Nonlinear Analysis: Real World Applications 11 (2010) 246–252

Contents lists available at ScienceDirect

Nonlinear Analysis: Real World Applications

journal homepage: www.elsevier.com/locate/nonrwa

Qualitative analysis of a predator–prey model with Holling type II

functional response incorporating a constant prey refuge
Liujuan Chen a,∗ , Fengde Chen b , Lijuan Chen b
a
Department of Mathematics and Physics, Fujian Institute of Education, Fuzhou, Fujian, 350001, PR China
b
College of Mathematics and Computer Science, Fuzhou University, Fuzhou, Fujian, 350002, PR China

article info abstract

Article history: A predator–prey model with Holling type II functional response incorporating a constant
Received 3 June 2008 prey refuge is investigated. Depending on a constant prey refuge m, which provides a
Accepted 29 October 2008 condition for protecting m of prey from predation, we show the instability and global
stability properties of the equilibria and the existence and uniqueness of limit cycles for
Keywords: the model. Two examples are carried out to illustrate the validity of our results.
Predator–prey model
© 2008 Elsevier Ltd. All rights reserved.
Prey refuge
Limit cycle
Holling type II
Global stability

1. Introduction

The aim of this paper is to consider the instability and global stability properties of the equilibria and the existence and
uniqueness of limit cycles of the following model with Holling type II functional response incorporating a constant prey
refuge m
 x β(x − m)y
ẋ = α x 1 − − ,
k 1 + a(x − m) (1.1)
c β(x − m)y
ẏ = −dy + ,
1 + a(x − m)
where x, y denote prey and predator population respectively at any time t, d, k, α, β, a, c , m are positive constants. Here α
represents the intrinsic growth rate and k the carrying capacity of the prey; m is a constant number of prey using refuges,
which protects m of prey from predation; d is the death rate of the predator; c is the conversion factor denoting the number
βx
of newly born predators for each captured prey. The term 1+ax denotes the functional response of the predator, which is
termed as Holling type II response function (see Holling CS [1]).
This work is motivated for the paper by Gonzälez-Olivares and Ramos-Jiliberto [2]. They studied the above system (1.1)
and showed the local stability of equilibria and existence of limit cycle, but they did not prove the global stability of the
positive equilibrium and the uniqueness of limit cycle. This paper gives the complete qualitative analysis for the model (1.1)
and the results improve and extend the corresponding results of [2].
It is well known that many more attentions have paid on the effects of a prey refuge for predator–prey system. Hassel [3]
showed that adding a large refuge to a model, which exhibited divergent oscillations in the absence of a refuge, replaced the
oscillatory behaviour with a stable equilibrium. McNair [7] obtained that a prey refuge with legitimate entry–exit dynamics

∗ Corresponding author.
E-mail addresses: [email protected] (L. Chen), [email protected] (F. Chen).

1468-1218/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.nonrwa.2008.10.056
 L. Chen et al. / Nonlinear Analysis: Real World Applications 11 (2010) 246–252 247

was quite capable of amplifying rather than damping predator–prey oscillations. McNair [6] showed that several kinds of
refuges could exert a locally destabilizing effect and create stable, large-amplitude oscillations which would damp out if
no refuge was present. Even now, prey refuges are widely believed to prevent prey extinction and damp predator–prey
oscillations. For example, Kar [4] and Huang, etc. [5] considered a Lotka–Volterra type predator–prey system incorporating
a constant proportion of prey using refuges mx, which protects mx of prey from predation, with Holling type II response
function and Holling type III response function, respectively. They all pointed that increasing the amount of refuge could
increase prey densities and lead to population outbreaks. Moreover, the results of this work also indicate that refuge had a
stabilizing effect on prey–predator interactions and the dynamic behavior very much depends on the prey refuge parameter
m. For more biological backgrounds and results on the effects of a prey refuge, one could refer to [8–12] and the references
therein.
The rest of the paper is organized as follows. Basic properties of the model are given in Section 2 and in Section 3, sufficient
conditions for the global stability of the unique positive equilibrium are obtained. In Section 4, we derive the existence and
uniqueness of limit cycle and give two examples to illustrate the validity of our results. This paper ends by a brief conclusion.

