Applied Mathematical Modelling 33 (2009) 2248–2256

Contents lists available at ScienceDirect

Applied Mathematical Modelling

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

A predator–prey model with disease in the prey and two impulses

for integrated pest management
Ruiqing Shi a,b,*, Xiaowu Jiang c, Lansun Chen a
a
Department of Applied Mathematics, Dalian University of Technology, Dalian, Liaoning 116024, People’s Republic of China
b
School of Mathematics and Computer Science, Shanxi Normal University, Linfen, Shanxi 041004, People’s Republic of China
c
Department of Mathematics, Xinyang Normal University, Xinyang 464000, Henan, People’s Republic of China

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, a predator–prey model with disease in the prey is constructed and investi-
Received 9 March 2008 gated for the purpose of integrated pest management. In the first part of the main results,
Received in revised form 28 May 2008 the sufficient condition for the global stability of the susceptible pest-eradication periodic
Accepted 5 June 2008
solution is obtained, which means if the release amount of infective prey and predator sat-
Available online 14 June 2008
isfy the condition, then the pest will be controlled. The sufficient condition for the perma-
nence of the system is also obtained subsequently, which means if the release amount of
infective prey and predator satisfy the condition, then the prey and the predator will coex-
Keywords:
Predator–prey model
ist. At last, we interpret our mathematical results.
Impulsive Ó 2008 Elsevier Inc. All rights reserved.
Permanent
Susceptible pest-eradication periodic
solution

1. Introduction

Pests outbreak often cause serious ecological and economic problems, and the warfare between human and pests (such as
locust, Aphis and cotton bollworm) has sustained for thousands of years. With the development of society and the progress
of science and technology, man has adopted some advanced and modern weapons such as chemical pesticides, biological
pesticides, remote sensing and measuring, and so on. Some brilliant achievements have been obtained. However, the warfare
is not over, and will continue. A great deal of and a large variety of pesticides were used to control pests, because they can
quickly kill a significant portion of a pest population and sometimes provide the only feasible method for preventing eco-
nomic loss. However, pesticide pollution is also recognized as a major health hazard to human beings and beneficial insects.
At present, more and more people are concerned about the effects of pesticide residues on human health and on the envi-
ronment. In this regard, it has been observed that beneficial insects are often more susceptible to chemical pesticides than
the target pests are. In the same time, the concentration of the pesticides in use tends to increase with time and usage, since
many pests develop resistance to these chemicals (see [1–3] and the references therein).
An alternative to chemical control is biological control, including microbial control with pathogens, as diseases can be
important natural controls of some pests. Insects, like humans and other animals, can be infected by disease-causing organ-
isms such as bacteria, viruses and fungi. Under appropriate conditions, such as high humidity or high pest abundance, these
naturally occurring organisms may multiply to cause disease outbreaks or epizootics that can decimate an insect population.
Insect pathogens are used in two ways. In the first method, a small amount of pathogen is introduced in a pest population

* Corresponding author. Address: Department of Applied Mathematics, Dalian University of Technology, Dalian, Liaoning 116024, People’s Republic of
China.
E-mail addresses: shirq1979@163.com (R. Shi), lschen@amss.ac.cn (L. Chen).

0307-904X/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.apm.2008.06.001
 R. Shi et al. / Applied Mathematical Modelling 33 (2009) 2248–2256 2249

with the expectation that it will generate an epidemic which will persist at an endemic level. In the second method, an insect
pathogen is used like biopesticides. In this case, the pathogen is applied whenever a pest population is at an economically
significant level and there is no expectation that the pathogen will persist in the environment for a long time. There is a vast
amount of literatures on the applications of entomopathogens to suppress pests [4–7]. One of the successful cases of biolog-
ical control in greenhouses was the use of the parasitoid Encarsia formosa against the greenhouse whitefly Trialeurodes
vaporariorum on tomatoes and cucumbers [8,9]. There are also papers and books on mathematical models of the dynamics
of microbial diseases in pest control [3,10–13].
People also use natural enemy to control pest or regulate its density below the threshold for economical damage. Often
with augmentation or release, the natural enemy is applied like a pesticide after the pest has reached or exceeded the eco-
nomic threshold. There are many literatures concerning natural enemy for pest control [8,14–16].
Integrated pest management (IPM) is a long term management strategy that uses a combination of biological, cultural,
and chemical tactics to reduce pests to tolerable levels, with little cost to the grower and minimal effect on the environment
(for more details, we can see [17,18]). It has been proved by experiment that IPM is more effective than classical one (such as
biological control or chemical control) [19].
Systems with impulsive effects describing evolution processes are characterized by the fact that at certain moments of
time they abruptly experience a change of state. Impulsive effects are studied in almost every domain of applied science.
Impulsive equations [20,21] have been recently used in population dynamics in relation to impulsive vaccination [22], pop-
ulation ecology [23,24], the chemotherapeutic treatment of disease [25], the theory of the chemostat [26]. In addition, mod-
els with two impulses have been studied by [27,28].
Motivated by [17–19,27,28], in this paper, we will consider the integrated pest management by using microbial control
with pathogens and releasing natural enemy together. That is, we release infective pests and natural enemy to suppress the
pests. The infective pests can be cultivated in the laboratory and the natural enemy can be migrated from other regions. Once
the susceptible pest meets with the infective pest, there is a chance to be infected. The infective pests have more possibility
to death due to the disease. We suppose the natural enemy catch the susceptible pest, but do not catch the infective pest. In
our model, we call the pest and the natural enemy as prey and predator, respectively. For mathematical simplicity, we make
several assumptions in the models. Details are shown in the next section.
The main purpose of this paper is to construct a predator–prey model with disease in the prey for pest management. The
paper is organized as follows: in Section 2, the model is constructed and the main biological assumptions are formulated. In
Section 3, some preliminary lemmas and definitions are stated. In Section 4, by using Floquet’s theory for impulsive differ-
ential equations, small-amplitude perturbation methods and comparison techniques, we investigate the global asymptotic
stability of the susceptible pest-eradication periodic solution and the conditions for the permanence of the system. Finally,
a brief discussion and some possible future work for pest management are provided in the last section.

