Abstract
This paper concerns with a reaction–diffusion system modeling the population dynamics of the predator and prey, in which the predator moves toward the gradient of concentration of some chemical released by prey instead of moving directly toward the higher density of prey. The first objective is to investigate the global existence and boundedness of the unique classical solution. Then we study the asymptotic stabilities of nonnegative spatially homogeneous equilibria. Moreover, the convergence rates are also studied.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In this paper, we consider the following predator–prey model with nonlinear “indirect prey-taxis”:
In this model, \(\Omega \) is a bounded domain in \(\mathbb {R}^n\) with smooth boundary \(\partial \Omega \), \(\partial _{\nu }=\frac{\partial }{\partial \nu }\) and \(\nu \) is the unit outward normal vector of \(\partial \Omega \). Functions u and v are, respectively, population densities of the predator and prey, and w is the concentration of chemoattractant released by the prey. Here \(d_1,d_2,d_3,b,\mu ,r\) are positive constants. The decay rate of the chemical w is \(\mu \), and the parameter r is the production rate. The term \(\chi (w)\) is the chemotactic sensitivity which depends only upon w. The term uh(u) describes the population kinetic of the predator u. Function g(v) is the functional response accounting for the intake rate of the predator as a function of prey density. And f(v) is the growth function of prey.
The system (1.1), which was recently proposed by Tello and Wrzosek [17], describes “indirect prey-taxis” in the sense that the predator moves following the gradient of some chemicals which indicate the presence of prey instead of moving directly toward the higher density of prey. The substance released by the prey, such as pheromones, chemical alarm cues, sexual signals, can be viewed as the chemoattractant for the foraging predator. The known example is that the wolf spider Pardosa milvina responses to chemical cues left by the prey [7]. For the detailed biological background, please refer to [17] and the references therein. For the special case \(b=d_3=0\), \(h(u)=0\) and ug(v) is replaced by \(vF_0(u)\), where \(F_0\) is positive, bounded, smooth function and satisfies
with positive constant \(F_m\), the global existence of solutions, linearized stability and asymptotic behavior of steady states in two dimensional case for (1.1) were established. It was proved in [17] that the positive constant steady state may be unstable if chemotactic sensitivity or the rate of release of the chemoattractant is big enough. However, to our best knowledge, no other results are available. Studies concerning the model (1.1) with general functional responses and nonlinear indirect prey-taxis are required.
In order to better understand the system (1.1), it is worth mentioning some studies for the prey-taxis system in which the movement of the predator is determined by the prey density gradient. In the spatial predator–prey interaction, in addition to the random diffusion of predator and prey, the predator has the tendency to move towards the area with higher density of prey population. Kareiva and Odell [10] first derived a prey-taxis model to describe the predator aggregation in high prey density areas. Since then, various reaction–diffusion models have been proposed to interpret the prey-taxis phenomenon [1, 4, 15]. The general predator–prey model with prey-taxis reads as follows
where the constant \(\chi _0>0\) and the term \(\chi _0\nabla \cdot (u\nabla v)\) describes the tendency of the predator moving towards the increasing prey gradient direction. This system has been studied by many authors. Lee et al. [13] studied the pattern formation of (1.2), they showed that prey-taxis in most cases tends to stabilize predator–prey interactions, which is an opposite result to the case of Keller-Segel chemotaxis system (the chemotaxis may lead to the formation of aggregates or inhomogeneous space patterns [3]). In [12], Lee et al. studied the continuous traveling waves for (1.2) and they showed that prey-taxis can reduce the likelihood of effective biocontrol. Wu et al. [27] investigated the global existence and boundedness of solutions of (1.2) under a smallness assumption on \(\chi _0\). Jin and Wang [9] proved the global boundedness of solution and stabilities of nonnegative spatially homogeneous equilibria of (1.2) in the two-dimensional case. Recently, It was shown in [24] that the prey-taxis destabilizes predator–prey homogeneity when prey repulsion is present (i.e. \(\chi _0<0\)). Moreover, the nonconstant positive steady states of a wide class of prey-taxis systems with general functional responses over 1-D domain were obtained in [24]. For more related works, we refer the readers to [5, 16, 18, 19, 25].
In the present paper, the initial data \(u_0,w_0,v_0\) are supposed to satisfy
And we suppose that \(\chi ,\,h,\,f\) and g satisfy the following hypotheses [9, 19, 27]:
-
(A1)
The function \(\chi \in C^2([0,\infty ))\), \(\chi \ge 0\). The well known examples are
$$\begin{aligned} \mathrm{(i)} \ \chi (s)=\chi _1, \ \ \mathrm{(ii)} \ \chi (s)=\frac{\chi _1}{s+\varepsilon }, \ \ \mathrm{(iii)} \ \chi (s)=\frac{\chi _1}{(s+\varepsilon )^2} \end{aligned}$$with positive constants \(\chi _1,\varepsilon \).