2. Basic properties of the model

Let Ω0 = {(x, y)|x > m, y > 0}, for practical biological meaning, we simply study system (1.1) in Ω0 or in Ω0 . From the
first equation of system (1.1), it is easy to derive lim supt →∞ x(t ) ≤ k.

Lemma 2.1. The solution x(t ), y(t ) of system (1.1) with initial values x(0) > m, y(0) > 0 are positive and bounded for all t ≥ 0.

Proof. Obviously the solutions (x(t ), y(t )) with initial values x(0) > m, y(0) > 0 are positive for all t ≥ 0. Define the
function ω(x, y) = cx + y. Given any  > 0, x(t ) ≤ k +  for t sufficiently large, from system (1.1), it follows that
 x
ω̇ = c α x 1 − − dy ≤ − min{d, α}ω + 2c α(k + ),
k
2c α(k+)
then ω ≤ min{d,α}
, for t sufficiently large, which completes the proof. 

For simplicity, we take the following scaling: x = x − m, y = β y, t = (1 + a(x − m))τ and then the system (1.1) takes
the following form (still denote x, y, τ as x, y, t)

1
ẋ = b(1 + ax)(m + x)(k − m − x) − xy = P (x, y),
1 (2.1)
ẏ = (−d + (c β − ad)x)y = Q (x, y),

where b = α/k. Clearly, Ω0 transforms to Ω = {(x, y)|x > 0, y > 0} and system (2.1) is bounded. System (2.1) has at most
two equilibria E (k − m, 0), E∗ (x∗ , y∗ ) in Ω , where x∗ = d/(c β − ad), y∗ = b(1 + ax∗ )(m + x∗ )(k − m − x∗ )/x∗ . E∗ (x∗ , y∗ )
is a unique positive equilibrium if and only if c β − ad > 0, 0 < m < k − d/(c β − ad). If c β − ad ≤ 0 holds, then it follows
from (2.1) that limt →+∞ y(t ) = 0. Similarly to the proof of following Theorem 2.3, we derive that the unique equilibrium
E (k − m, 0) of system (2.1) is globally asymptotically stable. In the following, we always suppose that c β − ad > 0.

Lemma 2.2. If m > k − c β− d

adh
holds, then E is stable node point; if 0 < m < k − c β−
d
ad
holds, then E is saddle point and
i
furthermore assume that m1 = (k + 2ax2∗ ) − k2 + 4x2∗ (1 + ax∗ )2 /2, then
p

(I) E∗ is locally asymptotically stable for all 0 < m < k − d

c β−ad
, if ak − 1 ≤ 2ad/(c β − ad) holds;
(II) E∗ is locally asymptotically stable for m1 < m < k − d
c β−ad
and E∗ is locally unstable for 0 < m < m1 ,
if ak − 1 > 2ad/(c β − ad) holds.

Proof. The Jacobian matrix of system (2.1) for the equilibrium point E is given by

−b[m + (1 + am)(k − m) + a(k − m)2 ] −(k − m)

 
J (E ) = .
0 (c β − ad)(k − m) − d

If m > k − c β−d
ad
holds, the eigenvalues of matrix −b[m + (1 + am)(k − m) + a(k − m)2 ] and (c β − ad)(k − m) − d are
negative, hence, E is locally asymptotically stable and furthermore E is stable node point. If 0 < m < k − c β− d
ad
holds, one
of the eigenvalues of matrix (c β − ad)(k − m) − d is positive, hence, E is locally unstable and furthermore E is saddle point.
The Jacobian matrix of system (2.1) for E∗ is given by
b
!
− P −x ∗
J (E∗ ) = x∗ ,
ωy∗ 0
248 L. Chen et al. / Nonlinear Analysis: Real World Applications 11 (2010) 246–252

where P = 2ax3∗ + (1 − ak + 2am)x2∗ + (k − m)m and ω = c β − ad. Note that

b b
tr(J (E∗ )) = − P = [m2 − (k + 2ax2∗ )m − (1 − ak)x2∗ − 2ax3∗ ],
x∗ x∗
consider the function R(x) = x2 − (k + 2ax2∗ )x − (1 − ak)x2∗ − 2ax3∗ , the discriminant of R(x) = 0 takes the following form

