You are on page 1of 26

CANADIAN APPLIED

MATHEMATICS QUARTERLY
Volume 19, Number 4, Winter 2011

BIFURCATIONS IN A RATIO-DEPENDENT
PREDATOR-PREY MODEL WITH PREY
HARVESTING

Dedicated to Prof. H. I. Freedman on the occasion of his 70th birthday

LILI CHEN, YILONG LI AND DONGMEI XIAO

ABSTRACT. The dynamics and bifurcations of a class of


ratio-dependent predator-prey models with prey harvesting are
investigated. The existence of all ecological feasible equilib-
ria for this model is determined and the topological classifi-
cations of these equilibria are derived. It is proved that the
model can undergo Hopf bifurcation and Bogdanov-Takens bi-
furcation near the corresponding positive equilibrium as some
parameters of the model vary, and unstable oscillation can be
observed. Some numerical simulations are provided to support
our theoretical results. These theoretical conclusions reveal the
effect of constant harvesting rate on the coexistence of the two
species, which not only provides predication whether the two
species will suffer from extinction one after the other but also
gets insight into the optimal management of exploitation re-
sources

1 Introduction Predator-prey mathematical models for interac-


tions between the prey biomass and predator population in theoretical
ecology is a very good tool in studying the dynamic processes of species
and in making predictions whether the two species will suffer from mu-
tual extinction. A key element in the predator-prey models is the func-
tional response, which describes the number of prey consumed by preda-
tors per unit time. In classical predator-prey models, it is considered as a
function of prey abundance only. The most common functional response
Research was supported by the National Natural Science Foundations of China
(No.10831003 & No.10925102) and by Program of Shanghai Subject Chief Scientist
(No.10XD1406200).
Keywords: Ratio-dependent predator-prey, prey harvesting, Hopf bifurcation,
Bogdanov-Takens bifurcations, limit cycle.
c
Copyright Applied Mathematics Institute, University of Alberta.

293
294 L. CHEN, Y. LI AND D. XIAO

is the Michaelis-Menten or Holling type II function of the form


bx
(1.1) p(x) = ,
1 + Ax
where b/A > 0 is the maximal growth rate of the predators, and 1/A > 0
is the half-saturation constant. Many papers have been published to
get a better understanding of dynamics of the classical prey dependent
models. An introduction to the prey dependent predator prey systems
and the relative research is given in the excellent book by Freedman [14].
On the other hand, there are growing explicit biological evidence that
functional responses over typical ecological timescales ought to depend
on the densities of both prey and predators, especially when predators
have to search for food and therefore have to share or compete for food;
see [1, 3, 4, 5, 6]. Such a functional response is called a ratio-dependent
response function. Based on the Michaelis-Menten or Holling type II
function, Arditi and Ginzburg [5] proposed a ratio-dependent response
function of the form
 
x bx
(1.2) p = .
y y + Ax

It has been found that the ratio-dependent predator-prey models are


more appropriate for predator-prey interactions when the predators in-
volve serious hunting processes. Hence, a significant number of predator-
prey models with ratio-dependent response function have been studied
extensively in the last few decades, and very rich dynamics have been
observed; see [2, 15, 16, 18–23, 25, 30, 31, 34] and references therein.
Recently Haque [17] modified the prey-dependent functional response
(1.1) in the classical Bazykin’s model to the ratio-dependent functional
response (1.2) and obtained the following system

dx bxy

 = ax − − ex2 ,
dt y + Ax





(1.3) dy dxy
= −cy + − f y2 ,



 dt y + Ax

x(0) = x0 > 0, y(0) = y0 > 0,

where x and y are scaled prey and predator population densities, re-
spectively. Parameter a > 0 is the prey natural growth rate, b/A > 0 is
the prey maximal consumption rate by predators, d/A > 0 is the max-
imal growth rate of predators, 1/A > 0 is the half-saturation constant
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 295

of predators, e > 0 and f > 0 are the prey and predator intraspecies
competition rates, respectively, and c > 0 is the predator natural mortal-
ity rate. By qualitative analysis, Haque [17] obtained some interesting
dynamics of system (1.3), such as paradox of enrichment, stability and
permanence, etc.
However, both harvesting and predation are processes in which mem-
bers of a population are removed by an external agency, sometimes for
population management, but more often for the benefit of the harvester
from the point of view of human needs. Hence, the exploitation of
biological resources and the harvesting of populations are commonly
practiced in fishery, forestry, and wildlife management, which is related
to the optimal management of renewable resources [12]. There have
been some works on optimal exploitation of the harvested stock in or-
der to maximize the profit (see [10, 28]), and a considerable amount
of research on dynamics of predator-prey system with harvesting; see
[7, 13, 24, 29, 32, 33] and references therein.
In this paper, we consider that the prey population is subjected to
harvesting at a constant rate in model (1.3) as follows:

dx bxy


 = ax − − ex2 − H,



 dt y + Ax
(1.4) dy dxy
= −cy + − f y2 ,



 dt y + Ax

x(0) = x0 > 0, y(0) = y0 > 0.

where H > 0 is a constant harvesting rate.


For simplicity, we re-scale the state and time variables of model (1.4)
as follows:
e  
be
t̄ = at, x̄ = x, ȳ = y.
a ad

Dropping the bars, we have the following system


 ǫxy 2
ẋ = x − αx + y − x − h,


(1.5)
ǫxy
ẏ = −γy + − δy 2 ,


αx + y

where γ = c/a, α = Ab/d, ε = b/a, δ = f d/be, and h = He/a2 .


296 L. CHEN, Y. LI AND D. XIAO

Since system (1.5) is not well-defined at the origin (0, 0), we redefine
system (1.5) as
 ǫxy
 ẋ = x − − x2 − h,


 αx +y
ǫxy

(1.6) ẏ = −γy + − δy 2 ,


 αx +y

ẋ = −h, ẏ = 0, when (x, y) = (0, 0),

where γ, α, ε, δ and h are positive parameters.