-
(A2)
The function \(g\in C^2([0,\infty )),\,g(0)=0,\,g(s)> 0\) in \((0,\infty )\). The typical examples are
$$\begin{aligned} (\mathrm{type\ I}) \ g(s)= & {} \gamma s,\ \ \ (\mathrm{type\ II})\ g(s)=\displaystyle \frac{\gamma s}{l+s},\\ (\mathrm{type\ III}) \ g(s)= & {} \displaystyle \frac{\gamma s^{\kappa }}{l^{\kappa }+s^{\kappa }}, \ \ \ (\mathrm{Ivlev\ type}) \ g(s)=\gamma (1-e^{-ls}), \end{aligned}$$where \(\gamma ,l,\kappa \) are positive constants and \(\kappa >1\).
-
(A3)
The function \(h\in C^2([0,\infty ))\) and there exist two constants \(a>0\) and \(\theta \ge 0\) such that \(h(s)\ge a\) and \(h'(s)\ge \theta \) in \([0,\infty )\). In some sense, the constant a can be regarded as the minimal death rate of the predator. The typical example is \(h(s)=a+\theta s\).
-
(A4)
The function \(f\in C^2([0,\infty ))\) satisfying \(f(0)=0\), and there exist two positive constants \(\eta ,K\) such that \(f(s)\le \eta s\) for \(s\ge 0\), \(f(K)=0\) and \(f(s)<0\) for \(s>K\). Some examples are
$$\begin{aligned} (\mathrm{logistic}) \ f(s)=\eta s\left( 1-\frac{s}{K}\right) ,\ \ (\mathrm{Allee\ effect}) \ f(s)=\eta 's\left( 1-\frac{s}{K}\right) \left( \frac{s}{N}-1\right) \end{aligned}$$with \(0<N<K\) and \(\eta '=\frac{4KN}{(K-N)^2}\eta \).
Throughout this paper we denote \(\Vert \cdot \Vert _p=\Vert \cdot \Vert _{L^p(\Omega )}\), and use C and \(C_i\) to denote the generic positive constants.
In contrast to the prey-taxis system (1.2), the model (1.1) involves chemoattractant which is released by the prey and attracts the predator. A natural question is: Does the chemoattractant affect the dynamical properties of the predator and prey? Our conclusions show that, in “most situations”, the chemoattractant does not affect the dynamical properties of the predator and prey.
The first result of this paper asserts that the solution of the prey-taxis system (1.1) exists globally and maintains bounded. This property is the same as that of the classical problem of predator–prey model without prey-taxis:
Theorem 1.1
Let \(n\ge 1\) and the hypotheses (A1)–(A4) hold. Then (1.1) has a unique nonnegative and bounded global solution (u, w, v), and
Remark 1.1
We note that the solution of (1.2) exists globally in two-dimensional case ([9, Theorem 1.1]). In the higher dimensional case (\(n\ge 3\)), if \(\chi _0\) is small and \(g(v)\le c\) for some \(c>0\), then (1.2) admits a unique nonnegative global bounded solution ([27, Theorem 1.1]). It remains unknown whether or not the solution of (1.2) blows up in higher dimensional case when \(\chi _0\) is large. However, for the system (1.1), Theorem 1.1 claims the global existence and boundedness of solution of (1.1). This also shows that, compared to the prey-taxis, the indirect prey-taxis will prevent the growth of the predator to ensure the global existence and boundedness of the solution.
The second goal of this paper is to understand the role of the indirect prey-taxis in the global stabilities of nonnegative spatially homogeneous equilibria of (1.1). The global stability of the prey-taxis system (1.2) has been studied in [9]. Therefore, we are able to compare the stability results of (1.1) with that of (1.2).
Let \(\varphi (v)=f(v)/g(v)\). In order to achieve our aim, we shall need other assumptions [9]:
- (A2)\('\):
-
Function \(g\in C^2([0,\infty )),\,g(0)=0,\,g(s)> 0\) in \((0,\infty )\), and \(g'(s)>0\) in \([0,\infty )\).
- (A5):
-
Function \(\varphi \in C^1([0,\infty ))\), \(\varphi (0)>0\) and \(\varphi '(s)<0\) in \([0,\infty )\).
Remark 1.2
The Holling type I, type II and Ivlev type response functions satisfy the condition (A2)\('\). Moreover, if f is of logistic type and g is of Holling type I or type II with \(l>K\), then (A5) is fulfilled. We should mention that (A5) can not be satisfied by the bistable function f(v) or the Holling type III response function g(v) (see [9]).
Let us first note that the possible homogeneous steady states of the system (1.1) are given by
where the positive constants \(u_*,\,v_*\) are determined by
It is easy to deduce that, if \(g,\,h\) and f take biological meaningful forms like some of those given in (A2)\('\), (A3)–(A5), then \((u_*,v_*)\) is uniquely determined and can be explicitly found. Hence, in what follows, we shall suppose that (1.4) has a unique positive solution \((u_*,v_*)\). Moreover, if f and g satisfy the assumptions (A2)\('\) and (A4), then by the second equation of (1.4) we have \(v_*<K\), and hence \(m=\max \{\Vert v_0\Vert _\infty ,\ K\}>v_*\).
In the case of \(bg(K)>a\), we shall show that if the chemotactic coefficient \(\chi (w)\) is small or one of the diffusion coefficients of the predator and chemical is large then the positive spatially homogeneous equilibrium \((u_*,rv_*/\mu ,v_*)\) is globally asymptotically stable.
Theorem 1.2
Assume \(bg(K)>a\) and the hypotheses (A1), (A2)\('\),(A3)–(A5) are satisfied. Let (u, w, v) be the unique bounded global solution of (1.1), which is given by Theorem 1.1. Set
and
If
then \((u_*,rv_*/\mu ,v_*)\) is globally asymptotically stable. Furthermore, if we further assume \(\theta >0\), then \((u_*,rv_*/\mu ,v_*)\) is exponentially stable, i.e., there exist constants \(C,\,\lambda >0\) such that
In the case of \(bg(K)\le a\), the following theorem asserts that the semi-trivial spatially homogeneous equilibrium \((0,\,{rK}/{\mu },\,K)\) is globally asymptotically stable.
Theorem 1.3
Let the hypotheses (A1), (A2)\('\),(A3)–(A5) be satisfied and (u, w, v) be the unique bounded global solution of (1.1), which is given by Theorem 1.1.
-
(i)
If \(bg(K)<a\), then \((0,\,{rK}/{\mu },\,K)\) is globally asymptotically stable with exponential rate, i.e., there exist constants \(C,\,\lambda >0\) such that
$$\begin{aligned} \Vert u\Vert _\infty +\Vert w-{rK}/{\mu }\Vert _\infty +\Vert v-K\Vert _\infty \le C e^{-\lambda t},\ \ \forall \ t>0. \end{aligned}$$(1.7) -
(ii)
If \(bg(K)=a\), then \((0,\,{rK}/{\mu },\,K)\) is globally asymptotically stable. Furthermore, if \(\theta >0\), then \((0,\,{rK}/{\mu },\,K)\) is algebraically stable, i.e., there exist constants \(C,\,\lambda >0\) such that
$$\begin{aligned} \Vert u\Vert _\infty +\Vert w-rK/\mu \Vert _\infty +\Vert v-K\Vert _\infty \le C(t+1)^{-\lambda },\ \ \forall \ t>0. \end{aligned}$$(1.8)
In the conditions (A3) and (A4), constants a and K can be considered as the minimal death rate of predator and carrying capacity of prey, respectively. Hence, the maximal value of the predation is g(K). The cases \(g(K)>a/b\) and \(g(K)\le a/b\) can be regarded as the strong and weak predation, respectively.
In the strong predation case (\(g(K)>a/b\)), under our assumptions, the problem (1.1) has a unique positive constant steady state \((u_*,\,{rv_*}/{\mu },\,v_*)\) and it is globally asymptotically stable. Furthermore, if \(\theta >0\), then \((u_*,\,{rv_*}/{\mu },\,v_*)\) is also exponentially stable (Theorem 1.2).
Noticing that the condition (1.5) involves the coefficients \(d_2\), \(\mu ,\,r\). Hence, the chemoattractant plays an important role in the stability of \((u_*,\,rv_*/\mu ,\,v_*)\). It is observed that the value of \(k_2\) also affects the stability of \((u_*,\,rv_*/\mu ,\,v_*)\). Moreover, from the condition (1.5) we discover that the diffusion rate of the prey does not influence the long time behavior of solution of (1.1). Since the predator responses to the chemoattractant released by prey rather than the prey itself, the diffusion of prey may be negligible in this case.
In the weak predation case (\(g(K)\le a/b\)), the problem (1.1) has no positive constant steady state and the semi-trivial constant steady state \((0,\,{rK}/{\mu },\,K)\) is globally asymptotically stable (Theorem 1.3). This shows that, in the weak predation case, the presence of the chemoattractant does not influence the steady states and their stabilities for the problem (1.1). In contrast to the prey-taxis system (1.2) in such a case, please refer to [9, Theorem 1.3 (1)].
For the asymptotic behavior of solution, in contrast to the classical predator–prey model (1.3), we have the following assertions:
-
(i)
in the weak predation case, the asymptotic dynamical properties of (1.1) are the same as those of (1.3).
-
(ii)
in the strong predation case, under the assumption (1.5), the asymptotic dynamical properties of (1.1) are the same as those of (1.3).
The proofs of Theorems 1.2 and 1.3 rely on two Lyapunov functionals. The constructions of these Lyapunov functionals are inspired by [9]. However, the arguments leading to Theorems 1.2 and 1.3 are different from that of [9] which are based on LaSalle’s invariant principle. Our method depends on an important lemma (see Lemma 3.1) and some basic arguments which seems friendlier to the readers.
The methods in the proofs of Theorems 1.2 and 1.3 can be applied to the model (1.2). The case \(bg(K)=a\) and \(\theta =0\) was not considered in [9] for the problem (1.2). Using the method in the proof of Theorem 1.3 (ii), we can show that the semi-trivial spatially homogeneous equilibrium \((0,\,K)\) is globally asymptotically stable for the problem (1.2) in this case.
The article is organized as follows. Section 2 provides the uniqueness, global existence and boundedness of the classical solution of (1.1). Section 3 is devoted to proving the global stability results in Theorems 1.2 and 1.3. In the last section, we present two examples.
2 Existence, Uniqueness, Boundedness and Uniform Estimates of Global Solution
2.1 Existence and Uniqueness of Local Solution, Some Preliminaries
We first give a claim concerning the local-in-time existence of classical solution to (1.1).
Lemma 2.1
There exists a \({\hat{T}}\in (0,\infty ]\) and a unique nonnegative solution (u, w, v) of (1.1) defined in \([0,{\hat{T}})\) and satisfies
and
Moreover, the “existence time \({\hat{T}}\)” can be chosen maximal: either \({\hat{T}}=\infty \), or \({\hat{T}}<\infty \) and
Proof
The local-in-time existence and uniqueness of classical solution to the problem (1.1) follow from Amann’s theorem [2, Theorem 7.3 and Corollary 9.3] (cf. [27, Lemma 2.1]). The estimates (2.1) can be derived by the maximum principle.
Lemma 2.2
The solution component w of (1.1) satisfies
where \(M=\max \left\{ \Vert w_0\Vert _\infty ,\ {rm}/{\mu }\right\} \). And for any \(p\in [2,\infty )\), there is \(K_p=K(p)>0\) such that
Moreover, there exists a positive constant C such that the solution component u of (1.1) satisfies
Proof
By using (2.1) and the maximum principle, one can deduce from the w-equation in (1.1) that
In view of the variation-of-constants formula, it yields
Making use of (2.1) and the well-known semigroup estimates [6, 8, 26] we have that, for some \(\lambda _1,\,C_i>0,\, i=1,...,5\),
This implies (2.3).
We next prove (2.4). It follows from the first and third equation in (1.1) that
Let \(N_0=\sup _{z\in [0,m]}|f(z)|\). Recall the assumption (A3) and the estimate for v in (2.1), it yields
where \(C_6=(bN_0+m)|\Omega |\). Applying the Gronwall’s inequality to (2.5) we have (2.4). \(\square \)
Next we provide a lemma which claims that the global existence and \(L^\infty \)-boundedness of u can be reduced to proving its \(L^p\)-boundedness for \(p>n/2\) and \(p\ge 1\).
Lemma 2.3
Let \(n\ge 1\) and (u, w, v) be the unique solution of (1.1) in \(\Omega \times (0,{\hat{T}})\). Suppose that there exists a number \(p\ge 1\) and \(p>n/2\) for which
Then \({\hat{T}}=\infty \) and
Proof
The estimate (2.2) implies
Note that \((bug(v)-uh(u))_+\le bug(v)\) and
where \(N=\sup _{z\in [0,m]}g(z)\). Thanks to (2.1), (2.3) and (2.4), similar to the proof of [9, Lemma 3.1] (see also [3, Lemma 3.2]), one can deduce that \({\hat{T}}=\infty \) and (2.7) holds. \(\square \)
2.2 Proof of Theorem 1.1
Let \(n\ge 2\) and \(p>n/2\). Clearly, \(p>1\). Note that
Hence, we can choose \(q>p\) such that
Let
Multiplying the first equation of (1.1) by \(u^{p-1}\) and integrating the results over \(\Omega \), we obtain
where we have used Young’s inequality, (2.1) and (2.2) and the assumption (A3). By use of Young’s inequality again and (2.3), it yields
with some \(C_1>0\), and there is \(C_2>0\) such that
Inserting (2.10) and (2.11) into (2.9) gives
for all \(t\in (0,{\hat{T}})\), where \(C_3=C_1+C_2\). Note that (2.8). Taking advantage of the Gagliardo-Nirenberg inequality and (2.4) firstly, and using the Young’s inequality secondly, we have
Combined (2.13) with (2.12) allows us to deduce
Thus we have, by the Gronwall inequality,
Using (2.4) and Lemma 2.3 with \(p=1\) when \(n=1\), and using (2.14) and Lemma 2.3 when \(n\ge 2\), we can get the conclusion of Theorem 1.1 immediately.
2.3 Uniform Estimates of the Global Solution
Theorem 2.1
Let (u, w, v) be the unique global bounded classical solution of (1.1), which is given by Theorem 1.1. Then for any given \(0<\alpha <1\), there exists \(C(\alpha )>0\) such that
Proof
This proof is based on the standard parabolic regularity for parabolic equations (cf. [20, Theorem 2.1], [21, Theorem 2.1] and [23, Theorem 2.2]). For the reader’s convenience, we sketch the proof here. Applying the interior \(L^p\) estimate ([14, Theorems 7.30 and 7.35]) to the equations of w and v firstly and using the Sobolev embedding theorem secondly we have
and hence
Note that w satisfies
By use of the interior Schauder estimate [11] and (2.16),
which implies
Rewrite the equation of u in (1.1) as
where
Due to (2.16), (2.17) and the boundedness of (u, w, v), we see that \(\Vert G\Vert _{L^\infty (\Omega \times [i+\frac{1}{3},i+3])}\le C_5\) and \(\Vert \chi (w)\nabla w\Vert _{L^\infty (\Omega \times [i+\frac{1}{3},i+3])}\le C_5\) for all \(i\ge 0\). Applying the interior \(L^p\) estimate to (2.18) we have \(\Vert u\Vert _{W^{2,1}_p(\Omega \times [i+\frac{1}{2},i+3])}\le C_6\) for all \(i\ge 0\). Then the embedding theorem gives
It then follows that
This combined with (2.17) yields
Applying the interior Schauder estimate to (2.18) we have \(\Vert u\Vert _{C^{2+\alpha ,1+\frac{\alpha }{2}}({\bar{\Omega }}\times [i+1,i+3])}\le C_{10}\) for all \(i\ge 0\). Thus,
Similarly, thanks to (2.16) and (2.19), we can apply the interior Schauder estimate to the equation of v and get
Then (2.15) follows from (2.17), (2.20) and (2.21). The proof is complete. \(\square \)
3 Global Stability
Throughout this section we always assume that (u, w, v) is a bounded global solution of (1.1). We shall prove Theorems 1.2 and 1.3 by constructing suitable Lyapunov functionals. Let us first recall two basic results.
Lemma 3.1
([22, Lemma 1.1]) Let \(\tau \ge 0\), \(c>0\) be constants, \(\psi (t)\ge 0,\,\int _\tau ^\infty \rho (t)\mathrm{d}t<\infty \). Assume that \(\varphi \in C^1([\tau ,\infty )),\,\varphi \) is bounded from below and satisfies
If either \(\psi \in C^1([\tau ,\infty ))\) and \(\psi '(t)\le k\) in \([\tau ,\infty )\) for some constant \(k>0\), or \(\psi \in C^\alpha ([\tau ,\infty ))\) and \(\Vert \psi \Vert _{C^\alpha ([\tau ,\infty ))}\le k\) for some constants \(0<\alpha <1\) and \(k>0\), then \(\lim _{t\rightarrow \infty }\psi (t)=0\).
Lemma 3.2
([9, Lemma 4.1]) Let g satisfy the conditions in (A2)\('\) and (u, w, v) be a solution of (1.1). Define
for the constant \(k>0\). Then \(\zeta \) is a convex function and \(\zeta \ge 0\) on \([0,\infty )\). Furthermore, if \(v\rightarrow k\) as \(t\rightarrow \infty \), then there exists a constant \(T_0>0\) such that for all \(t\ge T_0\) there holds
3.1 Global Stability of \((u_*,\,rv_*/\mu ,\,v_*)\): Proof of Theorem 1.2
In this subsection we always assume that \(bg(K)>a\) and (1.5) holds. The constant K is given in the assumption (A4), and \((u_*,v_*)\) is given by (1.4). For the convenience, let \(w_*=rv_*/\mu \). Due to (1.5), we fix a constant \(\delta \) such that
where \({{\hat{\chi }}},k_1,k_2\) are given by Theorem 1.2.
Lemma 3.3
Let \(\delta \) be given by (3.2). Let the conditions in Theorem 1.2 hold. Then there is \(\varepsilon >0\) such that functions \(E_1(t)\), \(F_1(t)\) defined by
satisfy
Proof
For the convenience, we set
Evidently, \(A_1(t),B_1(t),D_1(t)\ge 0\). Let \({{\hat{\chi }}}=\sup _{z\in [0,M]}\chi (z)\), where M is given by (2.2). Since (u, w, v) is the global bounded solution to (1.1), there is \(c>0\) such that
Let us recall from the assumption (A3) that \(h'(s)\ge \theta \) for \(s\in [0,\infty )\). The straightforward calculation gives
and
as well as
Thus we have
where
We first deal with \(I_1(t)\). An application of the Young inequality yields
Consequently,
It follows from (3.2) that \(\varepsilon _0>0\), and hence by (3.4),
We next handle \(I_2(t)\). It follows from the second equation of (1.4) that \(u_*=f(v_*)/g(v_*)\). Thanks to the definitions of \(k_1,\ k_2\), the last term in the right hand side of (3.6) can be estimated as
where \(\xi _1\), \(\xi _2\) are between v and \(v_*\), and \(k_1,k_2\) come from Theorem 1.2. Insert (3.8) into (3.6) yields
Note that \(bk_1k_2>\delta r^2/(4\mu )\) by (3.2), we can choose \(\varepsilon _2>0\) small such that
Again, by Young’s inequality, there holds
This combined with (3.9) allows us to derive
Finally, according to (3.5), (3.7) and (3.10), by choosing \(\varepsilon =\min \{\varepsilon _1,\,\varepsilon _2,\varepsilon _3\}\) we then get (3.3). \(\square \)
Lemma 3.4
Under the conditions of Theorem 1.2, for any \(0<\alpha <1\), the following holds:
Proof
Let \(E_1(t)\), \(F_1(t)\) be given Lemma 3.3. Clearly, \(E_1(t)\ge 0\) as \(g'(s)>0\) in \([0,\infty )\). Thanks to (2.15), it is easy to see that \(F_1(t)\in C^{\alpha /2}([1,\infty ))\) and \(\Vert F_1\Vert _{C^{\alpha /2}([1,\infty ))}\le k\) in \([1,\infty )\) for some constant \(k>0\). Recall (3.3), we can apply Lemma 3.1 to deduce \(\displaystyle \lim _{t\rightarrow \infty }F_1(t)=0\). That is,
and
Take \(0<\alpha<\alpha '<1\). According to Theorem 2.1, in the space \(C^{2+\alpha '}({\bar{\Omega }})\), \(u(\cdot ,t)\), \(w(\cdot ,t)\) and \(v(\cdot ,t)\) are bounded for \(t\ge 1\). Using the compact arguments and uniqueness of limits we can show that (3.11) holds when \(\theta >0\), and
when \(\theta =0\).
In the following we consider the case \(\theta =0\). Define \(\bar{f}=\frac{1}{|\Omega |}\int _\Omega f\mathrm{d}x\) for \(f\in L^1(\Omega )\). It follows from the third equation of (1.1) that
It follows from (3.12) that \(\displaystyle \lim _{t\rightarrow \infty }[J_1(t)+J_2(t)]=0\). Recall (2.15), we have \(\Vert {\bar{v}}'\Vert _{C^{\alpha /2}([1,\infty ))}\le k\) for some positive constant k. This combined with (3.12) yields \({\bar{v}}'(t)\rightarrow 0\) as \(t\rightarrow \infty \). Therefore, in view of (3.13), there holds \(J_3(t)\rightarrow 0\) as \(t\rightarrow \infty \), i.e.,
Making use of the Poincaré inequality \(\Vert u-{\bar{u}}\Vert _2\le C\Vert \nabla u\Vert _2\) with \(C>0\), we have \(\Vert u-{\bar{u}}\Vert _2\rightarrow 0\) as \(t\rightarrow \infty \). This combined with (3.14) implies
Similar to the above we can prove (3.11). This completes the proof. \(\square \)
Proof of (1.6)
Let \(\theta >0\). For the given positive constant \(y_*\), we define \(h(y)=y-y_*\ln y\) for \(y>0\). By L’Hôpital’s rule, one can easily check that
Remember the limit (3.11) and (3.1), it follows that there is \(t_0>1\) such that
for all \(t>t_0\). Recall the definitions of \(E_1(t)\) and \(F_1(t)\), it follows from the right inequalities in (3.15)–(3.16) that \(E_1(t)\le C_1F_1(t)\) for all \(t>t_0\) and some \(C_1>0\). Inserting this into (3.3) we get
Thus, \(E_1(t)\le C_2 e^{-\sigma t}\) for \(t>t_0\) and some \(C_2,\,\sigma >0\). In view of the left inequalities in (3.15)–(3.16), there exist \(C_3,\,C_4>0\) such that
Recall that \(u(\cdot ,t)\), \(w(\cdot ,t)\) and \(v(\cdot ,t)\) are bounded in \(W^{1,\infty }(\Omega )\) for \(t>1\). Thanks to the Gagliardo-Nirenberg inequality (with \(C_{gn}>0\))
we can find \(C,\,\lambda >0\) such that
Thus (1.6) holds, and the proof is complete. \(\square \)
3.2 Global Stability of \((0,\,{rK}/{\mu },\,K)\): Proof of Theorem 1.3
Throughout this subsection we always assume that \(bg(K)\le a\). For the convenience, we denote \({\hat{K}}=rK/\mu \).
Lemma 3.5
Assume that \(bg(K)\le a\). Let \(k_1,k_2\) be as in Theorem 1.2 and \(0<\delta _1<(2\mu bk_1k_2)/(r^2)\). Then functions \(E_2(t)\), \(F_2(t)\) defined by
satisfy
where \(\varepsilon _4=\min \{{\delta _1\mu }/2,\ bk_1k_2-{\delta _1r^2}/(2\mu )\}>0\).
Proof
In view of the assumption (A3), we have \(h(u)\ge a+\theta u\), which implies that
Similar to the proof of Lemma 3.3, by a series of calculations we can get
and
Hence, there holds
where
with \(\varphi (v)=f(v)/g(v)\) and \(\varphi (K)=f(K)/g(K)\). The last term in the right hand side of (3.20) can be estimated as
where \(\xi _3\) and \(\xi _4\) are between v and K, and \(k_1,k_2\) come from Theorem 1.2. Inserting (3.21) into (3.20) and applying the Young’s inequality to derive that
where \(\varepsilon _4=\min \{{\delta _1\mu }/2,\ bk_1k_2-{\delta _1r^2}/(2\mu )\}>0\). This combined with (3.19) gives (3.18). \(\square \)
Proof of Theorem 1.3 (i)
Assume that \(bg(K)<a\). Let \(E_2(t)\) and \(F_2(t)\) be given in Lemma 3.5, then \(E_2'(t)\le -F_2(t)\). Clearly, \(F_2(t)\ge 0\). Similar to the arguments in the proof of Lemma 3.4, one can deduce that, for any \(0<\alpha <1\),
\(\square \)
According to (3.22) and Lemma 3.2, there exists \(t_0>1\) such that
In view of the definitions of \(E_2(t),F_2(t)\) and the right inequality in (3.23), we get
It follows that
This implies that there exist \(C_2,\sigma >0\) such that \(E_2(t)\le C_2e^{-\sigma t}\) for \(t>t_0\). By the left inequality in (3.23) we have
In light of (3.22), in the space \(W^{1,\infty }(\Omega )\), \(u(\cdot ,t)\), \(w(\cdot ,t)\) and \(v(\cdot ,t)\) are bounded for \(t>1\). Making use of the Gagliardo–Nirenberg inequality
and (3.24), we have
Similarly, it follows from the Gagliardo–Nirenberg inequality (3.17) and (3.24) that
Thanks to (3.25)–(3.27), the statement in Theorem 1.3 (i) is followed immediately.