∆ = (k + 2ax2∗ )2 + 4x2∗ (1 − ak + 2ax∗ ) = k2 + 4x2∗ (1 + ax∗ )2 ,

√
(k+2ax2∗ )+ k2 +4x2∗ (1+ax∗ )2
therefore, R(x) has two roots m1 and m2 , where m2 = 2
. It is easy to see m2 > k.
(I) if ak − 1 ≤ 2ad/(c β − ad) holds, then m1 ≤ 0 and for all m1 < x < m2 , R(x) < 0, hence, for all 0 < m < k − c β− d
ad
,
tr(J (E∗ )) < 0, E∗ is locally asymptotically stable.
(II) if ak − 1 > 2ad/(c β − ad) holds, then m1 > 0 and when m1 < x < m2 , R(x) < 0, when 0 < m < m1 , R(x) > 0.
√ (k+2ax2∗ )−2x∗ (1+ax∗ )
Notice that ∆ > 2x∗ (1 + ax∗ ), then m1 < 2
= 2k − x∗ < k − c β−d ad . Hence, when m1 < m < k − c β−d ad ,
tr(J (E∗ )) < 0, E∗ is locally asymptotically stable; when 0 < m < m1 , tr(J (E∗ )) > 0, E∗ is locally unstable. This completes
the proof. 

Theorem 2.3. If m > k − c β−

d
ad
holds, E is globally asymptotically stable.

Proof. Notice that m > k − c β− d

ad
, E (k − m, 0) is the unique equilibrium of system (2.1). If system (2.1) exists a closed
orbit in Ω , then there must exist an equilibrium in the interior of the closed orbit, which is impossible. Hence, system (2.1)
does not exist limit cycle for all m > k − c β− d
ad
, from the boundedness of system (2.1), the stable node point E is globally
asymptotically stable. 
dx
= X (x, y), dy = Y (x, y). Along a limit cycle L, if div(X , Y )dt < 0(>0) holds,
H
Lemma 2.4 (Zhang, et al. [13]). Assume that dt dt L
then L is stable (unstable) limit cycle.

Lemma 2.5 (Zhang, et al. [13]). Assume that dx dt

= X (x, y, ), dt = Y (x, y, ), (X (x, y, ), Y (x, y, )) structures a rotation
dy

vector field in the wider sense. And furthermore suppose that the set of equations admit semi-stable limit cycle L0 for  = 0 .
Then, when parameter  changes according to appropriate direction, L0 at least decomposes to a stable and an unstable limit
cycle, and they lay in inboard and outboard of L0 , respectively; while  changes in opposite direction, L0 disappears.

3. Global stability for the positive equilibrium

√ √
(4ak+5)−3 1+8ak (4ak+5)+3 1+8ak
Denote m3 = 8a
, m4 = 8a
.

Theorem 3.1. Assume that

(H1 ) ak − 1 ≤ min{2am, c β−
2ad
ad
}, 0 < m < k − c β−d ad ;
(H2 ) c β−ad < ak − 1 ≤ 2am, m1 < m < k − c β−d ad ;
2ad