From the point of view of biology, we only restrict our attention to
system (1.6) in the closed first quadrant of the (x, y) plane, denoted by
R2+ . However, R2+ is not invariant for system (1.6). The solution curves
of system (1.6) can cross the positive y-axis out of R2+ . Based on the
ecological meaning, we adopt the following convention.

(H1): x(t) ≡ 0 for all t ≥ t0 if there exists a positive time t0 such that
the solution (x(t), y(t)) of system (1.6) satisfies x(t0 ) = 0.

The objective of this paper is to study the dynamics of system (1.6) in


R2+ . From the point of view of the optimal management of renewable
resources, we focus on how much we can harvest. There exists a maxi-
mum sustainable yield hM SY = 1/4 of system (1.6) such that the prey
population goes to extinction if h > hM SY , which leads to the collapse of
the predator-prey system. Biologically, over-harvesting of prey species
occurs for model (1.6) if h > hM SY . Hence, the case 0 < h < 1/4 is
of interest for the study on dynamics of system (1.6) in R2+ . By the
qualitative and bifurcation analysis, we investigative the existence of all
ecological feasible equilibria for system (1.6) and the topological classi-
fications of these equilibria even though it is algebraically impossible to
find expressions for some equilibria in terms of the parameters. We prove
that the model can undergo Hopf bifurcation and Bogdanov-Takens bi-
furcation near the corresponding positive equilibrium as some param-
eters of the model vary and provide numerical simulations to support
our theoretical results. These conclusions reveal the effect of constant
harvesting rate on the coexistence of the two species, which not only
provides predication whether the two species will suffer from extinction
one after the other but also gets insight into the optimal management
of exploitation resources.
The organization of this paper is as follows. In Section 2, we discuss
the existence of all ecological feasible equilibria for system (1.6), give the
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 297

the topological classifications of these equilibria, and obtain the saddle-


node bifurcation curves. In Section 3, we study the existence of a positive
cusp equilibrium for system (1.6) and prove that system (1.6) undergoes
Bogdanov-Takens bifurcation around this equilibrium. In Section 4, we
study Hopf bifurcation of system (1.6) and obtain an unstable limit
cycle. A brief discussion is given in the final section.

2 The existence of equilibria and their topological classifica-


tions In this section, we study the existence of all ecological feasible
equilibria and topological property of these equilibria of system (1.6) in
R2+ .
Setting the right-hand sides of system (1.6) to zero, we have the fol-
lowing equations
ǫxy
x− − x2 − h = 0,
αx + y
(2.1)
ǫxy
−γy + − δy 2 = 0.
αx + y

Therefore, the existence of all ecological feasible equilibria of system


(1.6) is equivalent to that of nonnegative solutions of equations (2.1).
Since h 6= 0, there does not exist any equilibrium on the positive y-axis.
All ecological feasible equilibria of system (1.6) are either on the positive
x-axis or in the interior of R2+ . If an equilibrium is at the positive x-axis,
we call it a boundary equilibrium. And if an equilibrium is in the interior
of R2+ , we call it a positive equilibrium.
It is clear that equations (2.1) have no solutions in R2+ if h > 1/4.
Hence, we focus on the solutions of (2.1) as 0 < h ≤ 1/4. Obvious, (2.1)
have at most two solutions (x1 , 0) and (x2 , 0) at the positive x-axis if
0 < h < 1/4,√ and a unique solution
√ (x0 , 0) = (1/2, 0) if h = 1/4, where
x1 = (1 − ∆1 )/2, x2 = (1 + ∆1 )/2 and ∆1 = 1 − 4h.
In the interior of R2+ the existence of solutions of equations (2.1) is
equivalent to that of the following equations

(x2 − x + h + ǫx)y = −αx(x2 − x + h),


(2.2)
(αγ − ǫ + αδy)x + y(γ + δy) = 0.

It is clear that equations (2.2) have no positive solutions if either


0 < ǫ ≤ αγ, or h = (1 − ǫ)2 /4 or h ≥ 1/4. And if equations (2.2) have
a positive solution (x∗ , y ∗ ), then one of the following three conditions
must hold.
298 L. CHEN, Y. LI AND D. XIAO

(C1) ǫ ≥ max{1, αγ}, 1/4 > h > 0 and (x∗ , y ∗ ) must be in the following
range
 
ǫ − αγ
Ω1 = (x, y) : x1 < x < x2 , 0 < y < .
αδ

(C2) 1 > ǫ > αγ, 1/4 > h > (1 − ǫ)2 /4 and (x∗ , y ∗ ) must be in the
range Ω1 .
(C3) 1 > ǫ > αγ, (1 − ǫ)2 /4 > h > 0 and (x∗ , y ∗ ) must be in the
following range

 
ǫ − αγ
Ω2 = (x, y) : x1 < x < x3 , 0 < y <
αδ
 
ǫ − αγ
∪ (x, y) : x4 < x < x2 , 0 < y < ,
αδ
√ √
where x3 = (1 − ǫ − ∆2 )/2, x4 = (1 − ǫ + ∆2 )/2 and ∆2 =
(1 − ǫ)2 − 4h.

In the ranges Ω1 and Ω2 , we consider the two curves Γ1 and Γ2 ,

y(γ + δy) αx(x2 − x + h)


Γ1 : x = , Γ2 : y = .
ǫ − αγ − αδy −x2 + (1 − ǫ)x − h

Therefore, equations (2.2) have a positive solution (x∗ , y ∗ ) if and only if


one of conditions (C1)–(C3) holds, and there exists an intersection point
(x∗ , y ∗ ) of curves Γ1 and Γ2 in the corresponding range, see Figure 1.
From the analysis of Γ1 and Γ2 , we have

Proposition 2.1. Equations (2.2) do not have a solution with multi-


plicity three or four, and do not have two solutions with multiplicity two,
respectively.

Based on the above analysis, we have the following lemma.

Lemma 2.2.

(i) System (1.6) has no equilibria if h > 1/4.


