rocco

Asymptotic behavior of the degenerate p Laplacian equation on bounded domains

EXISTENCE RESULTS FOR QUASILINEAR HEMIVARIATIONAL INEQUALITIES AT RESONANCE. Leszek Gasiński

On a non-autonomous stochastic Lotka-Volterra competitive system

Non-Autonomous Predator Prey Model. with Application

Simultaneous vs. non simultaneous blow-up

Journal of Inequalities in Pure and Applied Mathematics

SIMULTANEOUS AND NON-SIMULTANEOUS BLOW-UP AND UNIFORM BLOW-UP PROFILES FOR REACTION-DIFFUSION SYSTEM

EXISTENCE OF POSITIVE PERIODIC SOLUTIONS OF DISCRETE MODEL FOR THE INTERACTION OF DEMAND AND SUPPLY. S. H. Saker

6. Age structure. for a, t IR +, subject to the boundary condition. (6.3) p(0; t) = and to the initial condition

Equilibria with a nontrivial nodal set and the dynamics of parabolic equations on symmetric domains

NON-CONSTANT POSITIVE STEADY STATES FOR A STRONGLY COUPLED NONLINEAR REACTION-DIFFUSION SYSTEM ARISING IN POPULATION DYNAMICS

PERMANENCE IN LOGISTIC AND LOTKA-VOLTERRA SYSTEMS WITH DISPERSAL AND TIME DELAYS

NUMERICAL SOLUTIONS OF NONLINEAR ELLIPTIC PROBLEM USING COMBINED-BLOCK ITERATIVE METHODS

ESAIM: M2AN Modélisation Mathématique et Analyse Numérique M2AN, Vol. 37, N o 2, 2003, pp DOI: /m2an:

Applied Mathematics Letters. Stationary distribution, ergodicity and extinction of a stochastic generalized logistic system

Analysis of a predator prey model with modified Leslie Gower and Holling-type II schemes with time delay

Finite-time Blowup of Semilinear PDEs via the Feynman-Kac Representation. CENTRO DE INVESTIGACIÓN EN MATEMÁTICAS GUANAJUATO, MEXICO

Some asymptotic properties of solutions for Burgers equation in L p (R)

REACTION-DIFFUSION EQUATIONS FOR POPULATION DYNAMICS WITH FORCED SPEED II - CYLINDRICAL-TYPE DOMAINS. Henri Berestycki and Luca Rossi

Gas solid reaction with porosity change

Dynamics on a General Stage Structured n Parallel Food Chains

Permanency and Asymptotic Behavior of The Generalized Lotka-Volterra Food Chain System

Simultaneous vs. non simultaneous blow-up

1 2 predators competing for 1 prey

Breakdown of Pattern Formation in Activator-Inhibitor Systems and Unfolding of a Singular Equilibrium

Uniqueness of Positive Solutions for a Class of p-laplacian Systems with Multiple Parameters

LIST OF PUBLICATIONS Edited Books (Conference Proceedings):

PREPUBLICACIONES DEL DEPARTAMENTO DE MATEMÁTICA APLICADA UNIVERSIDAD COMPLUTENSE DE MADRID MA-UCM

Krein-Rutman Theorem and the Principal Eigenvalue

UPPER AND LOWER SOLUTIONS FOR A HOMOGENEOUS DIRICHLET PROBLEM WITH NONLINEAR DIFFUSION AND THE PRINCIPLE OF LINEARIZED STABILITY

EXISTENCE, UNIQUENESS AND QUENCHING OF THE SOLUTION FOR A NONLOCAL DEGENERATE SEMILINEAR PARABOLIC PROBLEM

LIFE SPAN OF BLOW-UP SOLUTIONS FOR HIGHER-ORDER SEMILINEAR PARABOLIC EQUATIONS

Analysis of a delayed predator-prey model with ratio-dependent functional response and quadratic harvesting. Peng Feng

Existence of Positive Solutions for Boundary Value Problems of Second-order Functional Differential Equations. Abstract.