(H3 ) 2am < ak − 1 ≤ c β− 2ad

ad
, m3 < m < min{m4 , k − c β−d ad };
(H4 ) ak − 1 > max{2am, c β− 2ad
ad
}, max{m1 , m3 } < m < min{m4 , k − c β−d ad }.
Then the positive equilibrium E∗ (x∗ , y∗ ) of system (2.1) is globally asymptotically stable, if any one of (H1 )–(H4 ) holds.
Proof. Define a Dulac function B(x, y) = x−1 y−1 , from system (2.1), we have
∂(BP ) ∂(BQ ) 1
D= + = −bx−2 y−1 [2ax3 + (1 − ak + 2am)x2 + m(k − m)] = −bx−2 y−1 ϕ(x).
∂x ∂y
Then ϕ 0 (x) = 2x[3ax + (1 − ak + 2am)] = 0 exists two roots
1 − ak + 2am
x1 = 0, x2 = − .
3a
If ak − 1 ≤ 2am holds, then x2 ≤ 0, for all x ≥ 0, ϕ 0 (x) ≥ 0 and ϕ 0 (x) = 0 holds if and only if x = 0. Hence, ϕ(x) increases
in [0, +∞), from ϕ(0) = m(k − m) > 0, we derive ϕ(x) > 0, for all x ≥ 0, therefore, for all x ≥ 0, D < 0, system (2.1) does
not exist limit cycle.
If ak − 1 > 2am holds, notice that ϕ 0 (k − m) = 2(k − m)(2ak + 1 − am) > 0, then 0 < x2 < k − m and when
0 < x < x2 , ϕ 0 (x) < 0, x > x2 , ϕ 0 (x) > 0, hence, ϕ(x2 ) is the lowest value in Ω . Now, it is sufficient to prove ϕ(x2 ) > 0.
Since ϕ 0 (x2 ) = 2x2 [3ax2 + (1 − ak + 2am)] = 0, then
ϕ(x2 ) = −ax32 + m(k − m) > −ax32 + mx2
1 − ak + 2am
= 2
[4a2 m2 − (4ak + 5)am + (1 − ak)2 ].
27a
 L. Chen et al. / Nonlinear Analysis: Real World Applications 11 (2010) 246–252 249

Consider the function φ(x) = 4a2 x2 − (4ak + 5)ax + (1 − ak)2 , the discriminant of φ(x) = 0 takes the following form
∆ = (4ak + 5)2 a2 − 16a2 (1 − ak)2 = 9a2 (1 + 8ak),
therefore, φ(x) has two positive roots m3 and m4 , where m3 and m4 denote the above. If m3 < x < m4 holds, then φ(x) < 0,
which means ϕ(x2 ) > 0 for all m3 < x < m4 , then for all x ≥ 0 and m3 < m < m4 , D < 0, system (2.1) does not exist limit
cycle.
If any one of (H1 )–(H4 ) holds, E∗ is locally asymptotically stable and system (2.1) does not exist limit cycle in Ω , from
the boundedness of system (2.1), E∗ is globally asymptotically stable. 

4. Existence and uniqueness of limit cycle

Theorem 4.1. Suppose that ak − 1 > 2ad/(c β − ad) and 0 < m < m1 . System (2.1) admits at least one limit cycle in Ω .

Proof. For system (2.1), constructing a Bendixson ring OABCD

\ including E∗ . Define OA as a length of line L1 = y = 0, AB as
a length of line L2 = x − (k − m) = 0. Define

ẋ = x(a0 − y),
(4.1)
ẏ = (−d + (c β − ad)x)y,
where a0 = maxx∗ ≤x≤k−m {b(1 + ax)(m + x)(k − m − x)/x}. The orbit of system (4.1) with the initial value point
B(k − m, a0 ) intersects with line x = x∗ and the intersection point C (x∗ , y1 ), then we derive the curve BCc.CD as a length
of line
 L 3 = y − y 1 = 0, DO as a length
 of line L 4 = x = 0. Since OA as a length of orbit for system (2.1) and
(2.1) = −(k − m)y < 0(y > 0), dt3  = (−d + (c β − ad)x)y1 < 0(0 < x < x∗ ), = bm(k − m) > 0, the orbits
dL2 dL  dL4
|

dt dt (2.1)
(2.1)
of system (2.1) go through into the interior of thedyBendixsondyring from the outer of AB, CD and DO; on BC , compared system
c
(2.1) to system (4.1): dt (2.1) < dt (4.1) < 0 and dt |(2.1) = rmdt |(4.1) > 0, then the orbits of system (2.1) go through into the
dx  dx 

c. On the other hand, under the assumption of Theorem 4.1, E∗ (x∗ , y∗ ) is an

interior of the Bendixson ring from the outer of BC
\ ⊂ Ω.
unstable equilibrium, by Poincaré–Bendixson Theorem, System (2.1) admits at least one limit cycle in region OABCD
This completes the proof of Theorem 4.1. 

