Loading [MathJax]/jax/output/SVG/jax.js
Research article

Dynamical transition for a 3-component Lotka-Volterra model with diffusion

  • Received: 23 October 2020 Accepted: 20 January 2021 Published: 08 February 2021
  • MSC : 35A09, 35Q35, 35J60, 35D35

  • The main objective of this article is to investigate the dynamical transition for a 3-component Lotka-Volterra model with diffusion. Based on the spectral analysis, the principle of exchange of stability conditions for eigenvalues is obtained. In addition, when δ0<δ1, the first eigenvalues are complex, and we show that the system undergoes a continuous or jump transition. In the small oscillation frequency limit, the transition is always continuous and the time periodic rolls are stable after the transition. In the case where δ0>δ1, the first eigenvalue is real. Generically, the first eigenvalue is simple and all three types of transition are possible. In particular, the transition is mixed if Ωe3k0dx0, and is continuous or jump in the case where Ωe3k0dx=0. In this case we also show that the system bifurcates to two saddle points on δ<δ1 as ˜θ>0, and bifurcates to two stable singular points on δ>δ1 as ˜θ<0 where ˜θ depends on the system parameters.

    Citation: Ruili Wu, Limei Li, Junyan Li. Dynamical transition for a 3-component Lotka-Volterra model with diffusion[J]. AIMS Mathematics, 2021, 6(5): 4345-4369. doi: 10.3934/math.2021258

    Related Papers:

    [1] Yaw Chang, Wei Feng, Michael Freeze, Xin Lu, Charles Smith . Elimination, permanence, and exclusion in a competition model under Allee effects. AIMS Mathematics, 2023, 8(4): 7787-7805. doi: 10.3934/math.2023391
    [2] Guillaume Cantin, Cristiana J. Silva . Complex network near-synchronization for non-identical predator-prey systems. AIMS Mathematics, 2022, 7(11): 19975-19997. doi: 10.3934/math.20221093
    [3] Haiping Pan, Yiqiu Mao . Transition and bifurcation analysis for chemotactic systems with double eigenvalue crossings. AIMS Mathematics, 2023, 8(10): 24681-24698. doi: 10.3934/math.20231258
    [4] Lini Fang, N'gbo N'gbo, Yonghui Xia . Almost periodic solutions of a discrete Lotka-Volterra model via exponential dichotomy theory. AIMS Mathematics, 2022, 7(3): 3788-3801. doi: 10.3934/math.2022210
    [5] M. Adel, M. M. Khader, Hijaz Ahmad, T. A. Assiri . Approximate analytical solutions for the blood ethanol concentration system and predator-prey equations by using variational iteration method. AIMS Mathematics, 2023, 8(8): 19083-19096. doi: 10.3934/math.2023974
    [6] Yinyin Wu, Fanfan Chen, Qingchi Ma, Dingbian Qian . Subharmonic solutions for degenerate periodic systems of Lotka-Volterra type with impulsive effects. AIMS Mathematics, 2023, 8(9): 20080-20096. doi: 10.3934/math.20231023
    [7] Nazmul Sk, Bapin Mondal, Abhijit Sarkar, Shyam Sundar Santra, Dumitru Baleanu, Mohamed Altanji . Chaos emergence and dissipation in a three-species food web model with intraguild predation and cooperative hunting. AIMS Mathematics, 2024, 9(1): 1023-1045. doi: 10.3934/math.2024051
    [8] Xuechao Zhang, Yuhan Hu, Shichang Lu, Haomiao Guo, Xiaoyan Wei, Jun li . Dynamic analysis and control of online information dissemination model considering information beneficiaries. AIMS Mathematics, 2025, 10(3): 4992-5020. doi: 10.3934/math.2025229
    [9] Li Wang, Hui Zhang, Suying Liu . On the existence of almost periodic solutions of impulsive non-autonomous Lotka-Volterra predator-prey system with harvesting terms. AIMS Mathematics, 2022, 7(1): 925-938. doi: 10.3934/math.2022055
    [10] Iqra Batool, Naim Bajcinca . Stability analysis of a multiscale model including cell-cycle dynamics and populations of quiescent and proliferating cells. AIMS Mathematics, 2023, 8(5): 12342-12372. doi: 10.3934/math.2023621
  • The main objective of this article is to investigate the dynamical transition for a 3-component Lotka-Volterra model with diffusion. Based on the spectral analysis, the principle of exchange of stability conditions for eigenvalues is obtained. In addition, when δ0<δ1, the first eigenvalues are complex, and we show that the system undergoes a continuous or jump transition. In the small oscillation frequency limit, the transition is always continuous and the time periodic rolls are stable after the transition. In the case where δ0>δ1, the first eigenvalue is real. Generically, the first eigenvalue is simple and all three types of transition are possible. In particular, the transition is mixed if Ωe3k0dx0, and is continuous or jump in the case where Ωe3k0dx=0. In this case we also show that the system bifurcates to two saddle points on δ<δ1 as ˜θ>0, and bifurcates to two stable singular points on δ>δ1 as ˜θ<0 where ˜θ depends on the system parameters.



    In this paper, we consider the following 3-component Lotka-Volterra model with diffusion

    {ut=d1Δu+u(a1b1uc1vk1w),vt=d2Δv+v(a2c2ub2vk2w),wt=d3Δw+w(r+α1k1u+α2k2v), (1.1)

    where u, v are the population densities of two competing prey and w is the population density of its predator. The habitat ΩRn is a bounded domain with smooth boundary Ω. r,ai,bi,ci,αi,ki(i=1,2) and dj(j=1,2,3) are positive constants. a1 and a2 represent the intrinsic growth rate, bi and ci(i=1,2) represent the intra-specific and inter-specific competition rates of u and v, k1 and k2 are the predation rate of w, α1 and α2 are the transformation rate of predation, and r is the death rate of w. d1, d2 and d3 are represent the diffusion rates of u, v and w respectively.

    Here, we focus on the system (1.1) supplemented with the following initial condition:

    u(x,0)=u0,v(x,0)=v0,w(x,0)=w0, (1.2)

    and the Neumann boundary condition:

    un|Ω=vn|Ω=wn|Ω=0, (1.3)

    where n is the outward normal derivative on Ω. The Neumann boundary condition in (1.3) was interpreted as the condition that the system is self-contained with zero population flux across the boundary.

    For two species Lotka-Volterra (LV) systems, there has been largely discussed in the past several decades. Kuto and Tsujikawa [1] considered a general stationary Lotka-Volterra competition model with diffusion. They obtained the existence of nonconstant solutions by the Leray-Schauder degree theory and derived a limiting system as diffusion of one of the species tending to infinity. Eilbeck et al.[2] studied the two species Lotka-Volterra competition model and obtained the existence and non-uniqueness of coexistence solutions for a wide range of parameters. Other results related to the two species Lotka-Volterra competition model, we refer to other studies [3,4,5,6] and the references therein.