NONLINEAR DIFFERENTIAL INEQUALITY. 1. Introduction. In this paper the following nonlinear differential inequality

SEMILINEAR ELLIPTIC EQUATIONS WITH DEPENDENCE ON THE GRADIENT

Nonlinear Analysis 71 (2009) Contents lists available at ScienceDirect. Nonlinear Analysis. journal homepage:

ASYMPTOTIC THEORY FOR WEAKLY NON-LINEAR WAVE EQUATIONS IN SEMI-INFINITE DOMAINS

LOTKA-VOLTERRA SYSTEMS WITH DELAY

STABILITY OF A POSITIVE POINT OF EQUILIBRIUM OF ONE NONLINEAR SYSTEM WITH AFTEREFFECT AND STOCHASTIC PERTURBATIONS

LOCAL AND GLOBAL STABILITY OF IMPULSIVE PEST MANAGEMENT MODEL WITH BIOLOGICAL HYBRID CONTROL

An analogue of Rionero s functional for reaction-diffusion equations and an application thereof

Regularity of the density for the stochastic heat equation

Oblique derivative problems for elliptic and parabolic equations, Lecture II

Some lecture notes for Math 6050E: PDEs, Fall 2016

On the long time behaviour of non-autonomous Lotka-Volterra models with diffusion via the sub-super trajectory method

Appearance of Anomalous Singularities in. a Semilinear Parabolic Equation. (Tokyo Institute of Technology) with Shota Sato (Tohoku University)

Partial Differential Equations

NON-EXTINCTION OF SOLUTIONS TO A FAST DIFFUSION SYSTEM WITH NONLOCAL SOURCES

Fitness Dependent Dispersal in Intraguild Predation Communities

BLOW-UP ON THE BOUNDARY: A SURVEY

Threshold behavior and non-quasiconvergent solutions with localized initial data for bistable reaction-diffusion equations

On asymptotically symmetric parabolic equations

Global Dynamics of Some Periodically Forced, Monotone Di erence Equations

Hopf bifurcations, and Some variations of diffusive logistic equation JUNPING SHIddd

On the uniqueness of heat flow of harmonic maps and hydrodynamic flow of nematic liquid crystals

Convergence and sharp thresholds for propagation in nonlinear diffusion problems

Global Stability Analysis on a Predator-Prey Model with Omnivores

ATTRACTORS FOR SEMILINEAR PARABOLIC PROBLEMS WITH DIRICHLET BOUNDARY CONDITIONS IN VARYING DOMAINS. Emerson A. M. de Abreu Alexandre N.

Non-Turing patterns in models with Turing-type instability

A NONLOCAL REACTION-DIFFUSION POPULATION MODEL WITH STAGE STRUCTURE

MEAN CURVATURE FLOW OF ENTIRE GRAPHS EVOLVING AWAY FROM THE HEAT FLOW

Asymptotic behavior of infinity harmonic functions near an isolated singularity

Interacting population models with nonlinear diffusions in ecology

NUMERICAL SOLUTIONS OF NONLINEAR PARABOLIC PROBLEMS USING COMBINED-BLOCK ITERATIVE METHODS

Global Qualitative Analysis for a Ratio-Dependent Predator Prey Model with Delay 1

Regularity of Weak Solution to Parabolic Fractional p-laplacian

Existence of Almost Periodic Solutions of Discrete Ricker Delay Models

Journal of the Vol. 36, pp , 2017 Nigerian Mathematical Society c Nigerian Mathematical Society

STABILITY ANALYSIS OF DELAY DIFFERENTIAL EQUATIONS WITH TWO DISCRETE DELAYS

Spreading-vanishing dichotomy in a diffusive epidemic model with Stefan condition

COMPETITION OF FAST AND SLOW MOVERS FOR RENEWABLE AND DIFFUSIVE RESOURCE

LYAPUNOV-RAZUMIKHIN METHOD FOR DIFFERENTIAL EQUATIONS WITH PIECEWISE CONSTANT ARGUMENT. Marat Akhmet. Duygu Aruğaslan

Workshop on Theoretical Ecology and Global Change March 2009

1.Introduction: 2. The Model. Key words: Prey, Predator, Seasonality, Stability, Bifurcations, Chaos.