Theorem 4.2. Suppose that ak − 1 > 2ad/(c β − ad) and 0 < m < m1 . System (2.1) admits at most one limit cycle which is
globally asymptotically stable in Ω .
Proof. In order to prove Theorem 4.2, we take the following change of variables v = x, u = ln y, τ = xt, still denote v, u, τ
as x, y, t, then system (2.1) takes the following form

b(1 + ax)(m + x)(k − m − x)

ẋ = −ey + ,
x (4.2)
−d + (c β − ad)x
ẏ = ,
x
and the positive equilibrium E∗ (x∗ , y∗ ) → E∗ (x∗ , ln y∗ ).
Taking translation transformation x = x0 + x∗ , y = y0 + ln y∗ , then E∗ transform to the origin O(0, 0), still denote x0 , y0 as
x, y, system (4.2) takes the following form

b(1 + a(x + x∗ ))(m + x + x∗ )(k − m − x − x∗ )

ẋ = −y∗ ey +
x + x∗
1
= −φ(y) − F (x) (x > −x∗ ) (4.3)
−d + (c β − ad)(x + x∗ ) 1
ẏ = = g (x).
x + x∗
where
b(1 + a(x + x∗ ))(m + x + x∗ )(k − m − x − x∗ )
F (x) = − ,
x + x∗
b[2a(x + x∗ )3 + (1 − ak + 2am)(x + x∗ )2 + m(k − m)]
f (x) = F 0 (x) = .
(x + x∗ )2
Notice that under the assumption of Theorem 4.2, E∗ (x∗ , y∗ ) is an unstable equilibrium and
b
tr(J (E∗ )) = − [2ax3∗ + (1 − ak + 2am)x2∗ + (k − m)m] > 0,
x∗
thus f (0) = −x−
∗ tr(J (E∗ )) < 0 and g (0) = 0, when x 6= 0, xg (x) > 0.
1
250 L. Chen et al. / Nonlinear Analysis: Real World Applications 11 (2010) 246–252

Under the assumption of Theorem 4.2, O(0, 0) is an unstable equilibrium of system (4.3). Thus, if there exists closed orbits
of system (4.3), then define L1 as the nearest orbit to the equilibriumH O and L1 is inboard stable. Suppose the parameter
equation of the closed orbit L1 : x = x1 (t ), y = y1 (t ), By Lemma 2.4, L f (x1 (t ))dt ≥ 0. Furthermore suppose that there
1
exists a closed orbit L2 ⊃ L1 of system (4.3), L2 is next to L1 and the parameter equation of the closed orbit L2 : x = x2 (t ),
y = y2H(t ). Thereinafter,Hwe depart the proof of Theorem 4.2 to two steps:
(I) L f (x2 (t ))dt > L f (x1 (t ))dt;
2 1
(II) L1 is impossible to be semi-stable limit cycle.
Therefore, if (II) holds, then L1 must be stable; by (I), there exists no other closed orbits except L1 , which completes the
proof of uniqueness. 

Proof (I). Constructing a function f1 (x) = f (x) + δ g (x). Suppose Q1 , P1 are the least and most x-coordinate points in L1 ,
f (xQ )
respectively, and they must lay on the curve −φ(y) − F (x) = 0. Let δ = − g (x 1 ) , then f1 (0) = f (0) < 0, f1 (xQ1 ) = 0. Do a
Q1
perpendicular line passing through Q1 , the perpendicular line intersects with x-axis and the foot of a perpendicular M, then
f1 (xM ) = f1 (xQ1 ) = 0. We shall show:
(T1 ) f1 (x) must intersect with positive x-semi-axis and the x-coordinate of the intersection point N satisfies xN ≤ xP1 ;
(T2 ) For all x ∈ (xM , xN ), f1 (x) < 0;
(T3 ) For all x ∈ (−x∗ , xM ) ∪ (xN , +∞), function fg1((xx)) is non-decreasing. 