2. Model formulation

In this paper, we study the following model for integrated pest management.
8   9
>
>
> S0 ðtÞ ¼ rSðtÞ 1  SðtÞþIðtÞ  bSðtÞIðtÞ  aSðtÞyðtÞ
1þxSðtÞ >
;>
>
>
K =
>
> t–ns; t–ðn þ lÞs;
>
> I0 ðtÞ ¼ bSðtÞIðtÞ  d1 IðtÞ; >
>
> >
;
>
> 0 daSðtÞyðtÞ
y ðtÞ ¼ 1þxSðtÞ  d2 yðtÞ;
>
>
>
> 9
>
< MSðtÞ ¼ 0; > =
MIðtÞ ¼ p; t ¼ ðn þ lÞs; n 2 Z þ ¼ f0; 1; 2; 3; . . .g; ð1Þ
>
> >
>
> ;
>
> MyðtÞ ¼ 0;
>
> 9
>
>
>
> MSðtÞ ¼ 0; > =
>
>
>
> MIðtÞ ¼ 0; t ¼ ðn þ 1Þs; n 2 Z þ :
>
> >
: ;
MyðtÞ ¼ q;
Here SðtÞ; IðtÞ represent the densities of the susceptible prey (pest) population and the infective prey (pest) population, respec-
tively. yðtÞ is the density of predator (natural enemy) population. The model is derived with the following assumptions.
ðH1 Þ There is a disease among the prey population, and the prey is divided into two classes, susceptible and infective. The
incidence rate is the classic bilinear bSðtÞIðtÞ, and b is the contact number per unit time for every infective prey with suscep-
tible prey.
ðH2 Þ The prey population grow with Logistic rate, r > 0 is the intrinsic growth rate, K > 0 is the carrying capacity. We sup-
pose that the infective preys cannot produce offsprings due to the disease; however, the infective preys still consume some
 
crop. Thus, the growth rate for the prey is rSðtÞ 1  SðtÞþIðtÞ
K
.
ðH3 Þ The predator only catch the susceptible prey, and the predation functional response is Holling type II. The parameters
a, x are positive constants, and d is the conversion rate from the prey to the predator.
ðH4 Þ d1 ; d2 is the death rate for the infective prey and the predator, respectively.
2250 R. Shi et al. / Applied Mathematical Modelling 33 (2009) 2248–2256

ðH5 Þ We release infective prey population periodically with releasing amount pðp > 0Þ at time t ¼ ðn þ lÞs, where n 2 Z þ ,
and 0 < l < 1. In addition, we release predator population periodically with releasing amount qð> 0Þ at time t ¼ ðn þ 1Þs,
n 2 Z þ . There are some examples in [29,30] about releasing natural enemy and infective pest.

3. Preliminary

We give some definitions, notations and lemmas which will be useful for stating and proving our main results. Let
Rþ ¼ ½0; 1Þ; R3þ ¼ fx ¼ ðx1 ; x2 ; x3 Þ 2 R3 : x1 ; x2 ; x3 > 0g.
Denote f ¼ ðf1 ; f2 ; f3 Þ the mapping defined by the right-hand side of system (1).
Let V : Rþ  R3 ! Rþ . Then V is said to belong to class V 0 if

(i) V is continuous in ðns; ðn þ 1Þs  R3 and for each x 2 R3þ ; n 2 Z þ ¼ f0; 1; 2; 3; . . .g and the limit
lim Vðt; yÞ ¼ Vðnsþ ; xÞ
ðt;yÞ!ðnsþ ;xÞ

exists and is finite.

(ii) V is locally Lipschitzian in x.

Definition 3.1. For V 2 V 0 and ðt; xÞ 2 ðns; ðn þ 1Þs  R3 , the upper right Dini derivative of Vðt; xÞ with respect to the
impulsive differential system (1) is defined as
1
Dþ Vðt; xÞ ¼ limþ sup ½Vðt þ h; x þ hf ðt; xÞÞ  Vðt; xÞ:
h!0 h

Definition 3.2. System (1) is said to be permanent if there exists a compact region D 2 intR3þ such that every solution of sys-
tem (1) with positive initial values will eventually enter and remain in region D.
The solution of system (1), denoted by XðtÞ ¼ ðSðtÞ; IðtÞ; yðtÞÞ : Rþ ! R3þ , is continuously differentiable on
ðns; ðn þ 1Þs  R3 ; n 2 Z þ and the limit Xðnsþ Þ ¼ limt!nsþ XðtÞ exists and is finite for n 2 Z þ . Obviously, the global existence
and the uniqueness of solution of system (1) are guaranteed by the smoothness properties of f (see [20,21] for details on
fundamental properties of impulsive systems). The proofs of the following lemmas are obvious.
Lemma 3.3. Suppose that XðtÞ is a solution of (1) with Xð0þ Þ P 0. Then XðtÞ P 0 for all t > 0. Further, if Xð0þ Þ > 0 then XðtÞ > 0
for all t > 0.

