242 Nonlinear Analysis: Modelling and Control, 2011, Vol. 16, No.

2, 242–253

Global dynamics of a predator-prey system

with Holling type II functional response∗
Xiaohong Tian, Rui Xu
Institute of Applied Mathematics
Shijiazhuang Mechanical Engineering College
Shijiazhuang 050003, P R China
[email protected]; [email protected]

Received: 25 November 2010 / Revised: 15 March 2011 / Published online: 30 May 2011

Abstract. In this paper, a predator-prey system with Holling type II functional response and
stage structure is investigated. By analyzing the corresponding characteristic equations, the local
stability of each of feasible equilibria of the system is studied. The existence of the orbitally
asymptotically stable periodic solution is established. By using suitable Lyapunov functions and
the LaSalle invariance principle, it is proven that the predator-extinction equilibrium is globally
asymptotically stable when the coexistence equilibrium is not feasible, and sufficient conditions are
derived for the global stability of the coexistence equilibrium.
Keywords: Holling type II functional response, stage structure, periodic solution, LaSalle invarian-
ce principle, global stability.

1 Introduction
Mathematical modeling and computer simulation provide an effective tool in the study of
contemporary population ecology [1, 2]. In population dynamics, the functional response
of predator to prey density refers to the change in the density of prey attacked per unit
time per predator as the prey density changes [3]. In [4], based on experiment, Holling
suggested three kinds of functional response for different species to model the phenomena
of predation, it seems more reasonable than the standard Lotka–Voltera type predator-prey
system. In [5], Bazykin proposed the following predator-prey system:
 
bv(t)
u̇(t) = u(t) a − εu(t) − ,
1 + αu(t)
  (1)
du(t)
v̇(t) = v(t) −c + − ηv(t) ,
1 + αu(t)
∗ This work was supported by the National Natural Science Foundation of China (No. 11071254) and the

Scientific Research Foundation for the Returned Overseas Chinese Scholars, State Education Ministry, and the
Science Research Foundation of JCB (No. JCB 1005).

c Vilnius University, 2011
Global dynamics of a predator-prey system with Holling type II functional response 243

where u(t), v(t) represent the densities of prey and predator population, respectively.
System (1) is called Holling type II predator-prey model in the literature. This system
is an extension of the familiar Lotka–Volterra system, in which the divisor 1 + αu is
missing, i.e., α = 0. α is interpreted as a constant handling time for each prey captured.
For low prey biomass (αu  1) this response approximates the classical Lotka–Volterra
one. In [5], the stability of equilibrium, Hopf bifurcation, global existence of limit cycles,
global attractivity of equilibria, and codimension two bifurcations are investigated. The
global behavior of system (1) has been discussed by many authors (see, e.g., [5–7]).
In system (1), it is assumed that each individual predator has the same ability to feed
on prey. However, in the natural world, many species go through two or more life stages
as they proceed from birth to death, and in different stages, they have different reactions
to the environment. For example, the immature predators are raised by their parents, and
the rate they attack the prey and the reproductive rate can not be ignored. Stage-structured
population models have received great attention in recent years (see, e.g., [8–13]). In [12],
Yu et al. studied the following strengthen type predator-prey model with stage structure


ẋ(t) = x(t) r − ax(t) − a1 y2 (t) ,
ẏ1 (t) = ey2 (t) − (r1 + D)y1 (t), (2)
ẏ2 (t) = Dy1 (t) − r2 y2 (t) + a2 x(t)y2 (t),

where x(t) represents the density of the prey at time t, y1 (t) and y2 (t) represent the
densities of the immature and the mature predator at time t, respectively; the parameters a,
e, r, a1 , a2 , r1 , r2 and D are positive constants in which a is the intra-specific competition
rate of the prey, a2 /a1 is the rate of conversing prey into new mature predator, e is the
birth rate of new predators, r is the intrinsic growth rate of the prey, r1 is the death
rate of the immature predator and r2 is the death rate of the mature predator, and D
denotes the rate of immature predator becoming mature predator. In system (2), it was
assumed that feeding on prey can only make contribution to the increasing of the physique
of the predator and does not make contribution to the reproductive ability. In [12], the
global asymptotic stability of the coexistence equilibrium was established by constructing
suitable Lyapunov functions.
Motivated by the works of Bazykin [5] and Yu et al. [12], in this paper, we are
concerned with the effect of functional response and stage structure on the dynamics of a
predator-prey system. To this end, we study the following differential equations
 