Proof (T1 ). If (T1 ) does not hold, then on region L1 , f1 (x) ≤ 0, 6≡ 0, and then
I I I
f (x1 (t ))dt = f1 (x1 (t ))dt − δ g (x1 (t ))dt
L1 L1 L1
I I I
= f1 (x1 (t ))dt − δ dy = f1 (x1 (t ))dt < 0,
L1 L1 L1

f (x1 (t ))dt ≥ 0. This completes the proof of (T1 ).

H
which contradicts to L1


From the proof of (T1 ), we derive (xi (t ))dt = f (xi (t ))dt, i = 1, 2. Thus, (I) can be rewrite as the following form
H H
f
Li 1 Li
I I
(I0 ) f1 (x2 (t ))dt > f1 (x1 (t ))dt .
L2 L1

Proof (T2 ).
f10 (x) = f 0 (x) + δ g 0 (x)
2b[a(x + x∗ )3 + m2 − km]
 
1
= + δd
(x + x∗ )2 x + x∗
1 ψ(x)
=
(x + x∗ )2
2b[2a(x + x∗ )3 + m(k − m)]
ψ 0 (x) = > 0, x > −x ∗ .
(x + x∗ )2
Clearly, ψ(x) increases in (−x∗ , +∞). If ψ(x) ≥ 0, then f10 (x) ≥ 0, which contradicts to f1 (xM ) = 0, f1 (0) < 0; if ψ(x) ≤ 0,
then f10 (x) ≤ 0, which contradicts to f1 (0) < 0, f1 (xN ) = 0. Hence, there must exist an x0 ∈ (xM , xN ) so that ψ(x0 ) = 0
and x < x0 , ψ(x) < 0, x > x0 , ψ(x) > 0, then f10 (x0 ) = 0 and x < x0 , f10 (x) < 0, x > x0 , f10 (x) > 0. Combined with
f1 (xM ) = f1 (xN ) = 0, for all x ∈ (xM , xN ), f1 (x) < 0. 

Proof (T3 ).
∂ f1 (x) ∂ f ( x) bΨ (x)
   
1
= = ,
∂x g (x) ∂x g (x) (x + x∗ )2 [(c β − ad)(x + x∗ ) − d]2
where
Ψ (x) = 2(c β − ad)(x + x∗ )[a(x + x∗ )3 − km + m2 ] − d[4a(x + x∗ )3 + (1 − ak + 2am)(x + x∗ )2 − km + m2 ].
From tr(J (E∗ )) > 0 and d = (c β − ad)x∗ , we derive
Ψ 0 (x) = 2(c β − ad)[4a(x + x∗ )3 + (m2 − km) − 6ax∗ (x + x∗ )2 − (1 − ak + 2am)x∗ (x + x∗ )],
Ψ 00 (x) = 2(c β − ad)[12ax(x + x∗ ) − (1 − ak + 2am)x∗ ].
 L. Chen et al. / Nonlinear Analysis: Real World Applications 11 (2010) 246–252 251