(ii) System (1.6) has only a boundary equilibrium E0 (x0 , 0) if h = 1/4.
(iii) system (1.6) has two boundary equilibria E1 (x1 , 0) and E2 (x2 , 0) if
0 < h < 1/4 and 0 < ǫ ≤ αγ; or if h = (1 − ǫ)2 /4 and 1 > ǫ > αγ.
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 299

(1−ǫ)2
(a) ǫ ≥ max{1, αγ}, 14 > h > 0 (b) 1 > ǫ > αγ, 1
4
>h> 4

(1−ǫ)2
(c) 1 > ǫ > αγ, 4
>h>0

FIGURE 1: The curves Γ1 and Γ2 in three cases

(iv) system (1.6) has at least three equilibria: boundary equilibria E1 (x1 ,
0), E2 (x2 , 0) and a positive equilibrium E ∗ (x∗ , y ∗ ) if one of condi-
tions (C1)–(C3) holds, and the curve Γ1 intersects Γ2 in Ω1 or Ω2 .

Even though it is algebraically impossible to find expressions for gen-


eral positive equilibria of system (1.6) in terms of the parameters, we
can obtain the expressions of positive equilibrium in special cases. For
example, system (1.6) has a unique positive equilibrium E+ (x+ , y+ ),
which is a tangent point of Γ1 and Γ2 (see Figure 2) if parameters
300 L. CHEN, Y. LI AND D. XIAO

(α, γ, δ, ǫ, h) ∈ SN , where
p
2 (α2 + 2)h

1
SN = (α, γ, δ, ǫ, h) : ǫ = αγ + 1, γ = ,
2 α
r
3α4 δ 2

3 2
h= , α δ<2 ,
16(α2 δ + 1) α2 δ + 1
3α2 δ αx+ (x2+ − x+ + h)
x+ = , y+ = .
4(α2 δ + 1) −x2+ + (1 − ǫ)x+ − h

FIGURE 2: System (1.6) has a unique positive equilibrium when


(α, γ, δ, ǫ, h) ∈ SN .

Now we study the topological classifications of equilibria of system


(1.6). Calculating the Jacobian matrix of system (1.6) at the boundary
equilibria and using the qualitative analysis, we have

Theorem 2.3.

(i) If γ 6= ǫ/α and h = 1/4, then E0 (1/2, 0) is a saddle-node.


(ii) If 1/4 > h > 0 and ǫ > αγ, then equilibrium E1 is an unstable node
and equilibrium E2 is a hyperbolic saddle.
(iii) If 1/4 > h > 0 and αγ > ǫ > 0, then equilibrium E1 is a hyperbolic
saddle and equilibrium E2 is a stable node.
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 301

y y

0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
x x
(a) saddle-node E0 (b) unstable E1 and saddle E2

0.2

0.15

0.1

0.05

0
0 0.2 0.4 0.6 0.8
x
(c) saddle E1 and stable E2

FIGURE 3: The phase portraits of boundary equilibria for system (1.6)

The phase portraits of boundary equilibria are shown in Figure 3.


On the other hand, if system (1.6) has a positive equilibrium E ∗ (x∗ , y ∗ ),
then we have

Theorem 2.4.

(i) Equilibrium E ∗ (x∗ , y ∗ ) is elementary if E ∗ is a cross point of the


curves Γ1 and Γ2 .
(ii) Equilibrium E ∗ (x∗ , y ∗ ) is degenerated if E ∗ is a tangent point of the
curves Γ1 and Γ2 .
302 L. CHEN, Y. LI AND D. XIAO

Proof. Since the Jacobian matrix J ∗ at E ∗ (x∗ , y ∗ ) is

ǫ y∗ 2 α ǫ x∗ 2
 
1 − − 2 x∗ −
(α x ∗ + y ∗ )2 (α x∗ + y ∗ )
2 

J = ,
 
ǫ y∗ 2 α ǫ x∗ 2
∗
−γ + − 2 δ y

2 2
(α x∗ + y ∗ ) (α x∗ + y ∗ )

det (J ∗ ) = 0 if Γ1 is tangent to Γ2 at E ∗ (x∗ , y ∗ ). Hence, E ∗ (x∗ , y ∗ ) is


degenerated.
If the curve Γ1 crosses Γ2 at E ∗ (x∗ , y ∗ ), then by computing we obtain
that det(J ∗ ) 6= 0. Moreover, if det(J ∗ ) < 0, then equilibrium E ∗ (x∗ , y ∗ )
is a hyperbolic saddle. If det(J ∗ ) > 0, then equilibrium E ∗ (x∗ , y ∗ ) is a
node, or focus or center. Hence, we finish the proof.
It is clear that system (1.6) may undergo some bifurcations if it has a
degenerated equilibrium, and it may undergo Hopf bifurcation if it has a
weak focus. In the following we study these bifurcations of system (1.6).

3 Bogdanov-Takens bifurcation In this section, we focus on the


positive equilibrium E+ (x+ , y+ ) of system (1.6) as parameters (α, γ, δ, ǫ,
h) ∈ SN , which is the tangent point of Γ1 and Γ2 .
We can check that E+ (x+ , y+ ) is a saddle-node equilibrium of system
(1.6) if (α, γ, δ, ǫ, h) ∈ SN and

(3.1) 2α(k +1)l −9k 4 +3αk 3 −12k 3 +αk 2 −10k 2 −4k +3αk +α−1 6= 0,
p p
where k = (α2 δ + 1)/3 and l = k(3k 2 + 1). Let

B = {(α, γ, δ, ǫ, h) : 2α(k + 1)l − 9k 4 + 3αk 3


− 12k 3 + αk 2 − 10k 2 − 4k + 3αk + α − 1 6= 0}.

Hence, system (1.6) can undergo saddle-node bifurcation in a small


neighborhood of E+ (x+ , y+ ) as parameters (α, γ, δ, ǫ, h) vary in a small
neighborhood of SN ∩ B which is transversal to the hypersurface SN .
If

(3.2) 2α(k +1)l −9k 4 +3αk 3 −12k 3 +αk 2 −10k 2 −4k +3αk +α−1 = 0,

then we have
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 303

0.2

0.15

0.1

0.05

0
0 0.2 0.4 0.6 0.8
x

FIGURE 4: System (1.6) has a unique positive equilibrium which is a


cusp of codimension 2.