a1 y2 (t)
ẋ(t) = x(t) r − ax(t) − ,
1 + mx(t)
ẏ1 (t) = ey2 (t) − (r1 + D)y1 (t), (3)
a2 x(t)y2 (t)
ẏ2 (t) = Dy1 (t) − r2 y2 (t) + ,
1 + mx(t)

where a1 x/(1 + mx) describes the Holling type II functional response, here a1 and m
represent the effects of capturing rate and handling time, respectively, and r2 m > a2 is

Nonlinear Anal. Model. Control, 2011, Vol. 16, No. 2, 242–253

244 X. Tian, R. Xu

reasonable for biological meaning, and implies that hunting has a reward in the sense of
diminishing their mortality rate.
It is easy to show that all solutions of system (3) are defined on [0, +∞) and remain
positive for all t ≥ 0.
The organization of this paper is as follows. In the next section, by analyzing the
corresponding characteristic equations, the local stability of each of nonnegative equilib-
ria of system (3) is discussed. In Section 3, we present conditions for the permanence
of system (3). Further, based on discussions above, by using the theory of monotone
flows for three-dimensional competitive system, the existence of orbitally asymptotically
stable periodic solution is obtained. In Section 4, by using suitable Lyapunov functions
and LaSalle invariance principle, sufficient conditions are derived for the global stability
of the predator- extinction equilibrium and the coexistence equilibrium of system (3).
A brief remark is given in Section 5 to conclude this work.

2 Local stability
In this section, we discuss the local stability of each equilibria of system (3) by analyzing
the corresponding characteristic equations. It is easy to show that system (3) always
has a trivial equilibrium E0 (0, 0, 0) and a predator-extinction equilibria E1 (r/a, 0, 0).
Furthermore, if the following holds:
eD
(H1) a2 r > (a + rm)[r2 − D+r1 ] > 0,
then system (3) has a unique coexistence equilibrium E ∗ (x∗ , y1∗ , y2∗ ), where

r2 (D + r1 ) − eD
x∗ = ,
(a2 − r2 m)(D + r1 ) + emD
e a2 (D + r1 )(r − ax∗ )x∗
y1∗ = y2∗ , y2∗ = .
D + r1 a1 [r2 (D + r1 ) − eD]
We now study the local stability of each of nonnegative equilibria of system (3).
By analyzing the characteristic equation of system (3) at the equilibrium E0 (0, 0, 0),
it is easy to show that E0 is always unstable.
The characteristic equation of system (3) at the equilibrium E1 (r/a, 0, 0) takes the
form
(λ + r) λ2 + g1 λ + g0 = 0,


(4)
where
a2 r
g0 = r2 (D + r1 ) − eD − (D + r1 ) ,
a + rm
r2 (a + rm) − a2 r
g1 = D + r1 + .
a + rm
Eq. (4) always has a negative real root λ = −r. If a2 (D + r1 ) < (a + rm)[r2 (D +
r1 ) − eD], then g0 > 0, g1 > 0. It is easy to show that roots of λ2 + g1 λ + g0 = 0

www.mii.lt/NA
Global dynamics of a predator-prey system with Holling type II functional response 245

have only negative real parts. Accordingly, the equilibrium E1 of system (3) is locally
asymptotically stable. If (H1) holds, Eq. (4) has at least one positive real root. Therefore,
E1 is unstable.
The characteristic equation of system (3) at the coexistence equilibrium E ∗ (x∗ , y1∗ , y2∗ )
is of the form
λ3 + p2 λ2 + p1 λ + p0 = 0, (5)
where
  a1 x∗ y2∗
p0 = (a2 − r2 m)(D + r1 ) + emD ,
(1 + mx∗ )2
a1 y2∗ a 2 x∗
 