Proof of Theorem 1.3 (ii)
We first consider the case \(bg(K)=a\) and \(\theta >0\). Let \(E_2(t)\) and \(F_2(t)\) be given in Lemma 3.5, then \(E_2'(t)\le -F_2(t)\). Clearly, \(F_2(t)\ge 0\). In the present situation,
Similarly to the above, we can show that (3.22) holds. Let \(t_0>1\) be as in the proof of Theorem 1.3 (i). Using (3.23), the Cauchy-Schwarz inequality and boundedness of (u, w, v) we can find \(C_9>0\) such that
\(\square \)
This combined with \(E_2'(t)\le -F_2(t)\) leads us to \(E_2'(t)\le -C_{10}E_2^2(t)\) for \(t>t_0\). Thus, \(E_2(t)\le \frac{C_{11}}{t+1}\) for \(t>t_0\). Recall the definition of \(E_2(t)\) and the left inequality in (3.23), we can have
By the similar arguments in the proof of Theorem 1.3 (i), there exist \(C>0\) and \(\lambda >0\) such that
This implies (1.8).
Now we consider the case \(bg(K)=a\) and \(\theta =0\). In this case,
Similarly to the above it can be shown that
Integrating the equation of v in (1.1) we have
Noticing \(f(K)=0\), the limit (3.28) implies \(\int _\Omega f(v)\mathrm{d}x+\int _\Omega u(g(K)-g(v))\mathrm{d}x\rightarrow 0\) as \(t\rightarrow \infty \). We have known \(\lim \limits _{t\rightarrow \infty }\frac{d}{dt}\int _\Omega v\mathrm{d}x=0\) (see the proof of Lemma 3.4). It follows from (3.29) that \(\lim \limits _{t\rightarrow \infty }\Vert u\Vert _1=0\). Similarly to the above (compact arguments and uniqueness of limits), we can show (3.22), which implies the globally asymptotically stability of \((0,\,{rK}/{\mu },\,K)\). Theorem 1.3 (ii) is proved.
4 Two Examples
To better understand our stability results, we shall use Theorems 1.2 and 1.3 to study two examples which are of biologically meaningful.
Let us first consider the Lotka-Volterra predator–prey system with indirect prey-taxis, i.e.,
where the constants \(\chi _0,\,a,\,\eta ,\,K>0\) and \(\theta \ge 0\). Note that \(g'(v)=1,\,g(K)=K\) and
It is easy to see that (A5) is satisfied, and if \(bK>a\) then the positive constant steady state reads
According to Theorems 1.2 and 1.3, we have
-
If \(bK>a\) and
$$\begin{aligned} \frac{\chi _0^2}{d_1d_2}<\frac{16\mu b(bK+\theta \eta )}{r^2K(bK-a)}, \end{aligned}$$then (1.1) admits a unique positive steady state \((u_*,rv_*/\mu ,v_*)\).
-
If \(bK\le a\), then the steady state \((0,\,rK/\mu ,\,K)\) is globally asymptotically stable. This implies that the problem (1.1) has no positive steady state.
We next study the Rosenzweig-MacArthur predator–prey system with indirect prey-taxis, i.e.,
where the constants \(\chi _0,\,a,\,L,\,\eta ,\,K>0\) and \(L>K\). Note that \(g'(v)=L/(L+v)^2,\,g(K)=K/(L+K)\) and
Then,
where \(m=\max \{\Vert v_0\Vert _\infty ,\,K\}\). It is easy to see that (A5) is satisfied and if \(bK/(L+K)>a\), then
Thanks to Theorems 1.2 and 1.3, we have
-
If \(bK/(L+K)>a\), \(L>K\) and
$$\begin{aligned} \frac{\chi _0^2}{d_1d_2}<\frac{16\mu (L-K)(b-a)^2}{r^2(L+m)^2[(b-a)K-aL]}, \end{aligned}$$then (1.1) admits a unique positive steady state \((u_*,\,rv_*/\mu ,\,v_*)\).
-
If \(bK/(L+K)\le a\), then the steady state \((0,\,rK/\mu ,\,K)\) is globally asymptotically stable. This shows that the problem (1.1) has no positive steady state.