Lemma 3.1. System (1.6) has a unique positive equilibrium E+ (x+ , y+ )


which is a cusp of codimension 2 if (α, γ, δ, ǫ, h) ∈ SN and (3.2) holds.
The phase portrait of system (1.6) is shown in Figure 4.

Proof. We first make a time scale change dt = (αx + y)dτ such that
system (1.6) is equivalent to the following system in the interior of the
first quadrant.

dx


 = −αhx − hy + αx2 + (1 − ǫ)xy − αx3 − x2 y,
(3.3) dτ
 dy = (ǫ − αγ)xy − γy 2 − αδxy 2 − δy 3 .

When (α, γ, δ, ǫ, h) ∈ SN , system (3.3) has a unique positive equilibrium


E+ (x+ , y+ ). Translating E+ (x+ , y+ ) to the origin by setting

x1 = x − x+ , y1 = y − y+ ,

we have

dx

 1 = ax1 + by1 + p11 x2 + 2p12 xy + O(|x1 , y1 |3 ),

(3.4) dτ
 dy1 = cx1 + dy1 + 2q12 xy + q22 y 2 + O(|x1 , y1 |3 ),


304 L. CHEN, Y. LI AND D. XIAO

where

2
a = −αh + 2αx+ + (1 − ǫ)y+ − 3αx+ − 2x+ y+ ,


b = −h + (1 − ǫ)x+ − x2+ ,




p11 = α − 3αx+ − y+ ,




p12 = 1 − ǫ − x+ ,



2 2
2


 c = −αγy + + ǫy+ − αδy+ ,
2
d = −αγx+ + ǫx+ − 2γy+ − 2αδx+ y+ − 3δy+ ,





 −αγ + ǫ
q12 = − αδy+ ,




 2
q = −γ − αδx − 3δy .
22 + +

Making the linear change of variables


x2 = x1 , y2 = ax1 + by1 ,
system (3.4) can be changed to
  
x˙2 = y2 + p11 − 2a p12 x22 + 2 p12 x2 y2 + O(|x2 , y2 |3 ),

(3.5) b b
y˙2 = d1 x2 + d˜1 x2 y2 + q22 y 2 + O(|x2 , y2 |3 ),

2
b 2
2 2
where d1 = ap11 − 2ab p12 − 2aq12 + 2ab q22 , d˜1 = 2a 2a
b p12 + 2q12 − b q22 .

By the theory of normal forms, there exists a C change of variables
of system (3.5) in a small neighborhood of (0, 0) such that system (3.5)
can be written as

x˙3 = y3 + O(|x3 , y3 |3 ),
(3.6)
y˙ = d x2 + d x y + O(|x , y |3 ),
3 1 3 2 3 3 3 3

2 2
where d1 = ap11 − 2ab p12 −2aq12 + 2ab q22 , d2 = 2q12 +2p11 − 2a
b (p12 +q22 ).
If d1 d2 6= 0, then the equilibrium E0 of system (1.6) is a cusp of
codimension 2 by results in [35].
We now claim that d1 6= 0 and d2 6= 0. √
In fact, if d1 = 0 then we can solve (α, δ) = ( 32(2+

3)
√ , 0) or (α, δ) =
34 2+2 3
(1, −2/3). Note that δ > 0. Thus, d1 6= 0. Similarly

if d2 = 0, then we
2(2+ 3)
can get four pairs of solutions (α, δ) = ( 3 √ √ , 0), (α, δ) = (1, −2/3),
34 2+2 3
(α, δ) = (0.7771872633, −1.545783170) and (α, δ) = (14.21963930,
0.5603420000). Note that δ > 0 and 0 < α2 δ < 2. Thus, d2 6= 0.
Hence, we complete the proof.
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 305

From the perspective of biology, Lemma 3.1 shows that extinction


of prey and predators can be explained as a deterministic process for
parameters in SN ∩ B.
Next, we show that system (1.6) can undergo Bogdanov-Takens bi-
furcation in a small neighborhood of E0 . We choose α, δ as bifurcation
parameters and consider the following system

ẋ = (α + λ1 )x2 + xy − ǫxy



−x3 (α + λ1 ) − x2 y − h(α + λ1 )x − hy,



(3.7)



 ẏ = −y(γ(α + λ1 )x + γy − ǫx
+(δ + λ2 )y(α + λ1 )x + (δ + λ2 )y 2 ),

where λ = (λ1 , λ2 ) are very small parameters with 0 ≤ |λ| ≪ 1, and


other parameters (α, γ, δ, ǫ, h) ∈ SN and satisfy (3.2).
Therefore, when λ1 = λ2 = 0, system (3.7) has a unique positive
equilibrium E0 which is a cusp of codimension 2. Following the process
of normal form in [30], we obtain the following equivalent system

ẋ = y,

(3.8) 2 d2 (λ)
ẏ = µ1 (λ1 , λ2 ) + µ2 (λ1 , λ2 )y + x + pd (0) xy + R(x, y, λ),

1

where d1 (0) = d1 , d2 (0) = d2 , µ1 (λ1 , λ2 ) and µ2 (λ1 , λ2 ) are smooth


functions with µ1 (0, 0) = 0 and µ2 (0, 0) = 0, and R(x, y, λ) is a power
series in (x, y) with power xi y j satisfying i + j ≥ 3, j ≥ 2.
By reductio ad absurdum, we can verify that