∗
p1 = (D + r1 + r2 ) + (r − 2ax ) − (D + r1 + r2 ) ,
(1 + mx∗ )2 1 + mx∗
eD a1 y2∗
p2 = D + r1 + 2ax∗ − r + + .
D + r1 (1 + mx∗ )2
It is readily seen that if p2 > 0, p1 p2 − p0 > 0, by the Routh–Hurwitz theorem, the
coexistence equilibrium E ∗ of system (3) is locally asymptotically stable; and E ∗ is
unstable if p1 p2 − p0 < 0.
Based on the discussions above, we have the following result.
Theorem 1. For system (3), we have:
(i) The equilibrium E0 (0, 0, 0) is always unstable.
(ii) If a2 r(D + r1 ) < (a + r2 m)[r2 (D + r1 ) − eD], then the equilibrium E1 (r/a, 0, 0)
is locally asymptotically stable; if (H1) holds, E1 is unstable.
(iii) Let (H1) hold. If p2 > 0, p1 p2 − p0 > 0, then the coexistence equilibrium
E ∗ (x∗ , y1∗ , y2∗ ) of system (3) is locally stable; if p1 p2 − p0 < 0, E ∗ is unstable.

3 Permanence and stable periodic solution

In this section, we establish the permanence of system (3) and the existence of orbitally
asymptotically stable periodic solution.
Before giving our main results, we need the following lemmas.
Lemma 1. Let r2 (D + r1 ) > eD. Then positive solutions of system (3) are ultimately
bounded.
Proof. Let (x(t), y1 (t), y2 (t)) be any positive solution of system (3).
Define
N (t) = a2 x(t) + ky1 (t) + a1 y2 (t),
where k = a1 [r2 (D + r1 ) + eD]/(2e(D + r1 )).
Calculating the derivative of N (t) along positive solutions of (3), it follows that

Ṅ (t) = a2 x(r − ax) − (k(D + r1 ) − a1 D)y1 − (a1 r2 − ke)y2

≤ a2 x(r − ax) − δ(ky1 + a1 y2 ), (6)

Nonlinear Anal. Model. Control, 2011, Vol. 16, No. 2, 242–253

246 X. Tian, R. Xu

where δ = min{D + r1 − a1 D/k, r2 − ke/a1 }. Noting that r2 (D + r1 ) > eD, it follows

from (6) that

a2 (r + δ)2
Ṅ (t) = a2 x(r + δ − ax) − δN (t) ≤ − δN (t). (7)
4a
We derive from (7) that

a2 (r + δ)2
lim sup N (t) ≤ := M.
t→+∞ 4aδ

Hence, for ε > 0 sufficiently small, there exists a T > 0 such that, if t > T ,

x(t) < M + ε, y1 (t) < M + ε, y2 (t) < M + ε.

This completes the proof.

Let X be a complete metric space. Suppose that X 0 ⊂ X, X0 ⊂ X, X 0 ∩ X0 = φ.
Assume that T (t) is a C0 semigroup on X satisfying

T (t) : X 0 → X 0 , T (t) : X0 → X0 . (8)

Let Tb (t) = T (t)|X0 and let Ab be the global attractor for Tb (t). The following lemma
was introduced by Wang and Chen [10].
Lemma 2. [10] Suppose that T (t) satisfies (8) and the following:
(i) there is a t0 ≥ 0 such that T (t) is compact for t > t0 ;
(ii) T (t) is point dissipative in X;
S
(iii) Ãb = ω(x) is isolated and has an acyclic covering M̃ , where
x∈Ab

M̃ = {M1 , M2 , . . . , Mn };
(iv) W s (Mi ) ∩ X 0 = φ for i = 1, . . . , n.
Then X0 is uniform repeller with respect to X 0 , i.e., there is an ε > 0 such that for any
x ∈ X 0 , lim inf t→+∞ d(T (t)x, X0 ) ≥ ε, where d is the distance of T (t)x from X0 .
We now investigate the permanence of system (3).
Theorem 2. If (H1) holds, then system (3) is permanent.
Proof. Define
3

U1 = (x, y1 , y2 ) ∈ R+ : x≡0 ,
3

U2 = (x, y1 , y2 ) ∈ R+ : y1 ≡ 0, y2 ≡ 0 .