Since tr(J (E∗ )) > 0, then 1 − ak + 2am < 0 and for all x > −x∗ , Ψ 00 (x) > 0, so, Ψ 0 (x) increases for all x > −x∗ . Since
Ψ 0 (−x∗ ) = 2m(c β − ad)(m − k) < 0, Ψ 0 (0) = 2(c β − ad)[−2ax3∗ − (1 − ak + 2am)x2∗ + m2 − km] > 0, then there exists
an x0 ∈ (−x∗ , 0) so that Ψ 0 (x0 ) = 0, where
Ψ 0 (x0 ) = 2(c β − ad)[4a(x0 + x∗ )3 + (m2 − km) − 6ax∗ (x0 + x∗ )2 − (1 − ak + 2am)x∗ (x0 + x∗ )],
and hence, when x0 > x > −x∗ , Ψ 0 (x) < 0, when x > x0 , Ψ 0 (x) > 0, so Ψ (x) takes the least value at x = x0 . Noticing that
x0 < 0 and
(1 − ak + 2am)x∗ (x0 + x∗ ) = 4a(x0 + x∗ )3 + (m2 − km) − 6ax∗ (x0 + x∗ )2 ,
then
Ψ (x0 ) = −(c β − ad)(m2 − km)x∗ + (c β − ad)(x0 + x∗ ) · [2a(x0 + x∗ )3 + 2(m2 − km)
− 4ax∗ (x0 + x∗ )2 − (1 − ak + 2am)x∗ (x0 + x∗ )]
= −(c β − ad)(m2 − km)x∗ + (c β − ad)(x0 + x∗ ) · [−2a(x0 + x∗ )3 + (m2 − km) + 2ax∗ (x0 + x∗ )2 ]
= (c β − ad)(m2 − km)x0 − 2a(c β − ad)(x0 + x∗ )3 x0 > 0,
f (x)
we derive Ψ (x) > 0 for all x > −x∗ ,which means for all x ∈ (−x∗ , xM ) ∪ (xN , +∞), function g1(x) is non-decreasing.
By (T1 ), (T2 ), (T3 ) and Green formula (the regions G1 , G2 , G3 , G4 refer to the figure in Zhang, et al. [13], page 274), we
have
∂ f 1 ( x) ∂ f 1 ( x)
I I 
Z Z  Z Z  
f1 (x2 (t ))dt − f1 (x1 (t ))dt = − dxdy + dxdy
L2 L1 G1 ∪G3 ∂ y −φ(y) − F (x) G2 ∪G4 ∂ x g ( x)
f1 (x)y∗ ey ∂ f 1 ( x)
Z Z Z Z  
= − dxdy + dxdy > 0,
G1 ∪G3 (φ( y ) + F ( x )) 2
G2 ∪G4 ∂ x g ( x)
which completes the proof of (I). 

Proof (II). Assume that L1 is a semi-stable limit cycle and line x = x3 < 0 intersects with L1 . Constructing a new function
F̄ (x) = F (x) −  r (x), where 0 <   1,
0, x ≥ x3 ,

r (x) =
(g (x) − g (x3 ))2 , x < x3 .

It is obvious that
I I
(I 00 ) f¯ (x)dt > f¯ (x)dt ,
L̄2 L̄1

holds, where f¯ (x) = F̄ 0 (x).

On the other hand, for parameters 0 < 1 < 2  1, we have
−φ(y) − F (x) + 2 r (x) g (x)
 
−φ(y) − F (x) + 1 r (x) = r (x)g (x)(2 − 1 ) ≤ 0.
g (x)


Therefore, (−φ(y) − F̄ (x), g (x)) structures a rotation vector field in the wider sense on parameter  . Considering

ẋ = −φ(y) − F̄ (x)
(4.4)
ẏ = g (x),
when  = 0, system (4.4) transforms to system (4.3), by Lemma 2.5, when 0 <   1, the semi-stable limit cycle L1 of
system (4.3) decomposes into at least two rings L̄2 ⊃ L̄1 of system (4.4) and L̄1 is inboard stable, L̄2 is outboard unstable,
then
Z Z
f¯ (x)dt ≥ 0, f¯ (x)d ≤ 0,
L̄1 L̄2

which contradicts to (I00 ). Then completes the proof of (II) and then Theorem 4.2. 

Now given the following examples to illustrate the main results.