Oscillatory Behavior of Third-order Difference Equations with Asynchronous Nonlinearities

GENERALIZED FRONTS FOR ONE-DIMENSIONAL REACTION-DIFFUSION EQUATIONS

Lecture No 1 Introduction to Diffusion equations The heat equat

On an initial-value problem for second order partial differential equations with self-reference

On the Dirichlet Problem for the Reaction-Diffusion. Equations in Non-smooth Domains

Numerical Solutions to Partial Differential Equations

Introduction to Nonlinear Control Lecture # 3 Time-Varying and Perturbed Systems

ANALYSIS AND APPLICATION OF DIFFUSION EQUATIONS INVOLVING A NEW FRACTIONAL DERIVATIVE WITHOUT SINGULAR KERNEL

Multiple positive solutions for a class of quasilinear elliptic boundary-value problems

INFINITE TIME HORIZON OPTIMAL CONTROL OF THE SEMILINEAR HEAT EQUATION

NONSTANDARD NUMERICAL METHODS FOR A CLASS OF PREDATOR-PREY MODELS WITH PREDATOR INTERFERENCE

On the Well-Posedness of the Cauchy Problem for a Neutral Differential Equation with Distributed Prehistory

DETERMINATION OF THE BLOW-UP RATE FOR THE SEMILINEAR WAVE EQUATION

Applied Math Qualifying Exam 11 October Instructions: Work 2 out of 3 problems in each of the 3 parts for a total of 6 problems.

Stability of solutions to abstract evolution equations with delay

Dynamics of Modified Leslie-Gower Predator-Prey Model with Predator Harvesting

NONTRIVIAL SOLUTIONS TO INTEGRAL AND DIFFERENTIAL EQUATIONS

Feedback-mediated oscillatory coexistence in the chemostat

ON THE GLOBAL EXISTENCE OF A CROSS-DIFFUSION SYSTEM. Yuan Lou. Wei-Ming Ni. Yaping Wu

A PDE MODEL OF INTRAGUILD PREDATION WITH CROSS-DIFFUSION. Robert Stephen Cantrell

The Maximum Principles and Symmetry results for Viscosity Solutions of Fully Nonlinear Equations

Adaptive methods for control problems with finite-dimensional control space

Transcription:

DISCRETE AND CONTINUOUS Website: www.aimsciences.org DYNAMICAL SYSTEMS SUPPLEMENT 7 pp. 36 37 DYNAMICS IN 3-SPECIES PREDATOR-PREY MODELS WITH TIME DELAYS Wei Feng Mathematics and Statistics Department University of North Carolina at Wilmington Wilmington, NC83-597, USA Abstract. We study a differential equation system with diffusion and time delays which models the dynamics of predator-prey interactions within three biological species. Our main focus is on the persistence (non-extinction) of u-species which is at the bottom of the nutrient hierarchy, and the permanence effect (long-term survival of all the predators and prey) in this model. When u-species persists in the absence of its predators, we generate a condition on the interaction rates to ensure that it does not go extinction under the predation of the v- and w-species. With certain additional conditions, we can further obtain the permanence effect (long-term survival of all three species) in the ecological system. Our proven results also explicitly present the effects of all the environmental data (growth rates and interaction rates) on the ultimate bounds of the three biological species. Numerical simulations of the model are also given to demonstrate the pattern of dynamics (extinction, persistence, and permanence)in the ecological model. 1. Introduction. A simple and realistic model for single species population dynamics (with the assumption that the growth rate decreases linearly according to the population size) is the famous Logistic Equation [9] X (t) = rx(t)[1 X(t)/K], X() = X ; (1) where K is the so-called carrying capacity (or the maximum sustainable population)for the biological species. We may further think of a single species grazing upon vegetation, which takes time τ to recover. In this case, the population size with time delay X(t τ) will directly affect the growth rate of the population at time t. The logistic delay differential equation X (t) = ax(t)[1 X(t τ)/k] () has also been the object of intensive analysis by numerous authors (see, for example, [9]). It is known [1](Wiener and Cooke, 1989)that time delay has a tendency to produce oscillations in equation (3) which is non-oscillatory when τ =. In recent years, researchers also paid attention to the boundedness of the habitat and the diffusion effect of the species within it. In 1997, Feng and Lu [] studied the following diffusive logistic equation with instantaneous and discrete delay effects u t = D[ u + u(a αu βu τ in Ω (, ), u = on Ω (, ), (3) Mathematics Subject Classification. Primary: 58F15, 58F17; Secondary: 53C35. Key words and phrases. Mathematical Ecology, Differential equation models, Predator-prey systems, Time delays, Long-term survival and permanence. 36