    However, very little result is known about the three species Lotka-Volterra model. In general, three species systems are very complicated even in the ordinary differential equations case. In recent years, the three species Lotka-Volterra model with diffusion was studied by some investigators. Ni et al.[7] considered the role of cross-diffusion in the 3 × 3 Lotka-Volterra competition model, and obtained the existence of non-constant steady states created by cross-diffusion in 3 × 3 systems. In [8], Pang and Wang studied a strongly coupled system of a two-predator-one-prey ecosystem, they demonstrated the emergence of stationary patterns for this system, and showed that it is due to the cross diffusion that arises naturally in the model. Moreover, Ali et al.[9] studied the prey-predator-top-predator system, in addition, they point out that the system exhibits Bogdanov-Takens bifurcation, saddle-node bifurcation, Hopf bifurcation for suitable choice of the relevant parameters. There are other related works on three species model, see [10,11,12,13,14,15,16,17] and references therein.

    Although considerable work has been done concerning three competition model and two-predator-one-prey ecosystem, it is worth noting that it is interesting to investigate two prey and one predator system. In some circumstances, predation may have a tendency to increase species diversity in competitive communities, which is called predation mediated coexistence. For instance, in [10], the authors considered the coexistence problem of two competing species mediated by the presence of predator, and speculate that the possibility. Furthermore, Kan-on and Mimura [11] proved the existence of stable spatially inhomogeneous positive stationary solutions of (1.1). In addition, Yukio Kan-on [12] studied the positive stationary solutions by using the singular perturbation method and the associated singular limit spectral analysis. Moreover, Wang [13] considered the strongly coupled version of (1.1) and established the existence and non-existence of non-constant positive solutions.

    Motivated by the above papers, what we are concerned in this paper is to describe the dynamic stability and transition for the system (1.1). The technical method for the analysis is the dynamical transition theory, which has been developed by Ma and Wang [18,19,20] and has been used to solve many interesting mathematical and physical problems, see [21,22,23,24,25,26]. As is well known, for the system (1.1), due to non-selfadjoint linear operator, the transition can be caused by real or complex eigenvalues crossing the imaginary axis. Let δ0 and δ1 be critical values given by (3.17) and (3.20). δ0 is derived from ABC=0, which determine the signs of the eigenvalues of the linearized eigenvalue equations (3.7). The same is δ1, which is derived from Ck=0. When δ0>δ1, the first eigenvalue is real and simple, and all three types of transition are possible depending on a non-dimensional number exactly given in terms of the system parameters. In particular, the transition is mixed if Ωe3k0dx0, and in the case where Ωe3k0dx=0 we show that the transition of the system is continuous as ˜θ<0, and is jump as ˜θ>0 where ˜θ is defined in (4.43). when δ0<δ1, the first eigenvalues are complex, and we show that the system undergoes a continuous or jump transition. In the small oscillation frequency limit, the transition is always continuous and the time periodic rolls are stable after the transition.

    The rest of this paper is organized as follows. In Section 2, we present some preliminaries on dynamical transition theory. Section 3 recapitulates (1) the nondimensional form and the nonnegative basic states of the steady-state equations for the system (1.1), (2) an abstract form for (1.1), and (3) linear theory and principle of exchange of stabilities(PES). The main theorems of this artical are stated and proved in Section 4.

    In this section, we introduce the dynamical transition theory for nonlinear dissipative systems developed by Ma and Wang [18,19,20], which provides the basic method for the following research of this paper.

    Let H and H1 be two Hilbert spaces, H1H be a compact and dense inclusion. Consider the following abstract nonlinear equation defined on H, given by

    {dχdt=Lλχ+G(χ,λ),χ(0)=χ0, (2.1)

    where χ(t) is an unknown function, Lλ:H1H is a linear operator, and G:H1H is a nonlinear operator, λ is the system parameter.

    Assume that Lλ:H1H is a parameterized linear completely continuous field depending continuously on λ. which satisfies

    {Lλ=A+B  is a sectorial operator,A:H1H    is a linear homeomorphism,B:H1H    is a compact operator. (2.2)

    Furthermore, we assume that the nonlinear term G:HσH(0σ<1) is a Cr bounded operators (r1) and depending continuously on λ, where Hσ is the fractional order space, and

    G(χ,λ)=o(χHσ). (2.3)

    Hereafter, we always assume the conditions (2.2) and (2.3) hold true, which imply that the system (2.1) has a dissipative structure.

    At first, we recall the mathematical definition of transition for the system (2.1).

    Definition 2.1. ([20]) We say that the system (2.1) has a transition from (χ,λ)=(0,λ0) at λ0 if the following two conditions hold true:

    (1) if λ<λ0, χ=0 is locally asymptotically stable for (2.1), and

    (2) if λ>λ0, there exists a neighborhood UH of χ=0 independent of λ, such that for any χ0UΓλ the solution χλ(t,χ0) of (2.1) satisfies that

    lim suptχλ(t,χ0)Hδ(λ)>0,limλλ0δ(λ)0, (2.4)

    where Γλ is the stable manifold of χ=0, with codim Γλ1 in H for λ>λ0.

    Let the eigenvalues (counting multiplicity) of Lλ be given by {βj(λ)C|j=1,2,}, and let

    Reβi(λ){ >0,ifλ>λ0,=0,ifλ=λ0,<0,ifλ<λ0,    for any  1im, (2.5)
    Reβj(λ0)<0,for any  jm+1. (2.6)

    The following theorem is a basic principle of transitions from equilibrium states, which provides sufficient conditions and a basic classification for transitions of nonlinear dissipative systems. The proof of this Lemma is given in Ma and Wang [19,20].

    Lemma 2.1. Let the conditions (2.5) and (2.6) hold true. Then the system (2.1) must have a transition from (χ,λ)=(0,λ0), and there is a neighborhood UH of χ=0 such that the transition is one of the following three types:

    (1) Continuous transition: There exists an open and dense set ~UλU such that for any χ0~Uλ, the solution χλ(t,χ0) of (2.1)satisfies

    limλλ0lim supt||χλ(t,χ0)||H=0.

    (2) Jump transition: For any λ0<λ<λ0+ε with some ε>0, there is an open and dense set UλU such that for any χ0Uλ,

    lim supt||χλ(t,χ0)||Hδ>0,for some δ>0 is independent of λ.

    (3) Mixed transition: For any λ0<λ<λ0+ε with some ε>0, U can be decomposed into two open sets Uλ1 and Uλ2 (Uλi not necessarily connected): ˉU=ˉUλ1+ˉUλ2,Uλ1Uλ2=, such that

    limλλ0lim supt||χλ(t,χ0)||H=0,χ0Uλ1,
    lim supt||χλ(t,χ0)||Hδ>0,χ0Uλ2,

    where Uλ1 and Uλ2 are called metastable domain.

    The forthcoming analysis is to formulate the evolution equations given in (1.1) using an abstract functional setting that is standard in the framework of dynamic transitions.

    First, we introduce a set of the following to nondimensionalize the system (1.1):

    u=a1b1u,v=a2b2v,w=a1k1w,x=lx,t=t0t(t0=b1α1k1a1),

    where the prime denotes nondimensionalized variables. Substituting these nondimensional variables into (1.1) and neglecting the prime for all variables for convenience, we obtain

    {ut=ε1Δu+ˉa1u(1uαvw),vt=ε2Δv+ˉa2v(1γuvκw),wt=ε3Δw+w(σ+u+qv), (3.1)

    here

    ε1=d1t0l2,ˉa1=a1t0,α=a2c1a1b2,ε2=d2t0l2,ˉa2=a2t0,γ=a1c2a2b1,ε3=d3t0l2,σ=rt0,κ=a1k2a2k1,q=a2b1α2k2a1b2α1k1,

    and the unknown functions are u, v, w0, ΩRn is a bounded domain, the parameters are positive constants:

    εi(1i3),ˉa1,ˉa2,α,γ,κ,σ,q.