Example 1. Let α = 5, k = 50, d = 1, a = c = 0.1, β = 2, by simple computation, we have m1 = 7.52, m3 = 7.238, m4 =

55.262, k − c β−
d
ad
= 40, ak − 1 = 4, c β−
2ad
ad
= 2. Then
(1) 0 < m < 7.52, the conditions of Theorems 4.1 and 4.2 hold, model (1.1) admits exactly one limit cycle;
(2) 7.52 < m < 20, (H4 ) holds, E∗ is globally stable;
252 L. Chen et al. / Nonlinear Analysis: Real World Applications 11 (2010) 246–252

(3) 20 ≤ m < 40, (H2 ) holds, E∗ is globally stable;

(4) m > 40, the conditions of Theorem 2.3 hold, E is globally stable.

Example 2. Let α = 5, k = 40, d = 2, a = 0.2, c = 0.1, β = 5, by simple computation, we have m3 = 8.0083, m4 =

38.2419, k − c β−
d
ad
= 20, ak − 1 = 7, c β−
2ad
ad
= 8. Then
(1) 8.0083 < m < 17.5, (H3 ) holds, E∗ is globally stable;
(2) 17.5 ≤ m < 20, (H1 ) holds, E∗ is globally stable.

5. Conclusions

This paper considers a predator–prey model with Holling type II functional response incorporating a constant prey refuge.
We give the complete qualitative analysis of the instability and global stability properties of the equilibria and the existence
and uniqueness of limit cycles for the model. Our results and examples indicate that dynamic behavior of the model very
much depends on the prey refuge parameter m and increasing the amount of refuge could increase prey densities and lead
to population outbreaks.

Acknowledgements

The research was supported by the Foundation of Education Department of Fujian Province (JA08253) and the National
Natural Science Foundation of China (10501007) and the Program for New Century Excellent Talents in Fujian Province
University (0330-003383).

References

[1] C.S. Holling, The functional response of predators to prey density and its role in mimicry and population regulations, Memoirs of the Entomological
Society of Canada 45 (1965) 3–60.
[2] E. González-Olivares, R. Ramos-Jiliberto, Dynamic consequences of prey refuges in a simple model system: More prey,fewer predators and enhanced
stability, Ecological Modelling 166 (2003) 135–146.
[3] M.P. Hassel, The Dynamics of Arthropod Predator–Prey Systems, Princeton University Press, Princeton, 1978.
[4] T.K. Kar, Stability analysis of a prey–predator model incorporating a prey refuge, Communications in Nonlinear Science and Numerical Simulation 10
(2005) 681–691.
[5] Y.J. Huang, F.D. Chen, Z. Li, Stability analysis of a prey–predator model with Holling type III response function incorporating a prey refuge, Applied
Mathematics and Computation 182 (2006) 672–683.
[6] J.N. McNair, The effects of refuges on predator–prey interactions: A reconsideration, Theoretical Population Biology 29 (1986) 38–63.
[7] J.N. McNair, Stability effects of prey refuges with entry–exit dynamics, Journal of Theoretical Biology 125 (1987) 449–464.
[8] W. Ko, K. Ryu, Qualitative analysis of a predator–prey model with Holling type II functional response incorporating a prey refuge, Journal of Differential
Equations 231 (2006) 534–550.
[9] T.K. Kar, Modelling and analysis of a harvested prey–predator system incorporating a prey refuge, Journal of Computational and Applied Mathematics
185 (2006) 19–33.
[10] J.B. Collings, Bifurcation and stability analysis of a temperature-dependent mite predator–prey interaction model incorporating a prey refuge, Bulletin
of Mathematical Biology 57 (1995) 63–76.
[11] A. Sih, Prey refuges and predator–prey stability, Theoretical Population Biology 31 (1987) 1–12.
[12] V. Krivan, Effects of optimal antipredator behavior of prey on predator–prey dynamics: The role of refuges, Theoretical Population Biology 53 (1998)
131–142.
[13] Z.F. Zhang, T.R. Ding, W.Z. Huang, Z.X. Dong, Qualitative Theory of Differential Equations, first ed., Science Publishing Company, Beijing, 1985.