∂(µ1 , µ2 )
J0 =

∂(λ1 , λ2 ) λ=0
2
d2 x + y + (y+ + αx+ )(−x+ + ǫx+ + h + x2+ )2 F1
= √ 6= 0,
2 d1 (αh + ǫy+ − y+ − 2αx+ + 3αx2+ + 2x+ y+ )2 F2

where

F1 = 14αδx2+ y+ − 12γx2+ y+ − 9δhy02 − 11αγx3+ − ǫhx+ + 8γx3+ y+


+ 7αγx4+ + αγh2 + 28αδhx20 y+ + 8αγhx20 − 9ǫδx+ y+
2

− 18αδhx0 y+ − 4γhy0 + 4αδh2 y+ + ǫx3+ + 2α2 δh2 x+


306 L. CHEN, Y. LI AND D. XIAO

− 6α2 δhx20 + 9ǫδhy02 + 4ǫγhy0 + 8α2 δhx30 − 4γǫx+ y+


− 5αγhx0 + 4ǫγx2+ y+ + ǫh2 + 2αδǫhx0 y+ − 2αδǫx2+ y+
+ 2αδǫx3+ y+ − 10α2 δx4+ − ǫx4+ + 6α2 δx5+ + 4αγx2+
+ 4α2 δx3+ + 9δx+ y+
2
+ 4γx+ y+ + 9ǫδx2+ y+
2 2
+ 18δhx0 y+
− 38αδx3+ y+ + 24αδx4+ y+ + 8γhx0 y+ − 27δx2+ y+
2
+ 18δx3+ y+
2
,
F2 = (−γy+ + 3x+ y+ − ǫx+ + 2ǫy+ − 3αx3+ + hy0 − 3x2+ y+
+ ǫ2 x+ + ǫh + ǫx2+ − ǫ2 y+ + αhx0 + 3αx2+
− αǫγx+ + αδhy0 + 9αδx2+ y+ + α2 δhx0 + αǫx+ − y+
2
− 5αδx+ y+ − αγx+ − 3δy+ − αx+ − 2α2 δx2+ − αδǫx+ y+
− 3ǫx+ y+ + 2αγx2+ − αǫh + 3δǫy+
2
+ ǫγy+
+ 3α2 δx3+ + 6δx+ y+
2
+ 2γx+ y+ )2 .

Hence, we have the following theorem by Bogdanov-Takens bifurcation


in [11] or [27].

Theorem 3.2. When the parameters (λ1 , λ2 ) vary in a small neighbor-


hood of the origin, system (3.7) undergoes Bogdanov-Takens bifurcation
in a small neighborhood of E0 . The local representations of the bifurca-
tion curves in the small neighborhood of the origin are given as follows.

(1) SN ± = {(λ1 , λ2 ) : µ1 (λ1 , λ2 ) = 0,pµ2 > 0 or µ2 < 0};


(2) H = {(λ1 , λ2 ) : µ2 (λ1 , λ2 ) = √dd2 −µ1 (λ1 , λ2 ), µ1 (λ1 , λ2 ) < 0};
1

(3) HL = {(λ1 , λ2 ) : µ2 (λ1 , λ2 ) = 75 √dd2 −µ1 (λ1 , λ2 ), µ1 (λ1 , λ2 ) < 0};


p
1

±
where SN , H, HL are saddle-node, Hopf, homoclinic bifurcation respec-
tively.

As an example, we take

√ √
q
α0 = 85 + 35 6 − 2 3564 + 1455 6 ≈ 1.87708925,
√ √
q
δ0 = (28831 + 11770 6 − 20 415552 + 1696497 6)−1 ≈ 0.283813,
√ √
q
3
γ0 = (66 6 − 12 + 5 4543 6 − 10212) ≈ 0.625339,
1444
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 307

1 √
4
ǫ0 = (8 + 3 6) ≈ 1.58691,
8
r
1 3 3
h0 = ≈ 0.114820.
32 2
Then the following system
 ǫ0 xy
 ẋ = x − − x 2 − h0 ,


 α 0x + y
ǫ0 xy

(3.9) ẏ = −γ0 y + − δ0 y 2 ,


 α 0x + y

ẋ = −h0 , ẏ = 0, when (x, y) = (0, 0)

has a unique positive equilibrium Ě(0.375, 0.177), which is a cusp. From


Lemma 3.1 we can compute the normal form of the cusp with d1 =
−0.124573 and d2 = −1.73141, i.e. d1 d2 > 0. We choose α and δ as
bifurcation parameters and consider the perturbation system of (3.9)
 1.58691xy
ẋ = x −
 − x2 − 0.114820,


 (1.87708925 + λ1 )x + y
(3.10) 1.58691xy
ẏ = −0.625339y +
(1.87708925 + λ1 )x + y




−(0.283813 + λ2 )y 2 ,

where λ1 and λ2 are very small parameters. We can check that system
(3.10) undergoes the Bogdanov-Takens bifurcation. The bifurcation dia-
gram of system (3.10) is sketched in Figure 5(a), and the corresponding
phase portraits are shown in Figure 5(b)–Figure 9.
When (λ1 , λ2 ) = (0, 0), system (3.10) has a unique positive equilib-
rium (x0 , y0 ) = (0.375, 0.177), which is cusp of codimension 2 (see Figure
5(b)).
When parameters lie on the curve SN + , for example, (λ1 , λ2 ) =
(0.1544, −0.3), the unique positive equilibrium becomes a saddle-node
(see Figure 6(a)). When parameters (λ1 , λ2 ) cross the curve SN + into
region I, system (3.10) undergoes saddle-node bifurcation. When param-
eters (λ1 , λ2 ) are in region I, for example, taking (λ1 , λ2 ) = (0.1653, −0.3),
system (3.10) has two positive equilibria: a hyperbolic saddle and a un-
stable focus (see Figure 6(b)).
When parameters lie on the curve H, for example, (λ1 , λ2 ) = (0.1811,
−0.3), system (3.10) has an unstable weak focus of order one and a
hyperbolic saddle (see Figure 7(a)). As parameters cross the curve H
308 L. CHEN, Y. LI AND D. XIAO

0.2

0.15

0.1

0.05

0
0 0.2 0.4 0.6 0.8
x

(a) The bifurcation diagram of BT (b) A cusp of codimension 2 as λ1 =


bifurcation. λ2 = 0.

FIGURE 5: The Bogdanov-Takens bifurcation of system (3.9).

y y

0.25 0.25

0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
x x
(a) Saddle-node equilibrium as (b) Saddle and unstable focus as
(λ1 , λ2 ) ∈ SN + . (λ1 , λ2 ) ∈ I.