References
Ainseba, B.E., Bendahmane, M., Noussair, A.: A reaction–diffusion system modeling predator-prey with prey-taxis. Nonlinear Anal. Real World Appl. 9, 2086–2105 (2008)
Amann, H.: Dynamic theory of quasilinear parabolic equations. II. Reaction–diffusion systems. Differ. Integral Equ. 3(1), 13–75 (1990)
Bellomo, N., Bellouquid, A., Tao, Y., Winkler, M.: Toward a mathematical theory of Keller–Segel models of pattern formation in biological tissues. Math. Models Methods Appl. Sci. 25, 1663–1763 (2015)
Grünbaum, D.: Advection–diffusion equations for generalized tactic searching behaviours. J. Math. Biol. 38(2), 169–194 (1999)
He, X., Zheng, S.: Global boundedness of solutions in a reaction–diffusion system of predator–prey model with prey-taxis. Appl. Math. Lett. 49, 73–77 (2015)
Henry, D.: Geometric Theory of Semilinear Parabolic Equations. Springer, Berlin (1981)
Hoefler, C.D., Taylor, M., Jakob, E.M.: Chemosensory response to prey in Phidippus audax and Pardosa milvina. J. Arachnol. 30, 155–158 (2002)
Horstmann, D., Winkler, M.: Boundedness vs. blow-up in a chemotaxis system. J. Differ. Equ. 215(1), 52–107 (2005)
Jin, H.Y., Wang, Z.A.: Global stability of prey-taxis systems. J. Differ. Equ. 262, 1257–1290 (2017)
Kareiva, P., Odell, G.: Swarms of predators exhibit “preytaxis” if individual predators use area-restricted search. Am. Nat. 130, 233–270 (1987)
Ladyzenskaja, O.A., Solonnikov, V.A., Ural’ceva, N.N.: Linear and Quasi-linear Equations of Parabolic Type. Academic Press, New York (1968)
Lee, J.M., Hillen, T., Lewis, M.A.: Continuous traveling waves for prey-taxis. Bull. Math. Biol. 70, 654–676 (2008)
Lee, J.M., Hillen, T., Lewis, M.A.: Pattern formation in prey-taxis systems. J. Biol. Dyn. 3(6), 551–573 (2009)
Lieberman, G.M.: Second Order Parabolic Differential Equations. World Scientific Publishing Co., Inc., River Edge (1996)
Sapoukhina, N., Tyutyunov, Y., Arditi, R.: The role of prey taxis in biological control: a spatial theoretical model. Am. Nat. 162, 61–76 (2003)
Tao, Y.S.: Global existence of classical solutions to a predator–prey model with nonlinear prey-taxis. Nonlinear Anal. Real World Appl. 11(3), 2056–2064 (2010)
Tello, J.I., Wrzosek, D.: Predator–prey model with diffusion and indirect prey-taxis. Math. Models Methods Appl. Sci. 26(11), 2129–2162 (2016)
Wang, J.P., Wang, M.X.: The diffusive Beddington-DeAngelis predator–prey model with nonlinear prey-taxis and free boundary. Math. Methods Appl. Sci. 41, 6741–6762 (2018)
Wang, J.P., Wang, M.X.: Boundedness and global stability of the two-predator and one-prey models with nonlinear prey-taxis. Z. Angew. Math. Phys. 69(3), 63(1-24) (2018)
Wang, M.X.: The diffusive logistic equation with a free boundary and sign-changing coefficient. J. Differ. Equ. 258, 1252–1266 (2015)
Wang, M.X.: A diffusive logistic equation with a free boundary and sign-changing coefficient in time-periodic environment. J. Funct. Anal. 270(2), 483–508 (2016)
Wang, M.X.: Note on the Lyapunov functional method. Appl. Math. Lett. 75, 102–107 (2018)
Wang, M.X., Zhang, Y.: Dynamics for a diffusive prey–predator model with different free boundaries. J. Differ. Equ. 264, 3527–3558 (2018)
Wang, Q., Song, Y., Shao, L.J.: Nonconstant positive steady states and pattern formation of 1D prey-taxis systems. J. Nonlinear Sci. 27, 71–97 (2017)
Wang, X.L., Wang, W.D., Zhang, G.H.: Global bifurcation of solutions for a predator–prey model with prey-taxis. Math. Methods Appl. Sci. 38, 431–443 (2015)
Winkler, M.: Aggregation vs. global diffusive behavior in the higher-dimensional Keller–Segel model. J. Differ. Equ. 248(12), 2889–2905 (2010)
Wu, S.N., Shi, J.P., Wu, B.Y.: Global existence of solutions and uniform persistence of a diffusive predator–prey model with prey-taxis. J. Differ. Equ. 260, 5847–5874 (2016)
Acknowledgements
The authors would like to thank the anonymous referees for their helpful comments and suggestions.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by NSFC Grant 11771110.
Rights and permissions
About this article
Cite this article
Wang, J., Wang, M. The Dynamics of a Predator–Prey Model with Diffusion and Indirect Prey-Taxis. J Dyn Diff Equat 32, 1291–1310 (2020). https://doi.org/10.1007/s10884-019-09778-7
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10884-019-09778-7
Keywords
- Diffusive predator–prey model
- Indirect prey-taxis
- Global existence and boundedness
- Global stability and convergence rate