If X0 = U1 ∪ U2 and X 0 = intR3+ , it is easy to show that X0 and X 0 are coexistently

invariant. Moreover, by Lemma 1, conditions (i) and (ii) of Lemma 2 are clearly satisfied.

www.mii.lt/NA
Global dynamics of a predator-prey system with Holling type II functional response 247

Hence, we only need to verify the conditions (iii) and (iv). There are two constant
solutions E0 (0, 0, 0) and E1 (r/a, 0, 0) in X0 , corresponding, respectively, to x(t) =
y1 (t) = y2 (t) = 0 and x(t) = r/a, y1 (t) = y2 (t) = 0. If (x(t), y1 (t), y2 (t)) is a
solution of system (3) initiating from U1 , then
ẏ1 (t) = ey2 (t) − (D + r1 )y1 (t),
ẏ2 (t) = Dy1 (t) − r2 y2 (t).
Obviously, if (H1) holds, y1 (t) → 0 and y2 (t) → 0 as t → +∞. If (x(t), y1 (t), y2 (t)) is
a solution of system (3) initiating from U2 with x(0) > 0, it is easy to see that x(t) → r/a
as t → +∞. This shows that if invariant set E0 and invariant set E1 are isolated, {E0 , E1 }
is isolated and is an acyclic covering. It is obvious that E0 is isolated invariant. The
isolated invariance of E1 will be a consequence of the following proof.
We now prove that W s (E0 ) ∩ X 0 = φ and W s (E1 ) ∩ X 0 = φ. We restrict our
attention to the second equation, since the proof for the first is simple. Assume the
contrary. Then there exists a positive solution (x̃(t), ỹ1 (t), ỹ2 (t)) of system (3) such that
 

r
x̃(t), ỹ1 (t), ỹ2 (t) → , 0, 0 , as t → +∞.
a
Choose ξ > 0 sufficiently small satisfying
a2 (D + r1 )(r/a − ξ)
r2 (D + r1 ) − eD < . (9)
m(r/a − ξ)
Let T > 0 be sufficiently large such that
r r
− ξ < x̃(t) < + ξ for t ≥ T.
a a
Then we have, for t ≥ T ,
ỹ10 (t) = eỹ2 (t) − (D + r1 )ỹ1 (t),
a2 (r/a − ξ) (10)
ỹ20 (t) ≥ Dỹ1 (t) − r2 ỹ2 (t) + ỹ2 (t).
1 + m(r/a − ξ)
We consider the matrix Aξ defined by
" #
−D − r1 e
Aξ = a2 (r/a−ξ) .
D 1+m(r/a−ξ) − r2

Since Aξ admits positive off-diagonal elements, the Perron–Frobenius theorem implies

that there is a positive eigenvector υ for the maximum eigenvalue α of Aξ . Moreover, by
computing, we see that the maximum eigenvalue α is positive since we have (9).
We now consider the following system:
y10 (t) = −(D + r1 )y1 (t) + ey2 (t),
(11)
 
0 a2 (r/a − ξ)
y2 (t) = Dy1 (t) − r2 − y2 (t).
1 + m(r/a − ξ)

Nonlinear Anal. Model. Control, 2011, Vol. 16, No. 2, 242–253

248 X. Tian, R. Xu

Let υ = (υ1 , υ2 ) and let l > 0 be small enough such that

ỹ1 (t0 ) > lυ1 , ỹ2 (t0 ) > lυ2 .

If (y1 (t), y2 (t)) is a solution of system (11) satisfying y1 (t0 ) = lυ1 , y2 (t0 ) = lυ2 , since
the semiflow of system (11) is monotone and Aξ υ > 0, it follows from [14] that yi (t) is
strictly increasing and yi (t) → +∞ as t → +∞. Note that ỹi (t) ≥ yi (t) for t > t0 .
We have ỹi (t) → +∞ as t → +∞. This contradicts Lemma 3.1. The above assertion
is thus proved. At this time, we are able to conclude from Lemma 3.2 that X0 repels
the positive solutions of (3) uniformly. As a consequence, there exists a ε > 0 such that
each positive solution (x(t), y1 (t), y2 (t)) of (3) satisfies lim inf t→+∞ x(t) ≥ ε and lies
eventually outside the set Q1 defined by

Q1 = (x, y1 , y2 ): x > 0, 0 < y1 ≤ ε, 0 < y2 ≤ ε .