FIGURE 6: The saddle-node bifurcation and the corresponding phase


portraits of system (3.9).
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 309

into the region II (i.e., the region between H and HL ), for example,
(λ1 , λ2 ) = (0.1811, −0.3), system (3.10) undergoes Hopf bifurcation and
a unstable limit cycle appears (see figure 7(b)).
When parameters lie on the curve HL, for example, (λ1 , λ2 ) = (0.186,
−0.3), system (3.10) has an unstable homoclinic loop and a stable focus
(see Figure 8(a)). And as parameters cross the curve HL into the region
III, for example, (λ1 , λ2 ) = (0.2, −0.3), the homoclinic loop disappears
and system (3.10) has a hyperbolic saddle and a stable focus (see Figure
8(b)).
When parameters lie on the curve SN − , for example, (λ1 , λ2 )) =
(−0.04401, 0.1), system (3.10) has a unique positive equilibrium which
is saddle-node (see Figure 9(a)). When parameters cross the curve SN −
into the region IV, system (3.10) undergoes saddle-node bifurcation.
When parameters lie the region IV, for example, (λ1 , λ2 ) = (0.1, −0.3),
system (3.10) has no positive equilibria (see Figure 9(b)).

y y

0.25 0.25

0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
x x
(a) An unstable weak focus as (b) An unstable limit cycle and stable
(λ1 , λ2 ) ∈ H. focus as (λ1 , λ2 ) ∈ II.

FIGURE 7: The Hopf bifurcation and the corresponding phase portraits


of system (3.9).
310 L. CHEN, Y. LI AND D. XIAO

y y

0.25 0.25

0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
x x
(a) An unstable homoclinic loop as (b) A saddle and stable focus as
(λ1 , λ2 ) ∈ HL. (λ1 , λ2 ) ∈ III.

FIGURE 8: The Homoclinic bifurcation and the corresponding phase


portraits of system (3.9).

y y

0.25

0.25

0.2
0.2

0.15
0.15

0.1
0.1

0.05 0.05

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
x x
(a) A saddle-node positive equilibrium (b) No positive equilibria as (λ1 , λ2 ) ∈
as (λ1 , λ2 ) ∈ SN − . IV .

FIGURE 9: The saddle-node bifurcation and the corresponding phase


portraits of system (3.9).
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 311

4 Hopf bifurcation In this section, we will discuss the Hopf bifur-


cation of system (1.6). Suppose that E0 (x0 , y0 ) is a positive equilibrium
which is center type, that is, the Jacobian matrix J0 at E0 (x0 , y0 ) has
the vanish trace. Thus, the following equations hold.

ǫx0 y0
x0 − − x0 2 − h = 0,
αx0 + y0
ǫx0
(4.1) −γ + − δy0 = 0,
αx0 + y0
ǫy0 2 − αǫx0 2
1 − 2x0 − γ − 2δy0 − = 0.
(αx0 + y0 )2

We cannot obtain the expressions of x0 and y0 from equations (4.1)


depending on all parameters. We now look for some conditions of param-
eters such that equations (4.1) has a pair of positive solutions (x0 , y0 ).
From the analysis in section 2, we know that x0 must satisfy the follow-
ing equation

(4.2) F (x) , Ax4 + Bx3 + Cx2 + Dx + E = 0,

where



 A = α2 δ + 1,


B = 2(ǫ − 1) − αγ − α2 δ,






C = αγ(1 − ǫ) + (ǫ − 1)2 + 2h + α2 δh,



D = h [2(ǫ − 1) − αγ] ,






 E = h2 .

Let

m , α + αγ − ǫ + (1 − α)α2 γ 2 ǫ + 2(α − 1)αγǫ2 + (1 − α)ǫ3 .

We have

Lemma 4.1. System (1.6) has a positive equilibrium E0 (x0 , y0 ), which


312 L. CHEN, Y. LI AND D. XIAO

is a center or focus (that is center type) if



0 < ǫ < 1,





 2(1 − ǫ)(1 − ǫ + αγ)
δ= > 0,






(4.3) m 
2α(1 − ǫ2 + αγǫ) − m > 0,

h= 2


 4α
2
− ǫ + αγ)2 − (1 − ǫ2 + αγǫ)2 > 0,




 αǫ (1



m > 0,

where

(1 − ǫ)(1 − ǫ + αγ) m
x0 = = ,
α2 δ 2α
(ǫ − αγ)(1 − ǫ) m(ǫ − αγ)
y0 = = > 0.
αδ 2(1 − ǫ + αγ)

Proof. By straight forward calculation, we can check that (x0 , y0 ) is


a solution of equations (4.1) if (4.3) holds. We compute the Jacobian
matrix J0 at E0 (x0 , y0 ) and get that
 
a −b
J0 =
c −a

2
ǫ(1−ǫ+αγ)2
where a = (ǫ−αγ)(1−ǫ
α
+αγǫ)
,b= α , and c = ǫ(ǫ−αγ)2 . Further,
we obtain that

tr (J0 ) = 0,
(ǫ − αγ)2  2
αǫ (1 − ǫ + αγ)2 − (1 − ǫ2 + αγǫ)2 .

det (J0 ) = 2
α

By the third inequality in (4.3), we know that det(J0 ) > 0. Therefore,


E0 (x0 , y0 ) is a center or focus (that is center type). We finish the proof.

Now we determine if the equilibrium E0 is a weak focus under the


condition (4.3). Translate E0 to the origin if (4.3) holds, we write system
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 313

(1.6) in Taylor expansion as follows.


ẋ = ax − by + a20 x2 + a11 xy + a02 y 2 + a30 x3 + a21 x2 y
+ a12 xy 2 + a03 y 3 + O(|x, y|4 ),
(4.4)
ẏ = cx − ay + b20 x2 + b11 xy + b02 y 2 + b30 x3 + b21 x2 y
+ b12 xy 2 + b03 y 3 + O(|x, y|4 ),
where
2αǫ
a20 = (1 − ǫ + αγ)(ǫ − αγ)2 − 1,
m
−4ǫ
a11 = (ǫ − αγ)(1 − ǫ + αγ)2 ,
m

a02 = (1 − ǫ + αγ)3 ,

−4ǫα2
a30 = (ǫ − αγ)2 (1 − ǫ + αγ)2 ,
m2
−4ǫ
a03 = (1 − ǫ + αγ)4 ,
m2 α
−4αǫ
a21 = (2 − 3ǫ + 3αγ)(ǫ − αγ)(1 − ǫ + αγ)2 ,
m2
−4ǫ
a12 = (1 − 3ǫ + 3αγ)(1 − ǫ + αγ)3 ,
m2
−2ǫα
b20 = (1 − ǫ + αγ)(ǫ − αγ)2 ,
m

b11 = (ǫ − αγ)(1 − ǫ + αγ)2 ,
m
−2(1 − ǫ + αγ)
b02 = [1 − ǫ(ǫ − αγ)(2 − ǫ + αγ)] ,

b30 = −a30 , b21 = −a21 , b12 = −a12 , b03 = −a03 .
From the formula of the first Liapunov number in [27], we have the
first Liapunov number V3 of system (4.4) at the origin as follows:
π
V3 = F,
4c3 d3
where
F = d4 (2a20 b20 + b11 b20 − 3ca30 − cb21 ) + a2 d2 (4a20 b20
314 L. CHEN, Y. LI AND D. XIAO

+ 2b11 b20 − 3ca30 − cb21 ) + acd2 (a20 b11 − 2a220


+ b211 + a11 b20 + 2b02 b20 − 2ca21 − 2cb12 )
− d2 (c2 b02 b11 − c2 a11 a20 − c3 a12 − 3c3 b03 )
− (2aa20 + ab11 + ca11 + 2cb02 )
× (a3 b20 − a2 ca20 + a2 cb11 − ac2 a11 + ac2 b02 − c3 a02 ).

By the condition (4.3), we simplify V3 and obtain that

π(1 − ǫ + αγ)2
V3 = H(ǫ, α, γ),
2α3 d3 m2 (ǫ − αγ)

where H(ǫ, α, γ) has a long expression and it can be written as

H(ǫ, α, γ) = α(ǫ − α)2 + 8ǫ3 + O(|ǫ, α, γ|3 ).

Therefore, the first Liapunov number V3 of system (4.4) at the origin


has the same sign to H(ǫ, α, γ). If V3 6= 0, then E0 is a weak focus with
multiplicity one. If V3 = 0, then E0 is a center or a weak focus with
multiplicity at least two. We would like to point out that there exist
parameter values in (4.3) such that V3 > 0. Thus, E0 is an unstable
weak focus of multiplicity one. System (1.6) undergoes subcritical Hopf
bifurcation as parameters vary in the small neighborhood of (4.3).
9
As an example, we take (ǫ1 , α1 , γ1 , δ1 , h1 ) = ( 10 , 2, 14 , 364
15 6279
, 62500 ) and
consider the following system
 ǫ1 xy
 ẋ = x − − x 2 − h1 ,

 α 1x + y
ǫ1 xy

(4.5) ẏ = −γ1 y + − δ1 y 2 ,


 α 1 x + y
ẋ = −h1 , ẏ = 0, when (x, y) = (0, 0).

System (4.5) has two positive equilibria E1 (0.364, 0.48533) and


E2 (0.30497, 0.4173), E1 (0.364, 0.48533) is unstable weak focus with mul-
tiplicity one and E2 (0.30497, 0.4173) is a hyperbolic saddle.
Choosing α as a bifurcation parameter, we consider the perturbation
system of system (4.5)
 ǫ1 xy 2
ẋ = x − (α1 + λ)x + y − x − h1 ,


(4.6)
ǫ1 xy
ẏ = −γ1 y + − δ1 y 2 ,


(α1 + λ)x + y
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 315

where λ is a very small parameter.


When parameter λ increases from zero, system (4.6) undergoes sub-
critical Hopf bifurcation and an unstable limit cycle appears. The bi-
furcation diagram and the corresponding phase portrait are shown in
Figure 10.

0.45

0.4

0.35

0.3

0.25
-0.03 -0.025 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015
lambda
(a) The bifurcation diagram x vs. λ.

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8
x
(b) an unstable limit cycle appears as λ = 0.006.

FIGURE 10: The Hopf bifurcation and phase portraits of system (4.6).
.
316 L. CHEN, Y. LI AND D. XIAO

5 Discussion In this paper, we have investigated the dynamics and


bifurcations of a class of ratio-dependent predator-prey models with prey
harvesting by modifying the classical Bazykin’s model. We have shown
the existence of positive equilibria, local stability and bifurcation be-
havior of the model, which reveals that the dynamics of populations are
sensitive to the corresponding net natural growth rates, consumption
rate by predators, intraspecies competition rates and harvesting rate.
For some values of these parameters and suitable initial populations, the
prey and predators can coexist at a positive equilibrium. However, in-
stead of the desired coexistence, the most frequent result is that the both
populations go extinct either after a couple of oscillations or predators
driven to extinction after prey. In particular, if the harvesting rate h is
larger than 1/4 or the parameters satisfy 1/4 > h > 0 and ǫ > αγ, then
extinction of both populations is a deterministic outcome for all initial
populations. From the perspective of biological control, over-harvesting
of prey species or the combined effects of predation and harvesting is
significant on the destruction of ecological systems. Therefore, it is very
essential to the optimal management of exploitation resources for the
sustainable development of eco-economic systems.

Acknowledgments. We are very grateful to Professor Shigui Ruan


and anonymous referees for their valuable comments and suggestions,
which led to an improvement of our original manuscript.