3-SPECIES MODELS WITH DELAYS 365 u(x, s) = u (x, s) in Ω [ τ, ], and obtained results on the asymptotic limits of the solution u (to be given in the Preliminary section). In this paper we extend the study of population dynamics from single species models to a model of three species predator-prey interaction [1, 3,, 6] with diffusion and delays. In an ecological system with three species: A, B and C where B and C both consume A. let u(x, t), v(x, t) and w(x, t) be the corresponding scaled density functions of A, B and C. We also would like to consider the diffusion effect and time delay effects in the interactions of the species, caused by the time period of food digesting and recovery of the preys. This leads to a differential equation system with time delays modeling the Lokta-Volterra type of interactions among the species [1]: u t = D 1 [ u + u(a 1 u l 11 u τ 11 k 1 v l 1 v τ 1 k 13 w l 13 w τ 13 v t = D [ v + v(a + k 1 u + l 1 u τ 1 v l v τ k 3 w l 3 w τ 3 w t = D 3 [ w + w(a 3 + k 31 u + l 31 u τ 31 k 3 v l 3 v τ 3 w l 33 w τ 33 in Ω (, ) () u = v = w = on Ω (, ) u(x, s) = u (x, s), v(s, x) = v (x, s), w(s, x) = w (x, s) in Ω [ τ, ], where Ω is a bounded region in R 3 with smooth boundary Ω. The diffusion term u is represented by the Laplacian Operator, and the diffusion coefficients D i and the growth rates a i (i = 1,, 3) are all positive constants. As in many real-life food webs, we assume that a 1 a > and a 1 a 3 >, namely, the species higher in the trophic levels have relatively smaller growth rates. The interspecific interaction rates k ij and l ij (1 i, j 3) are all nonnegative constants. Through appropriate scaling of the density functions, we can obtain that k 11 = k = k 33 = 1, for the convenience of our later discussions. The initial functions u, v, and w are assumed to be nonnegative smooth functions in [ τ, ]. In system () we also denote the delay terms u τij = u(x, t τ ij ), v τij = v(x, t τ ij ), and w τij = w(x, t τ ij ), with τ = max{τ ij, 1 i, j 3}. The main objective of this paper is to answer the following important questions in population dynamics concerning model (): Determine the existence and boundedness of the density functions (u, v, w) for all three of the species in [, ) Ω. Find the conditions for the persistence (long-term survival) of the prey species(u) at the bottom of the trophic hierarchy. Establish a permanence criterion (long-term survival of all three species) in the ecological system (). Find the the impact of the environmental data on the behavior of the biological species and their ultimate bounds.. Preliminaries. In the absence of the two predators, the prey species u satisfies the following single species model with time delays u t = D[ u + u(a αu βu τ in Ω (, ), u = on Ω (, ), (5) u(x, s) = u (x, s) in Ω [ τ, ],

366 WEI FENG where u τ (x, t) = u(x, t τ with τ >. The diffusion rate D, the scaled growth rate a, the interaction rates α and β are all positive constants, with α + β = 1. Denote by λ > the smallest eigenvalue of the eigenvalue problem: φ + λφ = in Ω, φ = on Ω. (6) Feng and Lu [] obtained the following result concerning the solutions of the Logisticdelay system () and its corresponding steady-state problem φ + φ(a φ) = in Ω, φ = on Ω. (7) Lemma 1. (i) If a λ, then the boundary value problem (7) has only the trivial solution and the solution u(x, t) of (5) converges to uniformly on Ω as t. (ii) If a > λ and α > β, then the boundary value problem (7) has a unique positive solution Φ(x). For every nonnegative and nontrivial u, the solution u(x, t) of (5) converges to Φ(x) uniformly on Ω as t. The question on existence and global boundedness of the non-negative density functions in the 3-species predator-prey model () can be dealt with by applying the theory of upper-lower solutions given by Pao in 199 [7]. Lemma. If there exist a pair of smooth functions (ũ, ṽ, w) and (û, ˆv, ŵ) such that (ũ, ṽ, w) (û, ˆv, ŵ) on [, ) Ω and they satisfy the following inequalities: ũ t D 1 [ ũ + ũ(a 1 ũ l 11 û τ11 k 1ˆv l 1ˆv τ1 k 13 ŵ l 13 ŵ τ13 ṽ t D [ ṽ + ṽ(a + k 1 ũ + l 1 ũ τ 1 ṽ l 1ˆv τ k 3 ŵ l 3 ŵ τ 3 w t D 3 [ w D 3 w(a 3 + k 31 ũ + l 31 ũ τ31 k 3ˆv + l 3ˆv τ3 w l 33 ŵ τ33 û t D 1 [ û D 1 û(a 1 û l 11 ũ τ 11 k 1 ṽ l 1 ṽ τ 1 k 13 w l 13 w τ 13 ˆv t D [ ˆv + ˆv(a + k 1 û + l 1 û τ 1 ˆv l 1 ũ τ 1 k 3 w l 3 w τ 3 ŵ t D 3 [ ŵ + ŵ(a 3 + k 31 û + l 31 û τ31 k 3 ṽ l 3 ṽ τ3 k 33 ŵ l 33 w τ33 (ũ, ṽ, w) (,, ) (û, ˆv, ŵ) on Ω (, ) (ũ, ṽ, w) (u, v, w ) (û, ˆv, ŵ) in Ω ( τ, ), then there exists a unique solution (u, v, w) for () on [, ) Ω with (ũ, ṽ, w) (u, v, w) (û, ˆv, ŵ). We let M 1 = max{ a 1, u }, M = max{ a + k 1 M 1, v }, M 3 = max{ a 3 + k 31 M 1, w }. It is not hard to verify that (M 1, M, M 3 ) and (,, ) satisfy all the requirements given in Lemma for coupled upper and lower solutions of (). Our result on the existence and global boundedness for the non-negative density functions is given as follows: Theorem 1. The system of differential equations with time delays () has a unique solution (u, v, w) on Ω [, ) with (,, ) (u, v, w) (M 1, M, M 3 ).

3-SPECIES MODELS WITH DELAYS 367 3. Our Main Results: Persistence and Permanence. As seen in Section, a previous result on diffusive logistic equation with time delays shows that in the absence of its predators v and w in (), the u-species at the bottom of the nutrient hierarchy will persist when a 1 > λ and l 11 < 1. Through a series of comparison analysis on the three differential equations in system (), we will show in this section that when a 1 > λ and δ = l 11 + (k 1 + l 1 )(1 + k 1 + l 1 ) + (k 13 + l 13 )(1 + k 31 + l 31 ) < 1, the u-species at the bottom of the trophic hierarchy will persist. This result involves the growth rate a 1 of u-species and all the interaction coefficients (k ij and l ij ) in (). Furthermore, we also obtain the ultimate lower bound of the density function u in term of the interaction rates by showing that lim inf u(, t) (1 δ)φ 1 ( ). Theorem. Persistence of the u-species. Let a 1 > λ and Φ 1 (x) be the unique positive solution of the boundary-value problem (7) when a = a 1. If δ = l 11 + (k 1 + l 1 )(1 + k 1 + l 1 ) + (k 13 + l 13 )(1 + k 31 + l 31 ) < 1, (8) then the density function u satisfies lim inf u(, t) (1 δ)φ 1( ) in C(Ω). (9) Proof: 1. Let (u, v, w) be the time-dependent solution of (). From the fact that u, v, and w are all non-negative functions, we know that u satisfies u t D 1 [ u + u(a 1 u in Ω (, ). Hence u is a lower solution for equation (5) in Section when β =. Lemma 1 indicates that the solution for equation (5) converges to Φ 1 (x) when t. Therefore u satisfies lim sup u(, t) Φ 1 ( ) in C(Ω).. Using the same argument as above and the ultimate upper bound Φ 1 for u, we know that for any arbitrarily small ɛ >, there exists a T ɛ >, v t D [ v + v(a + (k 1 + l 1 + ɛ)φ 1 v D [ v + v(a 1 + ɛ + (k 1 + l 1 )Φ 1 v in Ω (T ɛ, ). Noting that (1 + k 1 + l 1 )Φ 1 (x) is the solution of the boundary value problem V + V [a 1 + (k 1 + l 1 )Φ 1 V ] = in Ω, V = on Ω, we can conclude that the density function v satisfies: lim sup v(, t) (1 + k 1 + l 1 )Φ 1 ( ) in C(Ω). For the same reason, we can also obtain lim sup w(, t) (1 + k 31 + l 31 )Φ 1 ( ) in C(Ω). 3. Finally we apply the results in part 1 and part to the first equation in () for u. After substituting the ultimate upper bounds for u, v, w into the equation, we can obtain that for arbitrary ɛ >,there exists a T ɛ > such that u t D 1 [ u + u[a 1 u (δ ɛ)φ 1 ] in Ω (T ɛ, ).

368 WEI FENG Noting that (1 δ)φ 1 (x) is the unique positive solution of the boundary value problem U + U[a 1 U δφ 1 ] = in Ω, V = on Ω, we now reach the conclusion that lim inf u(, t) (1 δ)φ 1( ) in C(Ω). We now turn our attention to the long-term survival of the the two predators and the prey in (). In the following two theorems, we will discuss the permanence effect in the predator-prey model which ensures the survival of all three species. Theorem 3. Permanence in Model () with Equal Growth Rates. Let a 1 = a = a 3 > λ and δ < 1. If and γ = l + (k 1 + l 1 )(l + δ 1) + (k 3 + l 3 )(1 + k 31 + l 31 ) < 1, (1) θ = l 33 + (k 31 + l 31 )(l 33 + δ 1) + (k 3 + l 3 )(1 + k 1 + l 1 ) < 1, (11) then the predator-prey system () is permanent. In fact, the density functions satisfy lim inf [u(t, ), v(t, ), w(t, [(1 δ)φ 1( ), (1 γ)φ 1 ( ), (1 θ)φ 1 ( in C(Ω). Proof: 1. It is seen from Theorem that when a 1 > λ, lim sup[v(, t), w(, t [(1 + k 1 + l 1 )Φ 1 ( ), (1 + k 31 + l 31 )Φ 1 ( in C(Ω). Also, for δ < 1, lim inf u(t, ) (1 δ)φ 1( ).. Hence for each < ɛ < min{1 γ, 1 θ}, there exists a T ɛ > such that and v t D v D v[a ɛ v + (k 1 + l 1 )(1 δ)φ 1 l (1 + k 1 + l 1 )Φ 1 (k 3 + l 3 )(1 + k 31 + l 31 )Φ 1 ] = D v[a 1 ɛ v γφ 1 ] in Ω (T ɛ, ) w t D 3 w D 3 w[a 3 ɛ w + (k 31 + l 31 )(1 δ)φ 1 (k 3 + l 3 )(1 + k 1 + l 1 )Φ 1 l 33 (1 + k 31 + l 31 )Φ 1 ] = D 3 w[a 1 ɛ w θφ 1 ] in Ω (T ɛ, ). 3. By the fact that (1 γ)φ 1 (x) is the unique positive solution of the boundaryvalue problem we conclude from Lemma 1 that Similarly, V + V [a 1 V γφ 1 ] = in Ω, V = on Ω, lim inf v(, t) (1 γ)φ 1( ) in C(Ω). lim inf w(, t) (1 θ)φ 1( ) in C(Ω).

3-SPECIES MODELS WITH DELAYS 369 Theorem. Permanence in Model () with Unequal Growth Rates. Let a 1 > a > λ, a 1 > a 3 > λ, and δ < 1. If < a 1 γ < a λ and < a 1 θ < a 3 λ, (1) with γ and θ as defined in (1) and (11), then the predator-prey system () is permanent. Proof: 1. By Theorem, when δ < 1, the u-species will persist.. Condition (1) gives γ >. for arbitrarily small ɛ >, there exists a T ɛ > such that v t D v D v[a ɛ v γφ 1 ] [a ɛ γa 1 v] in Ω (T ɛ, ). 3. Since a a 1 γ > λ, Lemma 1 and the arbitrariness of ɛ imply that lim inf v(, t) v ( ) in C(Ω), where v is the positive solution of v = v [a γa 1 v ] in Ω, v = on Ω. Therefore the v-species will also persist.. Similarly, the w-species will persist when < a 1 θ < a 3 λ. Numerical Simulations of the Model. We now work on numerical simulations [8, 5]of the three species predator-prey model () on a one-dimensional spatial domain Ω = (, 1), where the principal eigenvalue λ = π. By discretizing the differential equation systems into finite-difference systems, we obtain numerical solutions through the monotone iterative scheme developed and employed in several earlier papers[,, 6].The diffusion coefficients in () are fixed at D 1 = 1.3, D =., D 3 = 1. and the initial functions are taken to be u (x) = 1.5 sin(πx), v (x) = 1.5 sin(πx), w (x) = 5. sin(πx). We also fix each time delay term as τ ij = 1.. Example 1 - Figure 1. Persistence of u-species. As seen in Theorem, when a 1 > λ and δ < 1, the prey species u will persist. We choose the growth rates for the three species as a 1 = 13., a = 1., a 3 = 1.. The set of interaction rates l 11 =.1, l = 1., l 33 = 1., k 1 =., l 1 =., k 13 =., l 13 =.1, k 1 =.1, l 1 =.1, k 3 =.8, l 3 =.9, k 31 =.1, l 31 =.1, k 3 =.5, l 3 =.5 makes δ =.9 < 1 which ensures the persistence of the u-species. Example - Figure. Extinction of u-species. Here we give a numerical example that u-species goes to extinction for δ > 1. We let the growth rates for the three species a 1 = a = a 3 = 11.5. The set of interaction rates l 11 =.3, l =.5, l 33 =.5, k 1 =., l 1 =.5, k 13 =..3, l 13 =., k 1 =., l 1 =.3, k 3 =.5, l 3 =.3, k 31 =.1, l 31 =., k 3 =., l 3 =. makes δ =.56 > 1. Example 3 - Figure 3. Permanence with equal growth rates. As stated in Theorem 3, for a 1 = a = a 3 > λ and δ, γ, θ < 1, all three species in model () will persist. We let the growth rates for the three species a 1 = a = a 3 = 11.. The set of interaction rates l 11 =.1, l =., l 33 =.3, k 1 =.1, l 1 =., k 13 =.1,

37 WEI FENG l 13 =.1, k 1 =., l 1 =.1, k 3 =., l 3 =.1, k 31 =.1, l 31 =., k 3 =., l 3 =. makes δ =.88, γ =.87, θ =, 87. Example - Figure. Permanence with unequal growth rates. In Theorem, we gave conditions to ensure permanence in model () with unequal growth rates: a 1 > a > λ, a 1 > a 3 > λ, δ < 1, < a 1 γ < a λ, and < a 1 θ < a 3 λ. We let the growth rates for the three species to be a 1 = 3., a =., a 3 = 19.. The set of interaction rates l 11 =.1, l =.1, l 33 =.15, k 1 =.1, l 1 =.1, k 13 =.1, l 13 =.1, k 1 =.1, l 1 =.1, k 3 =., l 3 =., k 31 =.1, l 31 =.1, k 3 =., l 3 =.1 makes δ =.58, γ =.516, θ =.56. Therefore the permanence conditions are satisfied. Fig. 1(a) U species Fig. 1(b) V species u axis v axis 1.5 1.5 1.5 1 1.5 1 Fig. 1(c) W species w axis 1.5 1 Figure 1. Persistence of u-species Fig. (a) U species Fig. (b) V species u axis 1 v axis 1 1.5 1 1.5 1 Fig. (c) W species w axis 1.5 1 Figure. Extinction of u-species

3-SPECIES MODELS WITH DELAYS 371 Fig. 3(a) U species Fig. 3(b) V species u axis 1.5 1.5 v axis 1.5 1.5 Fig. 3(c) W species w axis 1.5 Figure 3. Permanence with equal growth rates Fig. (a) U species Fig. (b) V species 1 8 6 u axis 5 v axis 1.5 1.5 Fig. (c) W species 3 w axis 1 1.5 Figure. Permanence with equal growth rates 5. Discussions and Conclusions. In Theorem of Section 3, we have shown that when u-species persists in the absence of the predation from v- and w-species (a 1 > λ, l 11 < 1)and the predation interactions are not too intensive (δ < 1), the u-species at the bottom of the trophic hierarchy will persist. This persistence condition depends on all interactions rates reflecting the predation of v-species and w-species on the u-species (k 1, l 1, k 1, l 1,k 13, l 13,k 31, l 31 ). Furthermore, one can also see in Theorem 3 and Theorem that the permanence effect in the predator-prey model () requires a balanced environment on all the growth rates and interaction rates. Theorem and Theorem 3 also gives the relation between the environmental data (growth rates and interaction rates) and the ultimate lower bounds for the density functions u, v and w. by showing that lim inf u(, t) (1 δ)φ 1 ( ) (in Theorem

37 WEI FENG ) and similarly lim inf v(, t) (1 γ)φ 1 ( ), lim inf w(, t) (1 θ)φ 1 ( ) (in Theorem 3). Those results, together with Theorem 1, reveal further details on the effects of all those environmental data on the dynamical behavior (ultimate upper and lower bounds) for the three biological species in Model (). REFERENCES [1] W. Feng, Coexistence, stability, and limiting behavior in a one- predator-two-prey model, Journal of Mathematical Analysis and Applications, 179 (1993), 59 69. [] W. Feng and X. Lu, Harmless Delays for Permanence in a class of Population Models with Diffusion Effects, Journal of Mathematical Analysis and Applications, 6 (1997), 57 566. [3] W. Feng, Permanence effect in a three-species food chain model, Applicable Analysis, 5 (199), 195 9. [] W. Feng and X. Lu, Some coexistence and extinction results in a three species ecological model, Differental and Integral Equations, 8 (1995), 617 66. [5] X. Lu, Monotone method and convergence acceleration for finite-difference solutions of parabolic problems with time delays, Numer. Meth. Part. Diff. Eqns. 11 (1995), 591 6. [6] X. Lu and W. Feng, Dynamics and numerical simulations of food chain populations, Appl. Math. Comp., 65 (199), 335 3. [7] C. V. Pao, On Nonlinear Parabolic and Elliptic equations, Plenum Press, New York, 199. [8] C. V. Pao, Numerical methods for semilinear parabolic equations,, SIAM J. Numer. Anal., (1987), 35. [9] E. C. Pielou, An Introduction to Mathematical Ecology, Wiley, New York, 1969. [1] J. Wiener and K. L. Cooke, Oscillations in Systems of Differential Equations with Piecewise Constant Arguments, Journal of Mathematical Analysis and Applications, 137 (1989), 1 39. Received September 6; revised May 7. E-mail address: fengw@uncw.edu