    Furthermore, let

    Rm+={(x1,,xm)Rm+|xi0,1im},λ=(ε1,ε2,ε3,ˉa1,ˉa2,α,γ,κ,σ,q)R10+.

    Then we define the following function spaces,

    H=L2(Ω)3,H1={χH2(Ω)3|χn=0onΩ},

    where χ=(u,v,w).

    Define the operators Lλ=Aλ+Bλ and Gλ:H1H by

    Aλχ=(ε1Δu,ε2Δv,ε3Δw),Bλχ=(ˉa1u,ˉa2v,σw),G(χ,λ)=(ˉa1u2ˉa1αuvˉa1uw,ˉa2γuvˉa2v2ˉa2κvw,uw+qvw).

    Obviously, Lλ:H1H is a parameterized linear completely continuous field depending continuously on λ, and Gλ:H1H represents the nonlinear terms of the Eq (3.1).

    Thus, the Eqs (3.1) with (1.2) and (1.3), take the following operator form:

    {dχdt=Lλχ+G(χ,λ),χ(0)=φ, (3.2)

    where λ=(ε1,ε2,ε3,ˉa1,ˉa2,α,γ,κ,σ,q)R10+.

    On the other hand, we study the steady-state solutions for the system (3.1). It is easy to check that the system (3.1) admits seven biologically realistic constant steady-state solutions:

    φ0=(0,0,0)T,φ1=(0,σq,qσκq)T,φ2=(σ,0,1σ)T,φ3=(1α1αγ,1γ1αγ,0)T,φ4=(1,0,0)T,φ5=(0,1,0)T,φ6=(u0,v0,w0)T, (3.3)

    where

    u0=σ(1ακ)q(κ1)q(γκ)(1ακ),v0=(κ1)+σ(γκ)q(γκ)(1ακ),w0=qγ1+σq+ασαγq(γκ)(1ακ).

    Biologically, only positive solutions (u0>0,v0>0,w0>0) are of interest in the competition of biological population. Hence, we make the natural assumption: u0>0,v0>0,w0>0.

    In this paper, we mainly focus on the bifurcation and transition problem of (3.1) at the more general positive steady-state solution φ6 in (3.3).

    For this purpose, we take the translation

    u=u+u0,v=v+v0,w=w+w0, (3.4)

    Omitting the primes, the system (3.1) with (1.2)(1.3) becomes

    {ut=ε1Δuˉa1u0uˉa1αu0vˉa1u0wˉa1u2ˉa1αuvˉa1uw,vt=ε2Δvˉa2γv0uˉa2v0vˉa2κv0wˉa2γuvˉa2v2ˉa2κvw,wt=ε3Δw+w0u+qw0v+uw+qvw, (3.5)

    with the initial-boundary conditions

    u(x,0)=u0u0,v(x,0)=v0v0,w(x,0)=w0w0,un|Ω=vn|Ω=wn|Ω=0. (3.6)

    Then it suffices to study the bifurcation solution of (3.5) at the steady-state solution χ=(0,0,0)T.

    The linearized eigenvalue equations of (3.5) are given by

    {ε1Δuˉa1u0uˉa1αu0vˉa1u0w=β(λ)u,ε2Δvˉa2γv0uˉa2v0vˉa2κv0w=β(λ)v,ε3Δw+w0u+qw0v=β(λ)w. (3.7)

    Let ρk and ek be the kth eigenvalue and eigenvector of the Laplace operator Δ with the Neumann boundary condition:

    {Δek=ρkek,   (ρk0),ekn|Ω=0. (3.8)

    Let Mk be the matrix given by

    Mk=(ε1ρkˉa1u0ˉa1αu0ˉa1u0ˉa2γv0ε2ρkˉa2v0ˉa2κv0w0qw0ε3ρk). (3.9)

    Thus, all eigenvalues β(λ)=βki(λ) of (3.7) satisfy

    Mkηki=βki(λ)ηki,   1i3,  k=1,2,, (3.10)

    where ηkiR3 is the eigenvector of Mk corresponding to βki(λ). Hence, the eigenvector ψki of (3.7) corresponding to βki(λ) is

    ψki(x)=ηkiek(x),    k=1,2,,  1i3, (3.11)

    where ek(x) is as in (3.8).

    In particular, ρ1=0 and e1 is a constant, and

    M1=(ˉa1u0ˉa1αu0ˉa1u0ˉa2γv0ˉa2v0ˉa2κv0w0qw00). (3.12)

    By simple calculation, it is not difficult to find that the eigenvalues β(λ)=β1i(λ)(i=1,2,3) satisfy the following equation:

    β1i(λ)3+Aβ1i(λ)2+Bβ1i(λ)+C=0, (3.13)

    where

    A=ˉa2v0+ˉa1u0,B=ˉa1ˉa2u0v0+ˉa1u0w0ˉa1ˉa2αγu0v0+ˉa2κqv0w0,C=ˉa1ˉa2(καγq+1+κq)u0v0w0. (3.14)

    It is known that all solutions of (3.13) have negative real parts if and only if

    A>0,   C>0,   ABC>0. (3.15)

    If we suppose q(γκ)(1κα)<0 in (3.3), then, direct calculation indicates that these two parameters A and C in (3.14) are positive

    A>0,   C>0. (3.16)

    Note that

    β11(λ)β12(λ)β13(λ)=C<0,

    which implies that all real eigenvalues of (3.12) do not change their signs, and at least one of these real eigenvalues is negative.

    In addition, let δ=αγ=c1c2b1b2, and we can derive from ABC=0, the critical number

    δ0=1+ˉa22κqv20w0+ˉa21u20w0+ˉa1ˉa2(κα+γq)u0v0w0ˉa1ˉa22u0v20+ˉa21ˉa2u20v0. (3.17)

    It is then clear that

    ABC{<0   ifδ>δ0,=0   ifδ=δ0,>0   ifδ<δ0. (3.18)

    Next, we check the other eigenvalues βkj(λ) with j=1,2,3,k2. By calculation, the eigenvalues βkj(λ)(j=1,2,3,k2) of (3.9) satisfy

    βkj(λ)3+Akβkj(λ)2+Bkβkj(λ)+Ck=0, (3.19)

    where

    Ak=A+(ε1+ε2+ε3)ρk,Bk=B+(ε1ε3+ε2ε3+ε1ε2)ρ2k  +(ˉa1u0ε3+ˉa2v0ε3+ˉa2v0ε1+ˉa1u0ε2)ρk,Ck=C+ε1ε2ε3ρ3k+(ε1ε3ˉa2v0+ε2ε3ˉa1u0)ρ2k  +ˉa1ˉa2(1δ)u0v0ε3ρk+ˉa1u0w0ε2ρk+ˉa2κqv0w0ε1ρk.

    We introduce another critical number

    δ1=minρk0[1+C+τ1ρ3k+τ2ρ2k+τ3ρkˉa1ˉa2u0v0ε3ρk], (3.20)

    where

    τ1=ε1ε2ε3,τ2=(ε1ε3ˉa2v0+ε2ε3ˉa1u0),τ3=ˉa1u0w0ε2+ˉa2κqv0w0ε1. (3.21)

    The following lemma characterizes the principle of exchange stability (PES) for the eigenvalue Eq (3.7).