Lemma 3.4. see Ref. [21] Let V : Rþ  R3 ! R and V 2 V 0 . Assume that

(
Dþ Vðt; XÞ 6 gðt; Vðt; XÞÞ; t–ns;
þ
Vðt; Xðt ÞÞ 6 Wn ðVðt; XðtÞÞÞ; t ¼ ns;

where g : Rþ  Rþ ! R is continuous in ðns; ðn þ 1Þs  Rþ and for each v 2 R3þ ; n 2 Z þ

lim gðt; yÞ ¼ gðnsþ ; vÞ
ðt;yÞ!ðnsþ ;vÞ

exists and is finite. Wn : Rþ ! Rþ is nondecreasing. Let RðtÞ be the maximal solution of the scalar impulsive differential equation
8 0
< U ðtÞ ¼ gðt; UÞ;
> t–ns;
þ
Uðt Þ ¼ Wn ðUðtÞÞ; t ¼ ns;
>
:
Uð0þ Þ ¼ U 0 ;
defined on ½0; 1Þ. Then Vð0þ ; X 0 Þ 6 U 0 implies that Vðt; XðtÞÞ 6 RðtÞ; t P 0, where XðtÞ is any solution of system (1).

Lemma 3.5. There exists a positive constant M such that SðtÞ 6 M, IðtÞ 6 M; yðtÞ 6 M, for each solution ðSðtÞ; IðtÞ; yðtÞÞ of system
(1) with positive initial values, where t is large enough.

Proof. Define a function V such that

VðtÞ ¼ SðtÞ þ IðtÞ þ yðtÞ:
By simple computation, we see that when t–ns; t–ðn þ lÞs,

SðtÞ þ IðtÞ ð1  dÞayðtÞ rS2 ðtÞ

Dþ Vjð1Þ þ dV ¼ ðr þ dÞSðtÞ  rSðtÞ  ðd1  dÞIðtÞ  ðd2  dÞyðtÞ  6 ðr þ dÞSðtÞ  ;
K 1 þ xSðtÞ K
where d ¼ minfd1 ; d2 g. Since the right-hand side of the above inequality is a quadratic with negative quadratic coefficient, it
is bounded from above for all ðSðtÞ; IðtÞ; yðtÞÞ 2 R3þ . Hence there exists a positive constant k such that
 R. Shi et al. / Applied Mathematical Modelling 33 (2009) 2248–2256 2251

Dþ Vjð1Þ þ dV < k for t–ns; t–ðn þ lÞs:

From the fourth to ninth equations of system (1) we see that,

Vðnsþ Þ 6 VðnsÞ þ l;
where l ¼ maxfp; qg. According to Lemma 2.2 of Ref. [20], we derive
Z t
k leds
VðtÞ ¼ Vð0Þedt þ kedðtsÞ ds þ R0<ks<t ledðtksÞ ! þ ds ; as t ! 1:
0 d e 1
Consequently, by the definition of VðtÞ we obtain that each solution of (1) with positive initial values is uniformly ultimately
bounded above. This completes the proof. h

Lemma 3.6. System


u0 ðtÞ ¼ wuðtÞ; t–ns; n 2 Zþ ;
ð2Þ
DuðtÞ ¼ l; t ¼ ns; n 2 Zþ ;
has a positive periodic solution u ðtÞ, and for every solution uðtÞ of this system with positive initial value uð0þ Þ, juðtÞ  u ðtÞj ! 0
as t ! 1, where
lewðtnsÞ l
u ðtÞ ¼ u ð0þ Þ ¼ :
1  ews 1  ews

Proof. The proof is obvious, in fact, since the solution of (2) is

 l 
uðtÞ ¼ uð0þ Þ  ewt þ u ðtÞ; ns < t 6 ðn þ 1Þs: 
1 ews

4. Main results

When SðtÞ  0 for all t P 0, we get the following subsystem of system (1):
8 0 
>
> I ðtÞ ¼ d1 IðtÞ;
>
> t–ns; t–ðn þ lÞs;
>
> y0 ðtÞ ¼ d2 yðtÞ;
>
> 
< MIðtÞ ¼ p;
t ¼ ðn þ lÞs; n 2 Z þ ; ð3Þ
>
> MyðtÞ ¼ 0;
>
> 
>
> MIðtÞ ¼ 0;
>
>
: t ¼ ðn þ 1Þs; n 2 Z þ :
MyðtÞ ¼ q;
In this system, we can see there is no relation between IðtÞ and yðtÞ. Thus, we can solve them independently. By Lemma 3.6,
we get the following result.
Theorem 4.1. System (3) has a unique positive periodic solution
(
I ed1 ðtnsÞ ; t 2 ðns; ðn þ lÞs;
I ðtÞ ¼
ðI ed1 s þ pÞed1 ðtðnþlÞsÞ ; t 2 ððn þ lÞs; ðn þ 1Þs;
d2 ðtnsÞ
y ðtÞ ¼ qe1ed2 s ; for t 2 ðns; ðn þ 1Þs;

where
: p : q
I ð0þ Þ ¼ I ¼ ; y ð0þ Þ ¼ y ¼ :
1  ed1 s 1  ed2 s
In addition, for every solution of system (3) with initial values Ið0þ Þ > 0; yð0þ Þ > 0, it follows that IðtÞ ! I ðtÞ; yðtÞ ! y ðtÞ as
t ! 1.
Thus, the complete expression for the susceptible pest-eradication periodic solution of system (1) is obtained as
ð0; I ðtÞ; y ðtÞÞ; t 2 ðns; ðn þ 1Þs; n 2 Z þ . The following theorems are results about the stability and attraction of the suscep-
tible pest-eradication periodic solution ð0; I ðtÞ; y ðtÞÞ.
 