REFERENCES

1. P. A. Abrams, The fallacies of Ratio-dependent predation, Ecology 75 (1994),


1842–1850.
2. M. Agarwal, T. Fatima and H. I. Freedman, Depletion of forestry resource
biomass due to industrialization pressure: a ratio-dependent mathematical
model, J. Biol. Dyn. 4 (2010), 381–396.
3. H. R. Akcakaya, Population cycles of mammals: evidence for a ratio-dependent
predation hypothesis, Ecol. Monogr. 62 (1992), 119–142.
4. H. R. Akcakaya, R. Arditi and L. R. Gingburg, Ratio-dependent predation: an
abstraction that works, Ecology 76 (1995), 995–1004.
5. R. Arditi and L. R. Ginzburg, Coupling in predator-prey dynamics: ratio-
dependence, J. Theoret. Biol. 139 (1989), 311–326.
6. R. Arditi, L. R. Ginzburg and H. R. Akcakaya, Variation in plankton densities
among lakes: a case of ratio-dependent models, American Naturalist, 138
(1991), 1287–1296.
7. C. Azar, J. Holmberg and K. Lindgren, Stability analysis of harvesting in a
predator-prey model, J. Theoret. Biol. 174 (1995) 13–19.
BIFURCATIONS IN A MODEL WITH PREY HARVESTING 317

8. A. Bazykin, Nonlinear dynamics of interacting populations, Series on Nonlin-


ear Science. Ser. A, Vol. 11, L. O. Chua (ed.), World Scientific, Singapore. 1998.
Original Russian version: Mathematical Biophysics of Interacting Populations,
Nauka, Moscow, 1985.
9. F. Brauer and C. Castillo-Chavez, Mathematical Models in Population Biology
and Epidemiology, Springer-Verlag, 2001.
10. K. S. Chaudhuri and S. Saha Roy, On the combined harvesting of a prey-
predator system, J. Biol. Syst. 4 (1996), 373–389.
11. S.-N. Chow, C. Z. Li and D. Wang, Normal Forms and Bifurcation of Planar
Vector Fields, Cambridge University Press, Cambridge, 1994.
12. C. W. Clark, Mathematical Bioeconomics: The Optimal Management of Re-
newable Resources, 2nd ed., John Wiley and Son, New York, Toronto, 1990.
13. G. Dai and M. Tang, Coexistence region and global dynamics of a harvested
predator-prey system, SIAM J. Appl. Math., 58 (1998), 193–210.
14. H. I. Freedman, Deterministic Mathematical Method in Population Ecology,
Dekker, New York, 1980.
15. H. I. Freedman and R.M. Mathsen, Persistence in predator-prey systems with
ratio-dependent predator influence, Bull. Math. Biol. 55 (1993), 817–827.
16. H. I. Freedman, M. Agarwal and S. Devi, Analysis of stability and persistence
in a ratio-dependent predator-prey resource model, Int. J. Biomath. 2 (2009),
107–118.
17. M. Haque, Ratio-dependent predator-prey models of interacting populations,
Bull. Math. Biol. 71 (2009), 430–452.
18. S. B. Hsu, T. W. Hwang and Y. Kuang, Global analysis of the Michaelis-
Menten-type ratio-dependent predator-prey system, J. Math. Biol. 42 (2001),
489–506.
19. S. B. Hsu, T. W. Hwang and Y. Kuang, Rich dynamics of ratio-dependent one
prey two predators model, J. Math. Biol. 43 (2001), 377–396.
20. S. B. Hsu, T. W. Hwang and Y. Kuang, A ratio-dependent food chain model
and its application to biological control, Math. Biosci. 181 (2003), 55–83.
21. C. Jost, O. Arino and R. Arditi, About deterministic extinction in ratio-
dependent predator-prey models, Bull. Math. Biol. 61 (1999), 19–32.
22. Y. Kuang, Rich dynamics of Gause-type ratio-dependent predator-prey system,
Fields Inst. Commun. 21 (1999), 325–337.
23. Y. Kuang and E. Bertta, Global qualitative analysis of a ratio-dependent predator-
prey system, Fields Inst. Commun. 36 (1998), 389–406.
24. K. Lan and C. Zhu, Phase portraits of predator-prey systems with harvesting
rates, Discrete Contin. Dyna Syst. 32 (2012), 901–933.
25. B. Li and Y. Kuang, Heteroclinic bifurcation in the Michaelis-Menten-type
ratio-dependent predator-prey system, SIAM J. Appl. Math. 67 (2007), 1453–
1464.
26. R. M. May, Stability and Complexity in Model Eco-System, Princeton Univer-
sity Press, Princeton, 1974.
27. L. Perko, Differential Equations and Dynamical Systems, Springer-Verlag,
New York, 1996.
28. P. D. N. Srinivasu, Bioeconomics of a renewable resource in presence of a
predator, Nonlinear Anal. RWA 2 (2001) 497–506.
29. P. N. V. Tu and E.A. Wilman, A generalized predator-prey model: Uncertainty
and management, J. Environ. Econ. Manag. 23 (1992) 123–138.
30. D. Xiao and S. Ruan, Global dynamics of ratio-dependent predator-prey sys-
tem, J. Math. Biol. 43 (2001), 268–290.
31. D. Xiao and W. Li, Stability and bifurcation in a delayed ratio-dependent
predator- prey system, Proc. Edinburgh Math. Soc. 46 (2003), 205–220.
32. D. Xiao and L. S. Jennings, Bifurcations of a ratio - dependent predatorCprey
system with constant rate harvesting, SIAM J. Appl. Math. 65 (2005), 737–753.
318 L. CHEN, Y. LI AND D. XIAO

33. D. Xiao, W. Li and M. Han, Dynamics in a ratio-dependent predator-prey


model with predator harvesting, J. Math. Anal. Appl. 324 (2006), 14–29.
34. R. Xu, M. A. J. Chaplain and F. A. Davidson, Persistence and global stability
of a ratio-dependent predator-prey model with stage structure, Appl. Math.
Compu. 158 (2004), 729–744.
35. Z. Zhang, T. Ding, W. Huang and Z. Dong, Qualitative Theory of Differential
Equations, Science Press, Beijing, (in Chinese). English Edition: Qualitative
Theory of Differential Equations, Translations of Mathematical Monographs
101, Amer. Math. Soc. Providence, 1992.

Department of Mathematics, Shanghai Jiao Tong University,


Shanghai 200240, China.

School of Science, East China University of Science and Technology,


Shanghai 200237, China.
E-mail address: ylli@ecust.edu.cn

Corresponding author
Department of Mathematics, Shanghai Jiao Tong University,
Shanghai 200240, China.
E-mail address: xiaodm@sjtu.edu.cn

You might also like