    Lemma 3.1. Assume that q(γκ)(1κα)<0, let δ0 and δ1 be the numbers given by (3.17) and (3.20), then the eigenvalues βki(i=1,2,3,k1) of Lλ satisfy the following properties:

    (1) If δ1<δ0 and k0 be the integer that δ1 in (3.20) reaches its minimum at ρk0, then βk01 is a real eigenvalue of (3.7), and

    βk01(δ){>0   ifδ>δ1,=0   ifδ=δ1,<0   ifδ<δ1,   forρk=ρk0,Reβij(δ1)<0  (i,j)(k0,1)withρk=ρk0.

    (2) If δ1>δ0, then β11(δ)=ˉβ12(δ) are a pair of complex eigenvalues of (3.7), and

    Reβ11(δ)=Reβ12(δ){>0   ifδ>δ0,=0   ifδ=δ0,<0   ifδ<δ0,Reβij(δ0)<0  (i,j)(1,1),(1,2).

    Proof. According to the assumption, C is positive. We see that

    A=ˉa2v0+ˉa1u0,B=ˉa1ˉa2u0v0+ˉa1u0w0ˉa1ˉa2αγu0v0+ˉa2κqv0w0.

    By the direct calculation, we can see that

    Ak>0,   AkBkCk>0,   k2.

    As δ1<δ0, we infer from (3.16) and (3.18) that

    Reβ1j(δ1)<0,   1j3.

    In addition, it is clear that there must exist a k0 satisfying (3.20), furthermore,

    Ck0(δ){<0   ifδ>δ1,=0   ifδ=δ1,>0   ifδ<δ1,Ck(δ1)>0   for all  kk0,

    thus, assertion (1) is approved.

    As δ1>δ0, through the analysis of above, we know that Ck>0 at δ=δ0 for all k2. Since all real eigenvalues β1j(1j3) of (3.12) do not change their signs, and at least one of these real eigenvalues is negative, so the condition (3.18) implies that there exists a pair of complex eigenvalues β11=ˉβ12 crossing the imaginary axis at δ=δ0. Then assertion (2) follows. The Lemma is proved.

    The following theorems will show the types of transition that the system (3.5) undergoes as the bifurcation parameter δ crosses the critical value δ0 or δ1 basing on Lemma 3.1. Hereafter, we will give different transition theorems basing on Lemma 3.1.

    First, we consider the case that δ0<δ1.

    By Lemma 3.1, as δ0<δ1, the first critical eigenvalues will be a pair of complex numbers β11 and β12, the problem (3.5) undergoes a dynamic transition to a periodic solution from δ0. To determine the types of transition, we introduce a parameter b0 which is defined by (4.30) as follows:

    b0=1D2D0[ρ2ˉa21u30(A(3F1+F3)E2+ρa˜b(F1+3F3)+ρF2E2)+ρw0ˉa1u0b˜bˉa2(3E5F1+E5F3+E7F2)+a˜b(3F1E6+F3E6F2E8ρ2Aˉa21u30F2)+(3F3+F1)(E3E7+E2E8)+F2(E3E5E2E6)]+2b˜bD4ˉa1u0[ρ2w0ˉa21u20a(ˉa2(a+κw0b)+qa)(E3η2(η2+κη3)E1E2)(w0ˉa2ξ2(γ+ξ2+κξ3)+ρ2ˉa1u0aξ3)(E3ξ2(γ+ξ2+κξ3)+E1E2)]+1D4ˉa1u0ˉa2[ξ2(γ+ξ2+κξ3)+η2(η2+κη3)](ρ3ˉa21u30˜b+ρˉa1u0ˉa2E7+E4˜a)(a2+2w0ab˜a+w20b2(˜a2+˜b2)), (4.1)

    where

    D0=ˉa1u0ˉa2v0w0(qα)(κα1)+A2(Aαqw0)qw20(ˉa2v0κq+ˉa1u0)ˉa1u0(Aαqw0)2,D1=[ˉa2ξ2(γ+ξ2+κξ3)ρ2qw0ˉa21u20abˉa1u0+ˉa2v0ακAαqw0],D2=ρˉa1u0[ρ2w0ˉa1u0b2ˉa1u0+ˉa2v0ακAαqw0ˉa1u0A+ˉa2v0w0κqu0(Aαqw0)ˉa2(γa+(a+κb)(ρ2ˉa1u0qb1q))],D3=ρˉa1u0a[q2w0ˉa1u0aˉa1u0+ˉa2v0ακAαqw0+ˉa2(ξ2+κξ3)],
    D4=ˉa1u0+ˉa2v0ακAαqw0qη2η3ˉa2η2(η2+κη3),D2=ρ2ˉa21u20(w20b2˜b2+(a+w0b˜a)2),E1=qρ3w0ˉa31u30ab,E2=a˜a+w0b(˜a2+˜b2),E3=ρˉa1u0(a+w0b˜a)ˉa2,E4=ρ2w0ˉa31u30(ρ2b2q2a2),E5=Av0κˉa1u20(γ(ξ2+ζ2)+2ξ2ζ2+κ(ξ2ζ3+ξ3ζ2)),E6=ρw0ˉa31u30ˉa1u0α+ˉa2v0qAαqw0(ρ2b+Aqa),E7=ρv0κˉa1u20(γη2+2η2ξ2+κ(η2ξ3+η3ξ2)),E8=ρ2w0ˉa31u30ˉa1u0α+ˉa2v0qAαqw0(qa+Ab),F1=D1A+2ρ2(D4D1)A(A2+4ρ2)ρ(D2+D3)A2+4ρ2,F2=D2+D3A4ρ2(D2+D3)A(A2+4ρ2)+2ρ(D1D4)A2+4ρ2,F3=D4A+2ρ2(D1+D4)A(A2+4ρ2)+ρ(D2+D3)A2+4ρ2. (4.2)

    Here A is as in (3.14), and a,b,˜a,˜b are as in (4.6) and (4.8).

    Then, we have the following dynamic transition theorem.

    Theorem 4.1. Consider b0 which is given by (4.1). Let δ0<δ1. Assume q(γκ)(1κα)<0 and assume that the critical index are (k,j)=(1,1) and (k,j)=(1,2), then the problem (3.5) undergoes a transition to periodic solutions at δ=δ0, and the following assertions holds true:

    (1) If b0<0, then the transition of (3.5) is continuous, and the system bifurcates to a periodic solution on δ>δ0, which is an attractor.

    (2) If b0>0, then the transition of (3.5) is jump, and the system bifurcates on δ<δ0 to a unique unstable periodic orbit.

    Proof. We shall prove the theorem in the following two steps.

    Step 1. Calculate the critical eigenvectors.