Theorem 4.2. If r s < Kr þ b dp1 þ aq d2
, then the periodic solution ð0; I ðtÞ; y ðtÞÞ is locally asymptotically stable for system (1).

Proof. To prove the local stability of this periodic solution, we use small amplitude perturbation methods. Let
SðtÞ ¼ uðtÞ; IðtÞ ¼ vðtÞ þ I ðtÞ; yðtÞ ¼ wðtÞ þ y ðtÞ;
2252 R. Shi et al. / Applied Mathematical Modelling 33 (2009) 2248–2256

where uðtÞ; vðtÞ; wðtÞ are small perturbations. Then system (1) can be linearized by using Taylor expansions and after
neglecting higher-order terms, the linearized equations read as
8 r  9
>
> u0 ðtÞ ¼ uðtÞ½r  þ b I ðtÞ  ay ðtÞ; >
=
>
> K
>
> v0 ðtÞ ¼ bI ðtÞuðtÞ  d1 vðtÞ; t–ns; t–ðn þ lÞs;
>
> >
>
> ;
>
> 0 
w ðtÞ ¼ day ðtÞuðtÞ  d2 wðtÞ
>
> 9
>
>
>
< uððn þ lÞsþ Þ ¼ uððn þ lÞsÞ; >
=
vððn þ lÞsþ Þ ¼ vððn þ lÞsÞ; t ¼ ðn þ lÞs; n 2 Zþ ; ð4Þ
>
> >
;
>
> wððn þ lÞsþ Þ ¼ wððn þ lÞsÞ;
>
> 9
>
>
>
>
> uððn þ 1Þsþ Þ ¼ uððn þ 1ÞsÞ; >=
>
>
>
> vððn þ 1Þsþ Þ ¼ vððn þ 1ÞsÞ; t ¼ ðn þ 1Þs; n 2 Zþ :
>
> >
;
: wððn þ 1Þsþ Þ ¼ wððn þ 1ÞsÞ;

Let UðtÞ be the fundamental solution matrix of system (4). Then UðtÞ must satisfy
0 r  1
r K
þ b I ðtÞ  ay ðtÞ 0 0
dUðtÞ B C
¼@ bI ðtÞ d1 0 AUðtÞ;
dt
day ðtÞ 0 d2
Uð0Þ ¼ I3 is the identical matrix. Hence the fundamental solution matrix is
0 Rt 1
½rðKr þbÞI ðtÞay ðtÞdt
Be 0 0 0 C
UðtÞ ¼ B @ u21 ed1 t 0 A;
C
d2 t
u31 0 e
where the exact expression of u21 and u31 are omitted, since they are not used subsequently. The fourth, fifth and sixth
equations in system (4) read as
0 1 0 10 1
uððn þ lÞsþ Þ 1 0 0 uððn þ lÞsÞ
B C B C B C
@ vððn þ lÞsþ Þ A ¼ @ 0 1 0 A@ vððn þ lÞsÞ A;
wððn þ lÞsþ Þ 0 0 1 wððn þ lÞsÞ
and the seventh, eighth and ninth equations in (4) read as
0 1 0 10 1
uððn þ 1Þsþ Þ 1 0 0 uððn þ 1ÞsÞ
B C B CB C
@ vððn þ 1Þsþ Þ A ¼ @ 0 1 0 A@ vððn þ 1ÞsÞ A:
wððn þ 1Þs Þ
þ
0 0 1 wððn þ 1ÞsÞ
Hence, if all eigenvalues of
0 10 1
1 0 0 1 0 0
B CB C
M ¼ @ 0 1 0 A@ 0 1 0 AUðsÞ
0 0 1 0 0 1
have absolute values less than one, then the periodic solution ð0; I ðtÞ; y ðtÞÞ is locally asymptotically stable. Since the eigen-
values of M are
Rs
½rðKr þbÞI ðtÞay ðtÞdt
k1 ¼ ed2 s < 1; k2 ¼ ed1 s < 1; k3 ¼ e 0 ;
 r
it follows that jk3 j < 1 if and only if rs < K þ b dp1 þ daq2 holds. According to the Floquet theory of impulsive differential equa-
tions, in this situation, the susceptible pest-eradication periodic solution ð0; I ðtÞ; y ðtÞÞ is locally asymptotically stable. The
proof is complete. h
r  
Theorem 4.3. If rs < K
þ b dp1 þ d2 ð1þ
aq 
xKÞ, then the periodic solution ð0; I ðtÞ; y ðtÞÞ is globally asymptotically stable for system (1).