Let  
Dε(1 + mε)
0 < ρ < min ε,
2(r2 + (r2 m − a2 )ε)
be fixed, where r2 > a2 ε/(1 + mε). Then y20 (t) > Dε/2 on region Q2 defined by

Q2 = (x, y1 , y2 ): x > 0, y1 ≥ ε, 0 < y2 ≤ ρ .

It follows that each positive solution (x(t), y1 (t), y2 (t)) of system (3) leaves Q2 eventu-
ally and lies eventually outside Q2 . Hence, y2 (t) ≥ ρ for t sufficiently large. In view of
lim inf t→+∞ x(t) ≥ ε, we see that for t sufficiently large,

ẏ1 (t) ≥ eρ − (D + r1 )y1 (t).

It follows that
eρ
lim inf y1 (t) ≥ .
t→+∞ D + r1
Consequently, system (3) is permanent. This completes the proof.

In the following, we show that there exists an orbitally asymptotically stable periodic
orbit in system (3).

Theorem 3. Let (H1) hold. If p2 p1 − p0 < 0, then system (3) has an orbitally asymptoti-
cally stable periodic solution.

Proof. Letting z1 = −x, z2 = y1 , z3 = −y2 , system (3) becomes

 
0 a1 z3
z1 = z1 r + az1 + ,
1 − mz1
z20 = −ez3 − (D + r1 )z2 , (12)
a2 z1 z3
z30 = −Dz2 − r2 z3 + .
1 − mz1

www.mii.lt/NA
Global dynamics of a predator-prey system with Holling type II functional response 249

If we write (12) as z 0 = f (z), the Jacobian matrix of f at z is as follows:

a1 z3 a1 z1
r + 2az1 + (1−mz 0
 
2
1) 1−mz1
J(z) =  0 −r1 − D −e .
a2 z3 a2 z 1
(1−mz1 )2 −D −r 2 + 1−mz1

Denote E = {(z1 , z2 , z3 ): z1 < 0, z2 > 0, z3 < 0}. J(z) has nonpositive off-diagonal
elements at each point of E. Thus, system (12) is competitive in E. Let z1∗ = −x∗ , z2∗ =
y1∗ and z3∗ = y2∗ . It is obvious that (z1∗ , z2∗ , z3∗ ) is the unique equilibrium of system (12).
Since p2 p1 − p0 < 0 holds, the analysis above shows that (z1∗ , z2∗ , z3∗ ) is unstable and det
J(z ∗ ) < 0. Moveover, since system (3) is permanent, there exists a compact subset B
of E such that for each z0 ∈ E, there exists a T (z0 ) > 0 such that z(t, z0 ) ∈ B for all
t ≥ T (z0 ). Hence, by Theorem 1.2 of [15], system (12) has an orbitally asymptotically
stable periodic solution. This completes the proof.
In the following, we give one example to illustrate the main results above.
Example 1. In system (3), we let a = 1, a1 = 1, a2 = 4, e = 2, m = 1, r = 5,
r1 = 1, r2 = 2, D = 1. System (3) with above coefficients has a unique coexistence
equilibrium E ∗ (1/3, 56/9, 28/9). Direct calculation shows that (H1) holds and p1 p2 −
p0 ≈ −4.8333 < 0. By Theorem 3, we see that system (3) has an orbitally asymptotically
stable periodic solution. Numerical simulation illustrates the above result (see Fig. 1).

40

30

20
y2

10

−10
30
20 5
10 0
−5 12
0 −10 x 10
y −10 −15
1 x

Fig. 1. The numerical solution of system (3) with a = 1, a1 = 1, a2 = 4, e = 2,

m = 1, r = 5, r1 = 1, r2 = 2, D = 1, and (φ1 , φ2 , φ3 ) = (1, 1, 1).

4 Global stability
In this section, we discuss the global stability of the coexistence equilibrium E ∗ and the
predator-extinction equilibrium E1 of system (3), respectively. The strategy of proofs is
to construct suitable Lyapunov functions and use LaSalle invariance principle.
Theorem 4. If a2 (D+r1 ) < (a+rm)[r2 (D+r1 )−eD], then the equilibrium E1 (r/a, 0, 0)
of system (3) is globally asymptotically stable.