    By Lemma 3.1, at δ0 there is a pair of imaginary eigenvalues β11=ˉβ12=iρ of (3.7). Let z=ξ+iη and z=ξ+iη be the eigenvectors and conjugate eigenvectors of (3.7) corresponding to iρ, i.e. z and z satisfy that

    (M1+iρ)z=0,(M1iρ)z=0. (4.3)

    For z=(z1,z2,z3), from the first equation of (4.3) we obtain

    {[(qα)iρqˉa1u0]z1=[q+iραw0]z3,z1+qz2=iρw0z3. (4.4)

    Thus, we derive from (4.4) the eigenvectors z=ξ+iη as follows:

    ξ=(ξ1,ξ2,ξ3)=(1,ρ2ˉa1u0qb1q,qw0ˉa1u0a),η=(η1,η2,η3)=(0,ρˉa1u0a,ρw0ˉa1u0b), (4.5)

    where

    a=ˉa1u0w0(qα)+ρ2αq2w20+ρ2α2,b=q2w0ˉa1u0α(qα)q2w20+ρ2α2. (4.6)

    In the same fashion, we derive from the second equation of (4.3) the conjugate eigenvectors z=ξ+iη as

    ξ=(ξ1,ξ2,ξ3)=(ρ˜bˉa2v0κˉa1u0,1,˜a),η=(η1,η2,η3)=(ρ˜aˉa1u0,0,˜b), (4.7)

    where

    ˜a=ˉa2v0(1ακ)qw0+ρ2αq2w20+ρ2α2,˜b=ˉa2v0ρα(1ακ)+ρqw0q2w20+ρ2α2. (4.8)

    It is easy to show that

    ξ,ξ=η,η=ρw0ˉa1u0b˜b,ξ,η=η,ξ=(ρˉa1u0a+ρw0ˉa1u0b˜a). (4.9)

    It is known that functions ψ1+iψ2 given by

    ψ1=1ξ,ξ[ξ,ξξ+ξ,ηη],ψ2=1η,η[η,ξξ+η,ηη], (4.10)

    also satisfy the second equation of (4.3) with

    ξ,ψ1=η,ψ20,ξ,ψ2=η,ψ1=0. (4.11)

    On the other hand, we know that

    β13(iρ)(iρ)=ρ2β13=C. (4.12)

    In addition, because ±iρ are solutions of (3.13), and ABC=0 at δ0, we deduce that

    ρ2=B=CA. (4.13)

    Then, we obtain

    β13=A=(ˉa2v0+ˉa1u0). (4.14)

    From the equation

    (M1β13)ζ=0, (4.15)

    we derive the eigenvector

    ζ=(ζ1,ζ2,ζ3)=(1,ˉa1u0w0+ˉa2v0Aˉa1u0(Aαqw0),w0ˉa1u0ˉa1u0α+ˉa2v0qAαqw0). (4.16)

    In the same fashion, from

    (M1β13)ζ=0, (4.17)

    we derive the conjugate eigenvector of β13 as follows:

    ζ=(ζ1,ζ2,ζ3)=(ˉa2v0w0κq+ˉa1u0Aˉa1u0(Aαqw0),1,ˉa1u0+ˉa2v0ακAαqw0). (4.18)

    Step 2. Derivation of evolution equation.

    Let χH be a solution of (3.5) expressed as

    χ=xξ+yη+Φ(x,y),    x,yR1,

    where Φ(x,y) is the center manifold function of (3.5) at δ0.

    Based on the center manifold reduction, the reduced equations of (3.5) on the center manifold are given by

    dxdt=ρy+1ξ,ψ1G(xξ+yη+Φ),ψ1,dydt=ρx+1η,ψ2G(xξ+yη+Φ),ψ2, (4.19)

    where G(χ)=G(χ,χ) is the bilinear operator defined by

    G(χ,χ1)=(ˉa1u1u2ˉa1αu1v2ˉa1u1w2,ˉa2γu1v2ˉa2v1v2ˉa2κv1w2,u1w2+qv1w2), (4.20)

    for χ=(u1,v1,w1), χ1=(u2,v2,w2)H1.

    Now we are in position to solve the center manifold function Φ(x,y). By the approximation formula (B.4.1) in [20], the center manifold function Φ satisfy

    Φ=Φ1+Φ2+Φ3+o(2), (4.21)

    where

    Φ1=x2P2G11xy(P2G12+P2G21)y2P2G22,(2+4ρ2)Φ2=2ρ2(x2y2)P2G11+4ρ2xy(P2G12+P2G21)+2ρ2(y2x2)P2G22,(2+4ρ2)Φ3=ρ(y2x2)(P2G12+P2G21)+2ρxy(P2G11P2G22), (4.22)

    here P2:HE2 is the canonical projection, E2 is the orthogonal complement of E1=span{ξ,η}, and is the linearized operator of (3.5).

    Remark 4.1. In (4.21), o(2) represent o(|(x,y)|2). hereafter, we make the following convention

    o(m)=o(|x|m)    for xRn near 0. (4.23)

    Direct calculation shows that

    ζ,ζ=D0,G11,ζ=D1,  G12,ζ=D2,G21,ζ=D3,  G22,ζ=D4

    and D0,D1,D2,D3,D4 are as in (4.2).

    By (4.5), (4.18) and (4.20), it is clear that

    P2G11=G11,ζζ=D1ζ,P2G12=G12,ζζ=D2ζ,P2G21=G21,ζζ=D3ζ,P2G22=G22,ζζ=D4ζ. (4.24)

    Hence, Φ1,Φ2,Φ3span{ζ}, which implies

    Φj=M1Φj=AΦj. (4.25)

    We infer from (4.21)(4.25) the center manifold function as follows:

    Φ=ζD0[(D1A+2ρ2(D4D1)A(A2+4ρ2)ρ(D2+D3)A2+4ρ2)x2+(D2+D3A4ρ2(D2+D3)A(A2+4ρ2)+2ρ(D1D4)A2+4ρ2)xy+(D4A+2ρ2(D1+D4)A(A2+4ρ2)+ρ(D2+D3)A2+4ρ2)y2]+o(2)=1D0(F1x2+F2xy+F3y2)ζ+o(2), (4.26)

    where F1, F2, F3 are as in (4.2).

    Inserting (4.26) into (4.19), by direct circulation, we have

    dxdt=ρy+1D2{[ξ,ξG11,ξ+ξ,ηG11,η]x2+[ξ,ξG22,ξ+ξ,ηG22,η]y2+[ξ,ξG12+G21,ξ+ξ,ηG12+G21,η]xy+ξ,ξG(ξ,ζ)+G(ζ,ξ),ξ1D0(F1x3+F2x2y+F3xy2)+ξ,ηG(ξ,ζ)+G(ζ,ξ),η1D0(F1x3+F2x2y+F3xy2)+ξ,ξG(η,ζ)+G(ζ,η),ξ1D0(F1x2y+F2xy2+F3y3)+ξ,ηG(η,ζ)+G(ζ,η),η1D0(F1x2y+F2xy2+F3y3)}+o(3), (4.27)
    dydt=ρx+1D2{[ξ,ηG11,ξ+ξ,ξG11,η]x2+[ξ,ηG22,ξ+ξ,ξG22,η]y2+[ξ,ηG12+G21,ξ+ξ,ξG12+G21,η]xyξ,ηG(ξ,ζ)+G(ζ,ξ),ξ1D0(F1x3+F2x2y+F3xy2)+ξ,ξG(ξ,ζ)+G(ζ,ξ),η1D0(F1x3+F2x2y+F3xy2)ξ,ηG(η,ζ)+G(ζ,η),ξ1D0(F1x2y+F2xy2+F3y3)+ξ,ξG(η,ζ)+G(ζ,η),η1D0(F1x2y+F2xy2+F3y3)}+o(3), (4.28)

    where D2=ξ,ξ2+ξ,η2.