Proof. By the given condition and Theorem 4.2, it is easy to know that ð0; I ðtÞ; y ðtÞÞ is locally asymptotically stable. There-
 
fore, we only need to prove its global attraction. Since rs < Kr þ b dp1 þ d2 ð1þ
aq
xKÞ, we can choose a e1 > 0 small enough such that
Z s r 
a :
r þ b ðI ðtÞ  e1 Þ  ðy ðtÞ  e1 Þ dt ¼ r < 0:
0 K 1 þ xK
Besides, we have
I0 ðtÞ ¼ bSðtÞIðtÞ  d1 IðtÞ P d1 IðtÞ:
 R. Shi et al. / Applied Mathematical Modelling 33 (2009) 2248–2256 2253

From Lemmas 3.4 and 3.6, there exists a n1 2 Z þ such that

IðtÞ P I ðtÞ  e1 ; for t P n1 s: ð5Þ
Similarly, there exists a n2 2 Z þ ðn2 > n1 Þ such that
yðtÞ P y ðtÞ  e1 ; for t P n2 s: ð6Þ
Thus, for t P n2 s, we have

 r 
SðtÞ þ IðtÞ aSðtÞyðtÞ aSðtÞyðtÞ
S0 ðtÞ ¼ rSðtÞ 1   bSðtÞIðtÞ  6 rSðtÞ  þ b SðtÞIðtÞ 
K 1 þ xSðtÞ K 1 þ xK
 r 
 aðy ðtÞ  e1 Þ
6 SðtÞ r  þ b ðI ðtÞ  e1 Þ  :
K 1 þ xK
From the above inequality, we get
Rt  r aðy ðtÞe Þ

rðK þbÞðI ðtÞe1 Þ 1þxK 1 dt
SðtÞ 6 Sðn2 sÞe n2 s 6 Sðn2 sÞekr ;
where t 2 ððn2 þ kÞs; ðn2 þ k þ 1Þ
s; k 2 Z þ .nSinceor< 0, we can easily see that SðtÞ ! 0 as k ! þ1. Thus, for an arbitrary po-
sitive constant e2 small enough e2 < min db1 ; dda2 , there exists a n3 2 Z þ ðn3 > n2 Þ such that SðtÞ < e2 for all t P n3 s. From
which we get
I0 ðtÞ ¼ bSðtÞIðtÞ  d1 IðtÞ 6 ðbe2  d1 ÞIðtÞ:
From Lemmas 3.4 and 3.6, there exists a n4 2 Z þ ðn4 > n3 Þ such that
IðtÞ 6 I1 ðtÞ þ e1 ; for t P n4 s; ð7Þ
where
(
I1 eðd1 be2 ÞðtnsÞ ; t 2 ðns; ðn þ lÞs;
I1 ðtÞ ¼
ðI1 ed1 s þ pÞeðd1 be2 ÞðtðnþlÞsÞ ; t 2 ððn þ lÞs; ðn þ 1Þs:

By similarly argument, there exists a n5 2 Z þ ðn5 > n4 Þ such that

yðtÞ 6 y1 ðtÞ þ e1 ; for t P n5 s; ð8Þ
ðd2 dae2 ÞðtksÞ
where y1 ðtÞ ¼ qe1eðd2 dae2 Þs ,
for t 2 ðks; ðk þ 1Þs; k 2 Z þ . Note that e1 ; e2 are positive constants small enough, and
I1 ðtÞ ! 
e2 ! 0. Together with equations(5)–(8), we get IðtÞ ! I ðtÞ and yðtÞ ! y ðtÞ as t ! þ1. There-
I ðtÞ; y1 ðtÞ ! y ðtÞ, as
fore, the periodic solution ð0; I ðtÞ; y ðtÞÞ is globally asymptotically stable. h

Corollary 4.4

(1) If p ¼ 0, then the condition of Theorem 4.3 becomes

r sd2 ð1 þ xKÞ :
q> ¼ q0 ; ð9Þ
a
which means that if only natural enemies are released periodically, then the release amount must larger than q0 to ensure
the eradication of the pest.
(2) If q ¼ 0, then the condition of Theorem 4.3 becomes
rsd1 :
p> ¼ p0 ; ð10Þ
b þ Kr
which means that if only infective pests are released periodically, then the release amount must larger than p0 to ensure the
eradication of the pest.

r  p
Theorem 4.5. If r s > K
þb d1
þ daq2 , then system (1) is permanent.

Proof. We will prove the theorem by several steps. By Lemma 3.5, without loss of generality, we suppose SðtÞ; IðtÞ; yðtÞ 6 M
for all t P 0, for simplifying the next proof process.

Step 1. We will prove IðtÞ and yðtÞ are ultimately positively bounded below.
:
Firstly, from Eq. (5), we get IðtÞ P I ðtÞ  e1 P I ed1 ls  e1 ¼ m1 > 0, for t P n1 s.
:
Secondly, from Eq. (6), we get yðtÞ P y ðtÞ  e1 P y ed2 s  e1 ¼ m2 > 0, for t P n2 s.
Therefore, IðtÞ and yðtÞ are ultimately positively bounded below.
2254 R. Shi et al. / Applied Mathematical Modelling 33 (2009) 2248–2256

Step 2. We will prove SðtÞ is ultimately positively bounded below.