Nonlinear Anal. Model. Control, 2011, Vol. 16, No. 2, 242–253

250 X. Tian, R. Xu

Proof. Let (x(t), y1 (t), y2 (t)) be any positive solution of system (3). Denote x0 = r/a.
Define  
a2 x
V1 (t) = x − x0 − x0 ln + ky1 + a1 y2 , (13)
1 + mx0 x0
where k = a1 D/(D + r1 ).
Calculating the derivative of V1 (t) along positive solutions of system (3), we obtain
that
 
d a2 x0 a1 xy2
V1 (t) = (1 − ) x(r − ax) −
dt 1 + mx0 x 1 + mx
 
  a2 xy2
+ k ey2 − (D + r1 )y1 + a1 Dy1 − r2 y2 + . (14)
1 + mx
On substituting r = ax0 into (14), it follows that
d aa2 a1 a2 xy2 a1 a2 xy2
V1 (t) = − (x − x0 )2 + −
dt 1 + mx0 1 + mx (1 + mx)(1 + mx0 )
 
a1 a2 x0 y2 a1 eD
+ + − a1 r2 y2
(1 + mx)(1 + mx0 ) D + r1
 
aa2 2 a1 a2 y2 x x0
=− (x − x0 ) + x− +
1 + mx0 1 + mx 1 + mx0 1 + mx0
a1  
+ eD − r2 (D + r1 ) y2 . (15)
D + r1
We derive from (15) that
d a2 a2 a1
V1 (t) = − (x − x0 )2 +
dt a+rm (a + rm)(D + r1 )
  
× a2 (D + r1 ) − (a + rm) r2 (D + r1 ) − eD y2 . (16)

Hence, if a2 (D + r1 ) < (a + rm)[r2 (D + r1 ) − eD], then V10 (t) ≤ 0. By Theorem 5.3.1

in [16], solutions limit M, the largest invariant subset of {V10 (t) = 0}. Clearly, we see
from (16) that V10 (t) = 0 if and only if x = x0 , y2 = 0. Noting that M is invariant, for
each element in M, we have x(t) = x0 , y2 (t) = 0. It therefore follows from the third
equation of system (3) that
0 = ẏ2 (t) = Dy1 (t),
which yields y1 (t) = 0. Hence, V10 (t) = 0 if and only if (x, y1 , y2 ) = (x0 , 0, 0). Ac-
cordingly, the global asymptotic stability of E1 follows from LaSalle invariance principle.
This completes the proof.
Theorem 5. Let (H1) hold. Then the coexistence equilibrium E ∗ (x∗ , y1∗ , y2∗ ) of system (3)
is globally stable provided that
r
(H2) x > 2a .
Here, x > 0 is the persistency constant for x satisfying lim inf t→+∞ x(t) ≥ x.

www.mii.lt/NA
Global dynamics of a predator-prey system with Holling type II functional response 251

Proof. Let (x(t), y1 (t), y2 (t)) be any positive solution of system (3). Since x > r/(2a),
it is seen that there is a T > 0 such that x(t) > r/(2a) for all t ≥ T and also that
x∗ > r/(2a). Clearly, p2 > 0. By calculations, we derive that
p1 p2 − p0
a1 y2∗ a1 y2∗
 
eD


= r2 D + r1 + + (D + r1 + r2 ) D + r1 +
(1 + mx∗ )2 D + r1 (1 + mx∗ )2
∗
 
a1 y2
+ (2ax∗ − r) eD + (D + r1 )2 + (D + r1 + r2 )
(1 + mx∗ )2
a1 y2∗
  
∗ eD ∗ eD
+ (2ax − r) D + r1 + 2ax − r + + > 0.
D + r1 D + r1 (1 + mx∗ )2
Accordingly, by Theorem 1, E ∗ is locally asymptotically stable.
Define
   
a2 ∗ ∗ x ∗ ∗ y1
V2 (t) = x − x − x ln ∗ + k y1 − y1 − y1 ln ∗
1 + mx∗ x y1
 
y2
+ a1 y2 − y2∗ − y2∗ ln ∗ , (17)
y2
where k = a1 D/(D + r1 ).
Calculating the derivative of V2 (t) along positive solutions of system (3), it follows
that
x∗
  
d a2 a1 xy2
V2 (t) = 1 − x(r − ax) −
dt 1 + mx∗ x 1 + mx
∗ 
 
y
+ k 1 − 1 ey2 − (D + r1 )y1

y1
y∗
  
a2 xy2
+ a1 1 − 2 Dy1 − r2 y2 + . (18)
y2 1 + mx
On substituting r = ax∗ + a1 y2∗ /(1 + mx∗ ) into (18), we derive that
x∗ a1 x∗ y2∗
  
d a2 ∗ ∗
V2 (t) = 1− x(r − ax) − x (r − ax ) +
dt 1 + mx∗ x 1 + mx∗
∗
 
a2 x a1 xy2 a1 eD a1 eD ∗ y2
− 1− + y2 − y
1 + mx ∗ x 1 + mx D + r1 D + r1 1 y1
y1 a1 a2 xy2 a1 a2 xy2∗
+ a1 Dy1∗ − a1 Dy2∗ − a1 r2 y2 + a1 r2 y2∗ + −
y2 1 + mx 1 + mx
∗ 2 ∗ ∗
x∗
 
a2 (x − x ) ∗
 a1 a2 x y2
= r − a(x + x ) + 1−
1 + mx∗ x (1 + mx∗ )2 x
∗
a1 a2 x a1 eD a1 eD ∗ y2 y1
+ y2 + y2 − a1 r2 y2 − y − a1 Dy2∗
1 + mx∗ D + r1 D + r1 1 y1 y2
∗
a a
1 2 xy
+ a1 Dy1∗ + a1 r2 y2∗ − 2
. (19)
1 + mx

Nonlinear Anal. Model. Control, 2011, Vol. 16, No. 2, 242–253

252 X. Tian, R. Xu

a2 x∗
Noting that ey2∗ = (D + r1 )y1∗ , eD
D+r1 − r2 + 1+mx∗ = 0, (19) can be rewritten as

(x − x∗ )2   a1 eDy2∗ y1∗ y2 y2∗ y1

 
d a2 ∗
V2 (t) = r − a(x + x ) + 2− ∗ − ∗
dt 1 + mx∗ x D + r1 y2 y1 y1 y2
a1 a2 x∗ y2∗ x∗ a1 a2 x∗ y2∗ a1 a2 xy2∗
 
+ 1− + − .
(1 + mx∗ )2 x 1 + mx∗ 1 + mx
(x − x∗ )2   a1 eDy2∗ y1∗ y2 y2∗ y1
 
a2 ∗
= r − a(x + x ) + 2 − −
1 + mx∗ x D + r1 y2∗ y1 y1∗ y2
a1 a2 x∗ y2∗ x∗ (1 + mx) (1 + mx∗ )x
 
+ ∗
2− − . (20)
1 + mx (1 + mx∗ )x x∗ (1 + mx)
Since the arithmetic mean is greater than or equal to the geometric mean, it is clear that
y1∗ y2 y ∗ y1 x∗ (1 + mx) (1 + mx∗ )x
2− ∗ − 2∗ ≤ 0, 2− − ≤ 0,
y2 y1 y1 y2 (1 + mx∗ )x x∗ (1 + mx)
and the equality hold only for x = x∗ , y1 = y1∗ , y2 = y2∗ . If x(t) > r/(2a) for t ≥ T , we
derive that
a2 (x − x∗ )2 
r − a(x + x∗ ) ≤ 0,

1 + mx ∗ x
with equality if and only if x = x∗ . This implies that if x(t) > r/(2a) for t ≥ T ,
V20 (t) ≤ 0, with equality if and only if x = x∗ , y1 = y1∗ , y2 = y2∗ . The largest
compact invariant subset in M = {(x, y1 , y2 ) | V̇2 (x, y1 , y2 ) = 0} is the singleton {E ∗ }.
Therefore, by LaSalle invariance principle, the coexistence equilibrium E ∗ is globally
asymptotically stable. This completes the proof.
Remark. From Theorem 5, we see that if (H1) holds and lim inf t→+∞ x(t) ≥ r/(2a),
then the coexistence equilibrium E ∗ is globally stable. We now give a sufficient condition
for this inequality. Denote δ = min{D +r1 −a1 D/k, r2 −ke/a1 }. By Lemma 1, one can
show that solutions of system (3) have an ultimately upper bound M = a2 (r +δ)2 /(4aδ).
Hence, we derive from the first equation of system (3) that ẋ(t) ≥ x(t)(r−a1 M −ax(t)),
which yields lim inf t→∞ x(t) ≥ x = (r − a1 M )/a = [4arδ − a1 a2 (r + δ)2 ]/(4a2 δ).
Obviously, we need only to choose parameters satisfying 2arδ > a1 a2 (r + δ)2 .