    In view of (4.9), Eqs (4.27) and (4.28) become

    dxdt=ρy+a20x2+a11xy+a02y2+a30x3+a21x2y+a12xy2+a03y3+o(3),dydt=ρx+b20x2+b11xy+b02y2+b30x3+b21x2y+b12xy2+b03y3+o(3), (4.29)

    where

    a20=1D2ρˉa1u0b˜b[w0ˉa2ξ2(γ+ξ2+κξ3)ρ2ˉa1u0aξ3],a02=1D2ρ3w0ˉa31u30ab˜b[ˉa2(a+κw0b)+qa],a11=1D2[E5E2+ρ2w0ˉa21u20b˜bˉa2E4E6a˜b],a30=F1D2D0[ρ2Aˉa21u30E2+ρw0ˉa1u0b˜bˉa2E7+E8a˜b],a03=F3D2D0[ρ3ˉa21u30E2+ρw0ˉa1u0b˜bˉa2E9E10a˜b],b20=1D2[E3ξ2(γ+ξ2+κξ3)E1E2],b02=1D2[E3η2(η2+κη3)+E1E2],
    b11=1D2[E5a˜b+ρˉa1u0E3E4+E2E6],b30=F1D2D0[ρ2Aˉa21u30a˜b+E3E7E2E8],b03=F3D2D0[ρ3ˉa21u30a˜b+E3E9+E2E10],a21=F2F1a30+F1F3a03,a12=F3F1a30+F2F3a03,b21=F2F1b30+F1F3b03,b12=F3F1b30+F2F3b03.

    The transition of (3.5)(3.6) is determined by the sign of the following number; see [20],

    b0=3(a30+b03)+(a12+b21)+2ρ(a02b02a20b20)+1ρ(a11a20+a11a02b11b20b11b02), (4.30)

    which is the same as that given by (4.1). From (4.1) and (4.2), it is easy to show that b0<0 in the limit of small ρ. Thus the proof is complete.

    Second, we consider the case that δ0>δ1.

    Thanks to Lemma 3.1, for δ0>δ1, the transition of the system (3.5) occurs at δ1, which is from real eigenvalues. The following theorems will show the types of transition that the system (3.5) undergoes as the bifurcation parameter δ crosses the critical value δ1

    δ1=minρk0[1+C+τ1ρ3k+τ2ρ2k+τ3ρkˉa1ˉa2u0v0ε3ρk].

    Let δ1 achieve it minimum at ρk0, ρk0 be the eigenvalues of (3.8), ek0 be the eigenvector of (3.8) corresponding to ρk0. Assume that βk01 is simple near δ1. Hereafter, we will give different transition theorems basing on the two cases about ek0.

    First, we consider the case where

    Ωe3k0dx0. (4.31)

    For simplicity, let

    θ=PkΩe3k0dxQkΩe2k0dx, (4.32)

    where

    Pk=ˉa21u0w0a2c2[ˉa1u0w0ac+αˉa2v0w0bd+w20ab]+ˉa22v0w0b2cd[γˉa1u0w0ac+ˉa2v0w0bd+κw20ab]+w20ab2e[ˉa1u0w0ac+qˉa2v0w0bd],Qk=[ˉa1u0w0a2c2ˉa2v0w0b2cd+w20ab2e], (4.33)

    and a,b,c,d,e in (4.33) are given by

    a=(ε2ρk0ˉa2v0)+ˉa2v0βq,b=(ε1ρk0ˉa1u0)q+ˉa1u0α,c=ε3ρk0α+qw0,d=ε3ρk0β+κw0,e=(ε2ρk0ˉa2v0)+ˉa2v0κα. (4.34)

    Then, under the condition (4.31), we have the second dynamic transition theorem.

    Theorem 4.2. Assume that q(γκ)(1κα)<0. Let δ0>δ1. If θ0, then we have the following assertions:

    (1) The transition of (3.5) at δ=δ1 is mixed. More precisely, there exists a neighborhood UH1 of χ=0 such that U is separated into two disjoint open sets U1 and U2 by the stable manifold Γ of χ=0 satisfying

    (a) U=U1+U2+Γ,

    (b) the transition in U1 is jump, and in U2 is continuous.

    (2) The system (3.5) bifurcates in U2 to a unique singular point ˉχ on δ>δ1, which is an attractor such that for any initial value φU2,

    limtχ(t,φ)ˉχH1=0.

    (3) The system (3.5) bifurcates on δ<δ1 to a unique saddle point ˉχ.

    (4) The bifurcated singular point ˉχ can be expressed as

    ˉχ=βk01θξek0+o(|βk01|), (4.35)

    where θ is as in (4.32) and ξ=(ξ1,ξ2,ξ3) are given by

    ξ1=ˉa1u0w0[(ε2ρk0ˉa2v0)+ˉa2v0γq][ε3ρk0α+qw0],ξ2=ˉa2v0w0[(ε1ρk0ˉa1u0)q+ˉa1u0α][ε3ρk0γ+κw0],ξ3=w20[(ε2ρk0ˉa2v0)+ˉa2v0γq][(ε1ρk0ˉa1u0)q+ˉa1u0α].

    Proof. We shall prove the theorem in the following two steps.

    Step 1. Calculate the critical eigenvectors and decompose space.

    Let ξ and ξR3 be the eigenvectors of Mk0 and Mk0 corresponding to βk01(δ1)=0, i.e.

    Mk0ξ=0,Mk0ξ=0,

    where Mk0 is the matrix (3.9) with k=k0. It is easy to see that ξ is as in (4.35), and

    ξ=(ξ1,ξ2,ξ3),ξ1=[(ε2ρk0ˉa2v0)+ˉa2v0γq][ε3ρk0α+qw0],ξ2=[(ε1ρk0ˉa1u0)q+ˉa1u0α][ε3ρk0α+qw0],ξ3=[(ε2ρk0ˉa2v0)+ˉa2v0κα][(ε1ρk0ˉa1u0)q+ˉa1u0α]. (4.36)

    On the other hand, based on the spectral theory of linear completely continuous field, the spaces H and H1 can be decomposed into the following form:

    H=E1E2,H1=E1ˉE2,

    where E1=span{ξek0}, E2=E1. Then near δ1, the solution of the Eq (3.5) can be expressed as

    χ=xξek0+z,z=3j=2xk0jψk0j+3kk0,j=1xkjψkj, (4.37)

    where xξek0E1, zE2. ψkj(j=1,2,3,kk0) is the eigenvector corresponding to the eigenvalue βkj.

    Thus, in the space E1, the Eq (3.5) can be reduced to

    ξek0,ξek0dxdt=Lδ(χ),ξek0+G(χ),ξek0=βk01ξek0,ξek0x+G(χ),ξek0. (4.38)

    Note that , denotes the inner product in H.

    Step 2. Derivation of evolution equation.

    According to the condition (4.31), we do not need to consider the influence of the center manifold function. That is to say, we let χ=xξek0 in (4.38).

    Hence, we derive the following reduced bifurcation equation

    dxdt=βk01x+G(xξek0),ξek0ξek0,ξek0. (4.39)

    For the operator G, we can derive that

    G(xξek0),ξek0ξek0,ξek0=θx2+o(2).

    where θ is as in (4.32).

    Thus, the Eq (4.39) can be rewritten as

    dxdt=βk01x+θx2+o(2). (4.40)

    It is known that the transition of the Eq (3.5) and its local topological structure are determined completely by (4.40). If θ0, it is clear that (4.40) has exactly a bifurcated solution as follows:

    ˉx=βk01θ+o(|βk01|). (4.41)

    Therefore,

    ˉχ=βk01θξek0+o(|βk01|)

    is the bifurcated singular point of (3.5).