 
Since r s > Kr þ b dp1 þ daq2 , we can select positive constant e and m3 small enough such that
h r  i  
m3 r p q :
rð1  Þ þ b e  ae s  þb þa ¼ r2 > 0:
K K K d1  bm3 d2  dam3
We claim that for an arbitrary N 1 2 Z þ ; SðtÞ < m3 cannot hold for all t P N 1 s. Otherwise, there exists a N 1 2 Z þ , such
that SðtÞ < m3 for all t P N 1 s. Then, we have
I0 ðtÞ ¼ bSðtÞIðtÞ  d1 IðtÞ 6 ðbm3  d1 ÞIðtÞ;
daSðtÞyðtÞ
y0 ðtÞ ¼  d2 yðtÞ 6 ðdam3  d2 ÞyðtÞ;
1 þ xSðtÞ
for all t P N 1 s. By Lemmas 3.4 and 3.6, we know there exists a N 2 2 Z þ ðN 2 > N 1 Þ such that
IðtÞ 6 I2 ðtÞ þ e; yðtÞ 6 y2 ðtÞ þ e;
for all t P N 2 s, where
(
I2 eðd1 bm3 ÞðtnsÞ ; t 2 ðns; ðn þ lÞs;
I2 ðtÞ ¼
ðI2 ed1 s þ pÞeðd1 bm3 ÞðtðnþlÞsÞ ; t 2 ððn þ lÞs; ðn þ 1Þs;
y2 ðtÞ ¼ y2 eðd2 dam3 ÞðtnsÞ
peðd1 bm3 Þð1lÞs q
I2 ¼ ; y2 ¼ :
1  eðd1 bm3 Þs 1  eðd2 dam3 Þs
are the unique positive periodic solutions of systems (11) and (12)
8 0
< I ðtÞ ¼ ðbm3  d1 ÞIðtÞ; t–ðn þ lÞs; t–ns;
>
MIðtÞ ¼ p; t ¼ ðn þ lÞs; ð11Þ
>
:
MIðtÞ ¼ 0; t ¼ ðn þ 1Þs;
8 0
< y ðtÞ ¼ ðdam3  d2 ÞyðtÞ; t–ðn þ lÞs; t–ns;
>
MyðtÞ ¼ 0; t ¼ ðn þ lÞs; ð12Þ
>
:
MyðtÞ ¼ q; t ¼ ðn þ 1Þs;
respectively. From which, we get

 h
SðtÞ þ IðtÞ aSðtÞyðtÞ rm3  r  i
S0 ðtÞ ¼ rSðtÞ 1   bSðtÞIðtÞ  P SðtÞ r   þ b ðI2 ðtÞ þ eÞ  aðy2 ðtÞ þ eÞ :
K 1 þ xSðtÞ K K
Thus,
R ðN2 þkÞs rm3 r
½r K ðK þbÞðI2 ðtÞþeÞaðy2 ðtÞþeÞdt
SððN2 þ kÞsÞ P SðN 2 sÞe N2 s P SðN 2 sÞekr2 :
We easily get SððN 2 þ kÞsÞ ! þ1 as t ! þ1, since r2 > 0. This is a contradiction with Lemma 3.5. Thus our claim is
true, and for an arbitrary N 1 2 Z þ , there exists at least a t 1 P N 1 s such that Sðt 1 Þ P m3 . There are two cases:
Case 1. SðtÞ P m3 for all t P t 1 . Then our aim is obtained. Otherwise, we consider the next case.
Case 2. We consider those solutions which leave the region C ¼ fðSðtÞ; IðtÞ; yðtÞÞ 2 R3þ : SðtÞ < m3 g and reenter it again. Let
t ¼ inf tPt1 fSðtÞ < m3 g. Then SðtÞ P m3 for t 2 ½t1 ; t  Þ and Sðt Þ ¼ m3 , since SðtÞ is continuous. Suppose
t 2 ðN 3 s; ðN 3 þ 1Þs; N 3 2 Z þ . Select N 4 ; N 5 2 Z þ such that
 e e 
 ln 2M  ln 2M
N4 s > max ; ;
d1  bm3 d2  dam3
N5 r2 > ðN 4 þ 1Þqs;
r 
q ¼ þ b þ a M:
K
Denote T ¼ ðN 4 þ N 5 Þs. We claim that SðtÞ < m3 cannot hold for all t 2 ½ðN 3 þ 1Þs; ðN 3 þ 1Þs þ T. Otherwise,
SðtÞ < m3 for all t 2 ½ðN 3 þ 1Þs; ðN 3 þ 1Þs þ T. Then
I0 ðtÞ ¼ bSðtÞIðtÞ  d1 IðtÞ 6 ðbm3  d1 ÞIðtÞ
holds for all t 2 ½ðN 3 þ 1Þs; ðN 3 þ 1Þs þ T. Consider the following system:

u0 ðtÞ ¼ ðbm3  d1 ÞuðtÞ; t–ns; n ¼ N 3 þ 1; N3 þ 2; . . . ;
ð13Þ
uðnsþ Þ ¼ uðns Þ þ p; t ¼ ns; n ¼ N 3 þ 1; N3 þ 2; . . . ;
with initial value uððN 3 þ 1Þsþ Þ ¼ IððN 3 þ 1Þsþ Þ. Obviously, the solution of system (13) is
 R. Shi et al. / Applied Mathematical Modelling 33 (2009) 2248–2256 2255

uðtÞ ¼ I2 ðtÞ þ ðuððN 3 þ 1Þsþ Þ  I2 Þeðd1 bm3 ÞðtðN3 þ1ÞsÞ :
And

juðtÞ  I2 ðtÞj 6 2Meðd1 bm3 ÞðtðN3 þ1ÞsÞ < e;

for all t 2 ½ðN 3 þ N 4 þ 1Þs; ðN 3 þ 1Þs þ T. So, by Lemma 3.4, we get
IðtÞ 6 uðtÞ 6 I2 ðtÞ þ e; ð14Þ
for all t 2 ½ðN 3 þ N 4 þ 1Þs; ðN 3 þ 1Þs þ T. Similarly, we have
yðtÞ 6 y2 ðtÞ þ e; ð15Þ
for all t 2 ½ðN 3 þ N 4 þ 1Þs; ðN 3 þ 1Þs þ T. Thus, we have

 h
SðtÞ þ IðtÞ aSðtÞyðtÞ rm3  r  i
S0 ðtÞ ¼ rSðtÞ 1   bSðtÞIðtÞ  P SðtÞ r   þ b ðI2 ðtÞ þ eÞ  aðy2 ðtÞ þ eÞ :
K 1 þ xSðtÞ K K
for all t 2 ½ðN 3 þ N 4 þ 1Þs; ðN 3 þ 1Þs þ T. And
R ðN3 þ1ÞsþT rm3
½r ð
 Kr þb ÞðI2 ðtÞþeÞaðy2 ðtÞþeÞdt
SððN3 þ 1Þs þ TÞ P SððN3 þ N4 þ 1ÞsÞe ðN 3 þN4 þ1Þs K
P SððN3 þ N4 þ 1ÞsÞeN5 r2 : ð16Þ
In the interval t 2 ½N 3 s; ðN 3 þ N 4 þ 1Þs, we have

 hr i
SðtÞ þ IðtÞ aSðtÞyðtÞ
S0 ðtÞ ¼ rSðtÞ 1   bSðtÞIðtÞ  P  þ b þ a MSðtÞ ¼ qSðtÞ ð17Þ
K 1 þ xSðtÞ K
and
R ðN3 þN4 þ1Þs
ðqÞdt
SððN3 þ N4 þ 1ÞsÞ P SðN 3 sÞe N3 s P SðN 3 sÞeðN4 þ1Þsq : ð18Þ
From Eqs. (16) and (18), we get

SððN3 þ 1Þs þ TÞ P SðN3 sÞeðN4 þ1Þsq eN5 r2 P m3 eN5 r2 ðN4 þ1Þsq :

By N 5 r2 > ðN 4 þ 1Þqs, we get SððN 3 þ 1Þs þ TÞ P m3 , which is a contradiction. Thus, there exists at least a t2 2
:
½ðN 3 þ 1Þs; ðN 3 þ 1Þs þ T such that Sðt 2 Þ P m3 . So, for t 2 ½t 1 ; t 2 ; SðtÞ P m3 eqðt2 N3 sÞ ¼ m4 . For t > t2 , the same argu-
ments can be continued since Sðt2 Þ P m3 . This proves that SðtÞ is ultimately positively bounded below.

Step 3. Denote m ¼ minfm1 ; m2 ; m4 g; D ¼ fR3þ : m 6 SðtÞ; IðtÞ; yðtÞ 6 Mg. Combining step 1, step 2 and Lemma 3.5, we know
that every solution of system (1) with positive initial values will eventually enter and remain in region D. By Def-
inition 3.2, we know system (1) is permanent. The proof is complete. h

Corollary 4.6

(1) If p ¼ 0, then the condition of Theorem 4.5 becomes

r sd2 :
q< ¼ q1 ; ð19Þ
a
which means that if only natural enemies are released periodically, and the release amount less than q1 , then the system is
permanent and the pest will not eradicate.
(2) If q ¼ 0, then the condition of Theorem 4.5 becomes
rsd1
p< ¼ p0 ; ð20Þ
b þ Kr
which means that if only infective pests are released periodically, and the release amount less than p0 , then the system is
permanent and the pest will not eradicate.

5. Discussion

In this paper, a predator–prey model with disease in the prey is investigated for the purpose of integrated pest manage-
ment. In Theorem 4.3, the sufficient condition for the global stability of the susceptible pest-eradication periodic solution is
obtained, which means that if the release amount of infective pest and natural enemy are large than some critical values,
then the susceptible pest will be doomed. By the result of Theorem 4.5, the sufficient condition for the permanence of system
2256 R. Shi et al. / Applied Mathematical Modelling 33 (2009) 2248–2256

(1) is also obtained, which means that the pest and the natural enemy will coexist for all time, if the release amount of infec-
tive pest and natural enemy are less than some critical values. Corollary 4.4 shows that if only one measure is taken, i.e.,
either only infective pests are released or only natural enemies are released, then the release amount must satisfy Eqs.
(9) or (10). Obviously, our results shows that integrated pest management strategy is superior to those where only infective
pests are released or only natural enemies are released. Therefore, our results present a more prior strategy for pest man-
agement. In the model of this paper, we suppose infective pests and natural enemies are released at different times. In an-
other paper [31], the same model with one impulse is considered, i.e., infective pests and natural enemies are released at the
same moment. An interesting result is that we get similar results with that paper, which is not the case from intuition. In fact,
from Theorems 4.3 and 4.5, we see that the conditions for the global asymptotical stability of susceptible pest-eradication
periodic solution and the permanence of system (1) have no relation with the parameter l, which means that we can freely
choose l ð0 < l < 1Þ. This result is important, since we cannot always release infective pests and natural enemies at the same
time. In practice, infective pests and natural enemies may be released at the same time or at different time. Sometimes, infec-
tive pests are released firstly and then predators are released; while at other time, predators are released firstly. Theorem 4.3
guarantees that the pests will be controlled no matter infective pests are released firstly or predators are released firstly.
Thus, our results are interesting and useful.

Acknowledgement

This work is partly supported by the National Natural Science Foundation of China (10471117).

References

[1] F.L. McEwen, R.G. Stephenson, The Use and Significance of Pesticides in the Environment, Wiley, New York, 1979.
[2] W. Phelps, Pesticide Environmental Fate: Bridging the Gap between Laboratory and Field Studies, American Chemical Society, USA, 2002.
[3] Zhang Hong, Jiao Jianjun, Chen Lansun, Pest management through continuous and impulsive control strategies, BioSystems 90 (2007) 350C361.
[4] H.D. Burges, N.W. Hussey, Microbial Control of Insects and Mites, Academic Press, New York, 1971.
[5] P.E. Davis, K. Myers, J.B. Hoy, Biological control among vertebrates, in: C.B. Huffaker, P.S. Messenger (Eds.), Theory and Practice of Biological Control,
Plenum Press, New York, 1976.
[6] L.A. Falcon, Problem associated with the use of arthropod viruses in pest control, Annu. Rev. Entomol. 21 (1976) 305–324.
[7] Y. Tanada, Epizootiology of insect disease, in: P. DeBach (Ed.), Biological Control of Insect Pests and Weeds, Chapman and Hall, London, 1964.
[8] J.C. Van Lenteren, Measures of success in biological control of anthropoids by augmentation of natural enemies, in: S. Wratten, G. Gurr (Eds.), Measures
of Success in Biological Control, Kluwer Academic Publishers, Dordrdcht, 2000, pp. 77–89.
[9] J.C. Van Lenteren, J. Woets, Biological and integrated pest control in greenhouses, Ann. Rev. Ent. (1988) 239–250.
[10] B.S. Goh, Management and Analysis of Biological Populations, Elsevier Scientific Publishing Company, Amsterdam–Oxford–New York, 1980.
[11] J. Grasman, O.A. Van Herwarrden, L. Hemerik, et al, A two-component model of host-parasitoid interactions: determination of the size of inundative
releases of parasitoids in biological pest control, Math. Biosci. 196 (2001) 207–216.
[12] Zhang Hong, Chen Lansun, J. Nieto Juan, A delayed epidemic model with stage-structure and pulses for pest management strategy, Nonlin. Anal. : Real
World Appl. (2007), doi:10.1016/j.nonrwa.2007.05.004.
[13] Jiao Jianjun, Meng Xinzhu, Chen Lansun, Global attractivity and permanence of a stage-structured pest management SI model with time delay and
diseased pest impulsive transmission, Chaos, Solitons & Fractals (2007), doi:10.1016/j.chaos.2007.01.003.
[14] P. Debach, D. Rosen, Biological Control by Natural Enemies, second ed., Cambridge University Press, Cambridge, 1991.
[15] H.J. Freedman, Graphical stability, enrichment, and pest control by a natural enemy, Math. Biosci. 31 (1976) 207–225.
[16] M.L. Luff, The potential of predators for pest control, Agric. Ecosyst. Environ. 10 (1983) 159–181.
[17] A.J. Cherry, C.J. Lomer, D. Djegui, F. Schulthess, Pathogen incidence and their potential as microbial control agents in IPM of maize stemborers in West
Africa, Biocontrol 44 (a) (1999) 301–327.
[18] L.F. Mary, V.B. Robert, Introduction to Intergrated Pest Management, Plenum Press, New York and London, 1981.
[19] J.C. Van Lenteren, Integrated pest management in protected crops, in: D. Dent (Ed.), Integrated Pest Management, Chapman and Hall, London, 1995.
[20] D.D. Bainov, P.S. Simeonov, Impulsive Differential Equations: Periodic Solutions and Applications, Longman, London, 1993.
[21] V. Lakshmikantham, D.D. Bainov, P.S. Simeonov, Theory of Impulsive Differential Equations, World Scientific, Singapore, 1989.
[22] J. Hui, L.S. Chen, Impulsive vaccination of SIR epidemic models with nonlinear incidence rates, Discrete Continuous Dyn. Syst. Ser. B 3 (2004) 595–606.
[23] X.N. Liu, L.S. Chen, Complex dynamics of Holling type II Lotka–Volterra predator–prey system with impulsive perturbations on the predator, Chaos,
Solitons & Fractals 16 (2003) 311–320.
[24] B. Liu, L.S. Chen, Y.J. Zhang, The dynamics of a prey-dependent consumption model concerning impulsive control strategy, Appl. Math. Comp. 169
(2005) 305–320.
[25] A. Lakmeche, O. Arino, Bifurcation of non trivial periodic solutions of impulsive differential equations arising chemotherapeutic treatment, Dyn.
Continuous Discrete Impulsive Syst. 7 (2000) 265–287.
[26] E. Funasaki, M. Kot, Invasion and chaos in a periodically pulsed mass-action chemostat, Theor. Pop. Biol. 44 (1993) 203–224.
[27] S.J. Gao, L.S. Chen, L.H. Sun, Optimal pulse fishing policy in stage-structured models with birth pulses, Chaos, Solitons & Fractals 25 (2005) 1209–1219.
[28] B. Liu, Z.D. Teng, L.S. Chen, The effect of impulsive spraying pestiside on stage-structured population models with birth pulse, J. Biol. Syst. 13 (1) (2005)
31–44.
[29] P.E. Davis, K. Myers, J.B. Hoy, Biological control among vertebrates, in: C.B. Huffaker, P.S. Messenger (Eds.), Theory and Practice of Biological Control,
Plenum Press, New York, 1976.
[30] Zhang Wei, Multiply methods for pest control (in chinese) <http://biological.aweb.com.cn/news/2007/9/12/10495060.shtml,2007-09-12>.
[31] Shi Ruiqing, Chen lansun, An impulsive predator–prey model with disease in the prey for integrated pest management, submitted for publication.