5 Conclusion
In this paper, we have studied the global dynamics of a predator-prey model with Holling
type II functional response and stage structure. The global stability of the predator-extinc-
tion equilibrium E1 and the coexistence equilibrium E ∗ of system (3) has been established
by using the Lyapunov–LaSalle type theorem. By Theorem 5, we see that if (H1) and (H2)
hold, the coexistence equilibrium E ∗ is globally asymptotically stable. Biologically, these
indicate that when the intrinsic growth rate of the prey, the birth rate of new predators and
the rate of immature predator becoming mature predator are large enough, and the death
rate of the immature predator and mature predator are small enough, then the prey and the
predator population coexist, and the ecological system is therefore permanent.

www.mii.lt/NA
Global dynamics of a predator-prey system with Holling type II functional response 253

Acknowledgments
The authors wish to thank the reviewers for their valuable comments and suggestions that
greatly improved the presentation of this work.

References
1. N. Apreutesei, A. Ducrot, V. Volpert, Competition of species with intra-specific competition,
Math. Model. Nat. Phenom., 3, pp. 1–27, 2008.
2. H. Malchow, S. Petrovskii, E. Venturino, Spatiotemporal Patterns in Ecology and Epidemiolo-
gy, Chapman & Hall/CRC Press, Boca Raton, 2008.
3. S.G. Ruan, D.M. Xiao, Global analysis in a predator-prey system with nonmonotonic functional
response, SIAM J. Appl. Math., 61, pp. 1445–1472, 2001.
4. C.S. Holling, The functional response of predator to prey density and its role in mimicry and
population regulation, Mem. Entomol. Soc. Can., 45, pp. 1–60, 1965.
5. A.D. Bazykin, Structural and Dynamic Stability of Model Predator-Prey Systems, Int. Inst.
Appl. Syst. Anal., Laxenburg. Res. Rep., IIASA, Laxenburg, 1976.
6. J. Hainzl, Stability and Hopf bifurcation in a predator-prey system with several parameters,
SIAM J. Appl. Math., 48, pp. 170–190, 1988.
7. Y. Kuang, Delay Differential Equations with Applications in Population Dynamics, Academic
Press, New York, 1993.
8. A. S. Ackleh, P. Zhang, Competitive exclusion in a discrete stage-structured two species model,
Math. Model. Nat. Phenom., 4, pp. 156–175, 2009.
9. S.A. Gourley, Y. Kuang, A stage structured predator-prey model and its dependence on through-
stage delay and death rate, J. Math. Biol., 49, pp. 188–200, 2004.
10. W.D. Wang, L.S. Chen, A predator-prey system with stage-structure for predator, Comput.
Math. Appl., 33, pp. 83–91, 1997.
11. R. Xu, Z.E. Ma, Stability and Hopf bifurcation in a ratio-dependent predator-prey system with
stage structure, Chaos Solitons Fractals, 38, pp. 669–684, 2008.
12. Y.M. Yu, Y. Zhang, W.D. Wang, The asymptotic behavior of a strengthen type predator-prey
model with stage structure, J. Southeast Univ., Nat. Sci. 26, pp. 363–367, 2001.
13. X.A. Zhang, L.S. Chen, A.U. Neumann, The stage-structured predator-prey model and optimal
harvesting policy, Math. Biosci., 168, pp. 201–210, 2000.
14. H.L. Smith, Monotone semiflows generated by functional differential equations, J. Differ.
Equations, 66, pp. 420–442, 1987.
15. H.R. Zhu, H.L. Smith, Stable periodic orbits for a class of three-dimensional competitive
system, J. Differ. Equations, 110, pp. 143–156, 1994.
16. J.K. Hale, Theory of Functional Differential Equations, Springer-Verlag, New York, 1976.

Nonlinear Anal. Model. Control, 2011, Vol. 16, No. 2, 242–253