    Obviously, ˉχ is a locally asymptotically stable singular point on δ>δ1, which implies the problem (3.5)(3.6) has a continuous transition in U2. Meanwhile, the original equilibrium state loses its stability and the problem (3.5)(3.6) has a jump transition in U1. And ˉχ is an unstable saddle point on δ<δ1 (see Figure 1). Thus, the Theorem is proved.

    Figure 1.  If Ωe3k0dx0, then the local topological structure of the transitions of (3.5) is : (1) when δ<δ1, the system bifurcates from stable equilibrium point χ=0 to an unstable saddle point ˉχ; (2) when δ>δ1, the system bifurcates from χ=0 to a attractor ˉχ.

    In the following, we consider the case that

    Ωe3k0dx=0. (4.42)

    Let

    ˜θ=~PkQkΩe2k0dx, (4.43)

    here

    ~Pk={[2ˉa21u0w0a2c2+ˉa1ˉa2αv0w0abcd+ˉa1w20a2bc+ˉa22γv0w0b2cd+w20ab2e]Ωϕ1e2k0dx+[ˉa21αu0w0a2c2+ˉa1ˉa2γu0w0abc2+2ˉa22v0w0b2cd+ˉa2κw20ab2c+qw20ab2e]Ωϕ2e2k0dx+[ˉa21u0w0a2c2+ˉa22κv0w0b2cd+ˉa1u0w0abce+qˉa2v0w0b2de]Ωϕ3e2k0dx}, (4.44)

    Qk is as in (4.33), ϕ=(ϕ1,ϕ2,ϕ3) satisfies

    Lϕ=G(ξ)e2k0, (4.45)

    and the operators L and G are defined by

    Lϕ={ε1Δuˉa1u0uˉa1αu0vˉa1u0w,ε2Δvˉa2γv0uˉa2v0vˉa2κv0w,ε3Δw+w0u+qw0v,
    G(ξ)={ˉa21u0w0ac(ˉa1u0w0ac+ˉa2αv0w0bd+w20ab),ˉa22v0w0bd(ˉa1γu0w0ac+ˉa2v0w0bd+κw20ab),w20ab(ˉa1u0w0ac+ˉa2qv0w0bd).

    Then we have the following theorem,

    Theorem 4.3. Assume that q(γκ)(1κα)<0. Let ˜θ0 be the number given by (4.43), and δ0>δ1. Then the transition of (3.5) at δ1 is continuous as ˜θ<0, and is jump as ˜θ>0. Moreover, we have the following assertions:

    (1) When ˜θ>0, the system (3.5) bifurcates from (χ,δ)=(0,δ1) to two steady-state solutions χ+ and χ on δ<δ1, which are saddles, and no bifurcation solutions on δ>δ1.

    (2) When ˜θ<0, the system (3.5) bifurcates from (χ,δ)=(0,δ1) to two steady-state solutions χ+ and χ on δ>δ1, which are attractors, and no bifurcation solutions on δ<δ1.

    (3) The bifurcated solutions χ± can be expressed as

    χ±=±[βk01˜θ]12ξek0+o(|βk01|12),

    where ξ is as in (4.35), βk01 as in Lemma 3.1.

    Proof. We shall prove the theorem in the following two steps.

    Step 1.We deduce the evolution equation.

    Space decomposition is the same as the step 1 in the proof of Theorem 4.2, so we omit it.

    Analogously, in the space E1, the Eq (3.5) can be reduced to

    ξek0,ξek0dxdt=Lδ(χ),ξek0+G(χ),ξek0=βk01ξek0,ξek0x+G(χ),ξek0. (4.46)

    Let

    χ=xξek0+Φ(x),   xR1, (4.47)

    and Φ(x) is the center manifold function. To evaluate the last term in (4.46), we need to know the center manifold function Φ:E1E2. Let Φ=x2ϕ=o(2), then by the approximation formula of center manifolds (see (A.10) in[27]), ϕ satisfies

    Lϕ=G(ξek0)=G(ξ)e2k0. (4.48)

    Direct circulation to get

    G(xξek0+Φ),ξek0=Ω[ˉa1(xξ1ek0+Φ1)2ξ1ek0ˉa1α(xξ1ek0+Φ1)(xξ2ek0+Φ2)ξ1ek0ˉa1(xξ1ek0+Φ1)(xξ3ek0+Φ3)ξ1ek0ˉa2γ(xξ1ek0+Φ1)(xξ2ek0+Φ2)ξ2ek0ˉa2(xξ2ek0+Φ2)2ξ2ek0ˉa2κ(xξ2ek0+Φ2)(xξ3ek0+Φ3)ξ2ek0+(xξ1ek0+Φ1)(xξ3ek0+Φ3)ξ3ek0+q(xξ2ek0+Φ2)(xξ3ek0+Φ3)ξ2ek0]dx=x3{[2ˉa1ξ1ξ1ˉa1αξ2ξ1ˉa1ξ3ξ1ˉa2γξ2ξ2+ξ3ξ3]Ωϕ1e2k0dx+[ˉa1αξ1ξ1ˉa2γξ1ξ22ˉa2ξ2ξ2ˉa2κξ3ξ2+qξ3ξ3]Ωϕ2e2k0dx+[ˉa1ξ1ξ1ˉa2κξ2ξ2+ξ1ξ3+qξ2ξ3]Ωϕ3e2k0dx}+o(3).

    Hence, we have

    G(xξek0+Φ),ξek0ξek0,ξek0=˜θx3+o(3), (4.49)

    where ˜θ is defined by (4.43).

    Combining (4.46) and (4.49), we deduce that the following reduced bifurcation equation:

    dxdt=βk01x+˜θx3+o(3). (4.50)

    Step 2. Bifurcation analysis.

    Obviously, when ˜θ>0, the Eq (4.50) bifurcates two saddle points on δ<δ1, and when ˜θ<0, the Eq (4.50) bifurcates two stable singular points on δ>δ1. The bifurcated solutions can be expressed as

    x±=±[βk01˜θ]12+o(|βk01|12).

    It is known that the transition and local topological structure of Eq (3.5) are determined completely by (4.50). Therefore,

    χ+=+[βk01˜θ]12ξek0+o(|βk01|12)

    and

    χ=[βk01˜θ]12ξek0+o(|βk01|12)

    are the bifurcated singular points of (3.5). The stability of χ+ and χ are the same as that of x±, see Figure 2. The Theorem is proved.

    Figure 2.  If Ωe3k0dx=0, then the local topological structure of (3.5) is: (1) when ˜θ>0, δ<δ1, the system bifurcates from an stable equilibrium point χ=0 to two saddle points χ+ and χ; (2)when ˜θ<0, δ>δ1, the system bifurcates from an equilibrium point χ=0 to two stable singular points χ+ and χ.

    In this work, we study the dynamical transition for a 3-component Lotka-Volterra model with diffusion from the perspective of dynamic transition recently developed by Ma and Wang [18,19,20]. By using the Principle of Exchange of Stabilities condition for 3-component Lotka-Volterra model, we note that the model produces stationary periodic in space patterns for δ<min{δ0,δ1}, where δ0 and δ1 are determined by parameters λ=(ε1,ε2,ε3,ˉa1,ˉa2,α,γ,κ,σ,q)R10+. However, when δ>min{δ0,δ1}, i.e., the ratio of the inter-specific competition rates and the intra-specific rates are greater than the specified value, the stability is broken. Then we obtained a few main characteristics are as follows.

    First, when δ0<δ1, the first eigenvalues are complex, and we show that the system undergoes a continuous or jump transition. In the small oscillation frequency limit, the transition is always continuous and the time periodic rolls are stable after the transition. That provide some sufficient conditions for the stable coexistence equilibrium and periodic solution.

    Second, when δ0>δ1, it was demonstrated that chaotic coexistence bifurcates from the periodic when Ωe3k0dx0. If Ωe3k0dx=0, we show that the permanent coexistence was existed for the two-prey-one-predator model with intra-specific competition for predator.

    Limei Li acknowledges the support from the National Science Foundation of China (NSFC) Grant No. 11701399. The authors are also grateful to the reviewer for the constructive comments.

    The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.



    [1] K. Kuto, T. Tsujikawa, Limiting structure of steady-states to the Lotka-Volterra competition model with large diffusion and advection, J. Differ. Equations, 258 (2015), 1801–1858. doi: 10.1016/j.jde.2014.11.016
    [2] J. C. Eilbeck, J. E. Furter, J. Lopezgomez, Coexistence in the competition model with diffusion, J. Differ. Equations, 107 (1994), 96–139. doi: 10.1006/jdeq.1994.1005
    [3] L. Lou, W. M. Ni, S. Yotsutani, On a limiting system in the Lotka-Volterra competition with cross-diffusion, Discrete Contin. Dyn. Syst., 10 (2004), 435–458.
    [4] Y. Wu, The instability of spiky steady states for a competing species model with cross diffusion, J. Differ. Equations, 213 (2005), 289–340. doi: 10.1016/j.jde.2004.08.015
    [5] W. M. Ni, Y. Wu, Q. Xu, The existence and stability of nontrivial steady states for S-K-T competition model with cross diffusion, Discrete Contin. Dyn. Syst., 34 (2014), 5271–5298. doi: 10.3934/dcds.2014.34.5271
    [6] L. Lou, W. M. Ni, S. Yotsutani, Pattern formation in a cross-diffusion system, Discrete Contin. Dyn. Syst., 35 (2015), 1589–1607. doi: 10.3934/dcds.2015.35.1589
    [7] W. M. Ni, M. Salomˊe, Y. Lou, On 3×3 Lotka-Volterra competition systems with cross-diffusion, Discrete Contin. Dyn. Syst., 6 (1999), 175–190.
    [8] P. Y. H. Pang, M. Wang, Strategy and stationary pattern in a three-species predator-prey model, J. Differ. Equations, 200 (2004), 245–273. doi: 10.1016/j.jde.2004.01.004
    [9] N. Ali, M. Haque, E. Venturino, S. Chakravarty, Dynamics of a three species ratio-dependent food chain model with intra-specific competition within the top predator, Comp. Biol. Med., 85 (2017), 63–74. doi: 10.1016/j.compbiomed.2017.04.007
    [10] T. Ikeda, M. Mimura, An interfacial approach to regional segregation of two competing species mediated by a predator, J. Math. Biol., 31 (1993), 215–240. doi: 10.1007/BF00166143
    [11] Y. Kan-on, M. Mimura, Singular perturbation approach to a 3-component reaction-diffusion system arising in population dynamics, SIAM J. Math. Anal., 29 (1998), 1519–1536. doi: 10.1137/S0036141097318328
    [12] Y. Kan-on, Existence and instability of Neumann layer solutions for a 3-component Lotka-Volterra model with diffusion, J. Math. Anal. Appl., 243 (2000), 357–372. doi: 10.1006/jmaa.1999.6676
    [13] M. X. Wang, Stationary patterns of strongly coupled prey-predator models, J. Math. Anal. Appl., 292 (2004), 484–505. doi: 10.1016/j.jmaa.2003.12.027
    [14] W. Chen, R. Peng, Stationary patterns created by cross-diffusion for the competitor-competitor-mutualist model, J. Math. Anal. Appl., 291 (2004), 550–564. doi: 10.1016/j.jmaa.2003.11.015
    [15] W. Ko, K. Ryu, I. Ahn, Coexistence of Three Competing Species with Non-negative Cross-diffusion rate, J. Dynam. Contr. Syst., 20 (2014), 229–240. doi: 10.1007/s10883-014-9219-6
    [16] K. Ryu, I. Ahn, Coexistence theorem of steady states for nonlinear self-cross diffusion systems with competitive dynamics, J. Math. Anal. Appl., 283 (2003), 46–65. doi: 10.1016/S0022-247X(03)00162-8
    [17] L. Li, Coexistence Theorems of Steady States for Predator-Prey Interacting Systems, Trans. Am. Math. Soc., 305 (1988), 143–166. doi: 10.1090/S0002-9947-1988-0920151-1
    [18] T. Ma, S. H. Wang, Bifurcation Theory and Applications, World Scientific, Singapore, 2005.
    [19] T. Ma, S. H. Wang, Stability and Bifurcation of Nonlinear Evolutions Equations, Science Press, Beijing, 2007.
    [20] T. Ma, S. H. Wang, Phase Transition Dynamics, New York: Springer-Verlag, 2014.
    [21] T. Ma, S. H. Wang, Phase transitions for Belousov-Zhabotinsky reactions, Math. Methods. Appl. Sci., 34 (2011), 1381–1397. doi: 10.1002/mma.1446
    [22] C. H. Hsia, T. Ma, S. H. Wang, Rotating Boussinesq equations: dynamic stability and transitions, Discrete Contin. Dyn. Syst., 28 (2010), 99–130. doi: 10.3934/dcds.2010.28.99
    [23] T. Ma, S. H. Wang, Dynamic transition and pattern formation for chemotactic systems, Discrete Contin. Dyn. Syst. Ser B., 19 (2014), 2809–2835.
    [24] Z. G. Pan, T. Sengul, Q. Wang, On the viscous instabilities and transitions of two-layer model with a layered topography, Commun. Nonlinear Sci. Numer. Simul., 80 (2020), 104978. doi: 10.1016/j.cnsns.2019.104978
    [25] C. H. Lu, Y. Mao, Q Wang, D. M. Yan, Hopf bifurcation and transition of three-dimensional wind-driven ocean circulation problem, J. Differ. Equations, 267 (2019), 2560–2593. doi: 10.1016/j.jde.2019.03.021
    [26] R. Liu, Q. Wang, S1 attractor bifurcation analysis for an electrically conducting fluid flow between two rotating cylinders, Phys. D., 392 (2019), 17–33. doi: 10.1016/j.physd.2019.03.001
    [27] T. Ma, S. H. Wang, Dynamic transition theory for thermohaline circulation, Phys. D. Nonlinear Phenomena, 239 (2009), 167–189.
  • This article has been cited by:

    1. Taylan Şengül, Burhan Tiryakioglu, Esmanur Yıldız Akıl, First transition dynamics of reaction–diffusion equations with higher order nonlinearity, 2024, 0022-2526, 10.1111/sapm.12735
  • Reader Comments
  • © 2021 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(2688) PDF downloads(246) Cited by(1)

Figures and Tables

Figures(2)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog