According to integrated pest management strategies, we construct and investigate the dynamics of a Holling-Tanner predator-prey system with state dependent impulsive effects by releasing natural enemies and spraying pesticide at different thresholds. Applying the Dulacs criterion, the global stability of the positive equilibrium in the system without impulsive effect is discussed. By using impulsive differential equation geometry theory and the method of successor functions, we prove the existence of periodic solution of the system with state dependent impulsive effects. Furthermore, the stability conditions of periodic solutions are obtained. Some simulations are exerted to illustrate the feasibility of our main results.
1. Introduction
In 2012, the main maize area in Northern China embraced the large-scale plant diseases and insect pests [1]. With the good weather condition, Mythimna separata Walker, Ostrinia furnacalis, aphid, and other pests had a mass propagation which brought a large damage to the production of corn. How human beings effectively control the pests has been a significant task. Due to its simple operation and quick effect, spraying pesticide has always been the major way to kill pests for a long time. However, there may be pesticide residues in vegetables and crops threatening people’s good health and damaging the environment. With the increasing awareness of the environment, people are paying more attention to developing green agriculture. Releasing natural enemies and artificial capture are becoming significant means of controlling pests.
Different control strategies will be applied due to different behavior features of pests and their damages for crops in different stages. For instance, in the spawning period of Ostrinia furnacalis, artificial releasing of Trichogramma (the natural enemy of Ostrinia furnacalis) will be applied with the control effect of 70%–80%; if the hatchability is over 30%, pesticide control will be used instantly [2]. In consideration of the rapidity of chemical control and nonpollution of biological control, people control pests integrating with biological, physical, and chemical means under the EIL (economic injury level) to realize environmental, economic, and social profits together. Spraying pesticides and releasing natural enemies are instantaneous; these phenomena can be described as the impulsive differential equations. In recent decades, the theoretical research on the impulsive differential equation has represented a significant development and has been widely used in various mathematical ecological models [3–11] and many scholars made a deep analysis of the impulsive differential ecological system at a fixed time and have got some important products [12–26]. However, in the actual process of pest control, relevant measures will be used according to pest quantity and its damage to crops, which is the state impulsive differential system. Tang et al. [27, 28], Zeng et al. [29], Zhao et al. [30], Nie et al. [31, 32], and literatures [33–37] had a further exploration for the system and made great progress. Based on the study of a Holling-Tanner system, an integrated pest control model with two impulses is established aiming at the specific pest conditions in different threshold.
Let x=x(t), y=y(t) denote the population densities of pest (prey) and natural enemy (predator) at time t, respectively; then the predator-prey system usually can be expressed as [38]
(1)dxdt=xg(x)-yΦ(x),dydt=y[-q(x)+cΦ(x)],
in which g(x) denotes the relative growth rate of the prey; Φ(x) is the functional response function of the predator; q(x) is the mortality rate of the prey. The literature [39] studied a class of Holling-Tanner system with functional response function Φ(x)=mx/(A+x). This system was given by
(2)dxdt=rx(1-xK)-mxA+xy,dydt=ys(1-hyx),hhhhhx(0)>0,y(0)>0.
Here, let the prey population be growth in logistic and the environmental capacity is K; the intrinsic growth rate of predator is s and the carrying capacity is proportional to the number of prey. Introducing transformation t~=rt, x~(t~)=x(t)/K, y~(t~)=my(t)/rK, and letting δ=s/r, β=sh/m, a=A/K, the system (2) is changed into a dimensionless form:
(3)dxdt=x(1-x)-xa+xy,dydt=y(δ-βyx),hhhx(0)>0,y(0)>0.
In order to carry out integrated control of pests, we adopt strategies as follows.
(1) When the pest density x(t) reached a lower level x=h1, we release natural enemies to control pests for low damage of insect pests on crops; that is,
(4)Δx(t)=0,Δy(t)=λ,x=h1,
where λ is amount of natural enemies y(t) released one time.
(2) If the pest density x(t) reached a higher level x=h2, due to the fact that the damage of insect pests on crops is severe at this time, we effectively combine spraying pesticides with releasing natural enemies to control pests; that is,
(5)Δx(t)=-px(t),Δy(t)=-qy(t)+τ,x=h2.
Here, τ is the amount of natural enemies y(t) released one time, p, q are mortality rates of pests and natural enemies which die from spraying pesticides, and p,q∈(0,1).
Synthesizing systems (3), (4), and (5), the following integrated pest management model is obtained:
(6)dx(t)dt=x(t)(1-x(t))-x(t)a+x(t)y(t),dy(t)dt=y(t)(δ-βy(t)x(t)),x≠h1,h2,orx=h1,y>y~,Δx(t)=0,Δy(t)=λ,x=h1,y≤y~,Δx(t)=-px(t),Δy(t)=-qy(t)+τ,x=h2,x(0)>0,y(0)>0,
where x(t) denotes the population density of pests at time t and y(t) denotes the population density of natural enemies at time t; a, δ, β are positive numbers; p, q, λ, τ are control parameters and positive numbers; the point (h1,y~) is the intersection of the isoclinic line y=(1-x)(x+a) and the straight line x=h1.
2. Preliminaries
In order to analyze the dynamics of the system (6), we introduce the basic knowledge of the state impulsive differential equations.
Consider the state impulsive differential equation:
(7)dx(t)dt=P(x,y),dy(t)dt=Q(x,y),(x,y)∉M(x,y),Δx(t)=α(x,y),Δy(t)=β(x,y),(x,y)∈M(x,y),
where P(x,y) and Q(x,y) have order-one continuous partial derivatives.
Definition 1.
The dynamic system which is formed by solution mapping of system (7) is called semicontinuous dynamic system, denoted by (Ω,f,φ,M), where f is semicontinuous dynamical system mapping and f:Ω→Ω, φ(M)=N, and in which φ is pulse mapping. Here, M(x,y) and N(x,y) are straight or curved line in the plane. M(x,y) is called impulsive set, and N(x,y) is called corresponding image set.
In the system (6), let M1={(x,y)∣x=h1,0<y≤y~}, and the image set corresponding to impulsive mapping (4) is N1=φ1(M1)={(x,y)∣x=h1,0<y≤y~+λ}. Let M2={(x,y)∣x=h2,y>0}, and the image set corresponding to impulsive mapping (5) is N2=φ2(M2)={(x,y)∣x(t+)=(1-p)h2,y(t+)=(1-q)y(t)+τ}.
Definition 2.
Assume that impulsive set M and image set N are straight lines, the orbit Π(A,t) of system (7) starting from point A on N hits M at point A1 and then jumps onto point A1+, then the function f(A)=yA1+-yA is defined as a successor function about point A, and then point A1+ is called successor point of A.
Lemma 3 (see [10, 11]).
Successor function is continuous.
Definition 4.
If there exists a point P0 on the image set N, and a constant T>0 such that Π(P0,T)=P∈M, φ(P)=P0∈N, then the orbit Π(P0,t) starting from P0 is called an order-one periodic solution of the system (7).
Lemma 5 (Bendixson theorem of impulsive differential equations [10, 11]).
Assume that G is a Bendixson region of system (7); if G does not contain critical points of system (7), then system (7) contains a closed orbit in G.
For the system (6), from Lemma 5, the following conclusion is obtained.
Lemma 6.
In system (6), if there exist points A and B on the image set N, such that the successor function satisfies f(A)f(B)<0, then there must exist an order-one periodic solution in system (6).
Lemma 7 (Analogue of Poincaré Criterion [3, 4]).
Assume that x=ξ(t), y=η(t) is the T-periodic solution of the following impulsive differential equations:
(8)dx(t)dt=P(x,y),dy(t)dt=Q(x,y),Φ(x,y)≠0,Δx(t)=α(x,y),Δy(t)=β(x,y),Φ(x,y)=0,
where P(x,y) and Q(x,y) contain order-one continuous partial derivatives and Φ(x,y) is a sufficiently smooth function with gradΦ(x,y)≠0.
If the multiplier μ satisfies the condition |μ|<1, then the periodic solution (ξ(t),η(t)) of the system (8) is orbitally asymptotically stable, where
(9)μ=∏j=1nκjexp[∫0T(∂P(ξ(t),η(t))∂xhμhhhhhhhhhhhhhh+∂Q(ξ(t),η(t))∂y)dt],
with
(10)κj=((∂β∂y·∂Φ∂x-∂β∂x·∂Φ∂y+∂Φ∂x)P+hκjhhh+(∂α∂x·∂Φ∂y-∂α∂y·∂Φ∂x+∂Φ∂y)Q+)×(∂Φ∂xP+∂Φ∂yQ)-1,
and P, Q, ∂α/∂x, ∂α/∂y, ∂β/∂x, ∂β/∂y, ∂Φ/∂x, ∂Φ/∂y are calculated at the point (ξ(τj),η(τj)), P+=P(ξ(τj+),η(τj+)), Q+=Q(ξ(τj+),η(τj+)), and τj(j∈N) is the time of the jth jump.
3. The Stability of System (6) without Impulsive Effect
In the system (6), if p=q=λ=τ=0, that is, the system without impulsive effect, the following system is obtained:
(11)dxdt=x(1-x)-xa+xy,dydt=y(δ-βyx),hhhx(0)>0,y(0)>0.
If set (x(t),y(t)) is an arbitrary solution of the system (11) satisfying the initial conditions, then the following lemma is obtained.
Lemma 8.
The solutions of the system (11) is bounded, which means ∃T>0 satisfies 0≤x(t)≤1 and 0≤y(t)≤δ/β for t≥T.
Obviously, the system (11) exhibits prey isocline L1:y=(1-x)(x+a), predator isocline L2:y=(δ/β)x, nontrivial equilibrium points E0(1,0), and E(x*,y*), and here;
(12)x*=(β-δ-aβ)+(β-δ-aβ)2+4aβ22β,y*=δβx*.
Calculating the variational matrix of the equilibrium point in the system (11), we get
(13)J(E0)=(-1-11+a0δ).
Obviously, E0 is saddle point. At E,
(14)J(E)=(x*x*+a(1-a-2x*)-x*x*+aδ2β-δ).
The characteristic equation of J(E) is
(15)f(λ)=λ2+pλ+q=0,
in which p=δ-(x*/(x*+a))(1-a-2x*), q=(δx*/(x*+a))(δ/β-(1-a-2x*)).
Thus,
(16)Δ=p2-4q=(δ-x*x*+a(1-a-2x*))2-4δx*x*+a(δβ-(1-a-2x*)),λ1λ2=δx*x*+a(δβ-(1-a-2x*))=δx*+a(x*2+a)>0,λ1+λ2=-(δ-x*x*+a(1-a-2x*))=-2x*2+(a-1+δ)x*+aδx*+a.
Let P(x)=2x2+(a-1+δ)x+aδ, and then
(17)Δ=P2(x*)(x*+a)2-4δ(x*2+a)x*+a=1(x*+a)2[P2(x*)-4δ(x*2+a)(x*+a)].
Based on the above analysis, we can get the following conclusion.
Theorem 9.
If P(x*)>0, the positive equilibrium point E(x*,y*) of the system (11) is locally asymptotically stable. Specially,
if (H1):0<P(x*)<(4δ(x*2+a)(x*+a))1/2, E(x*,y*) is a locally asymptotically stable focus,
if (H2):P(x*)⩾(4δ(x*2+a)(x*+a))1/2, E(x*,y*) is a locally asymptotically stable node.
Next, we discuss the global stability of E(x*,y*) about the system (11).
Theorem 10.
If (H3):a+δ≥1 or (H4):1-8aδ<a+δ<1 is true, then the positive equilibrium E(x*,y*) of the system (11) is globally asymptotically stable.
Proof.
From (H3) or (H4), we have P(x)=2x2+(a-1+δ)x+aδ>0 for x>0, and thus P(x*)>0. Structure a Dulac function as follows:
(18)B(x,y)=x+axy2,x>0,y>0.
Let f(x,y)=x(1-x)-(x/(a+x))y, g(x,y)=y(δ-β(y/x)), and thus
(19)∂(fB)∂x+∂(gB)∂y=-1xy2[2x2+(a-1+δ)x+aδ]∂(fB)∂x+∂(gB)∂y=-1xy2P(x)⩽0.
By the Bendixson-Dulac theorem, there does not exist closed orbit of the system (11) around E. Based on Lemma 8 and Theorem 9, the positive equilibrium E(x*,y*) is globally asymptotically stable.
Remark 11.
If (H1), (H3) or (H1), (H4) are true, then E(x*,y*) is a globally asymptotically stable focus.
Remark 12.
If (H2), (H3) or (H2), (H4) are true, then E(x*,y*) is a globally asymptotically stable node.
Assume that E(x*,y*) is globally asymptotically stable focal point of the system (11), and then the illustration of vector graph of the system is as follows (see Figure 1).
Illustration of vector graph of system (11), where a=0.05, δ=0.5, β=0.7.
4. The Geometric Analysis of System (6) with Two State Impulses
In this section, we will discuss the existence and stability of periodic solution of system (6) only at focal point situation. So, we assume that the conditions (H1), (H3) or (H1), (H4) are true. According to the practical significance of the integrated pest management model, the condition (H5):h1<(1-p)h2<h2<x* is always given as such. By the analysis of system (6), the curve L1:y=(1-x)(x+a) is X-isocline, and the line L2:y=(δ/β)x is Y-isocline. Let points P, Q, R be the intersection of the curve L1 and lines x=h1, x=(1-p)h2, x=h2, respectively. Obviously E(x*,y*) is the intersection point of L1 and L2. From the previous discussion, we know that the first impulsive set is M1={(x,y)∣x=h1,0<y≤y~}, and the image set corresponding to M1 is N1={(x,y)∣x=h1,0<y≤y~+λ}; the second impulsive set is M2={(x,y)∣x=h2,y>0}, and the image set corresponding to M2 is N2={(x,y)∣x(t+)=(1-p)h2,y(t+)=(1-q)y(t)+τ}. The structure of the system can be shown as in Figure 2.
The structure graph of system (6).
Using successor function and geometric theory of impulsive differential equations and according to different positions of orbit initial points, the existence and stability of periodic solution of system (6) are discussed as follows.
4.1. The Initial Point on N1
Let C0(h1,yC0) be an initial point of the orbit of the system (6); if yC0<y~, point C0 is below point P(h1,y~), then C0∈M1 (M1 is impulsive set), and the image point C0+ of C0 must be above point P with n times impulses; therefore, we only need to discuss the cases of yC0>y~.
The orbit Π(C0,t) starting from C0(h1,yC0) hits the impulsive set M1 at point C1(h1,yC1), and then C1 jumps to point C11(h1,yC11). If yC11<y~, C11 continued to jump to point C12(h1,yC12), and after n times it reaches the point C1n(h1,yC1n), where yC11=yC1+λ, yC12=yC1+2λ,…, yC1n=yC1+nλ, and y~<yC1n<y~+λ. The situation of point C1n has three cases as follows.
(a) If yC1n=yC0(see Figure 3(a)), C1n is coincident with C0, then the curve C0C1C11⋯C1n is closed orbit, and the system (6) exhibits a 1-periodic solution.
The orbit starting from the point C0 on N1.
(b) If yC1n>yC0 (see Figure 3(b)), C1n is above C0; in this time, the successor function of C0 satisfies f(C0)=yC1n-yC0>0. In the meantime, choose a point D0(h1,yD0) on N1 satisfying yD0>y~+λ. The orbit Π(D0,t) starting from D0 hits the impulsive set at point D1(h1,yD1), and D1 jumps sometimes to point D1m(h1,yD1m), where y~<yD1m<y~+λ. Thus, the successor function of D0 satisfies f(D0)=yD1m-yD0<0.
According to Lemma 6, the system (6) exhibits a periodic solution, and the initial point of the periodic solution is between C0 and D0.
(c) If yC1n<yC0 (see Figure 3(c)), C1n is below C0; in this time, the successor function of C0 is f(C0)=yC1n-yC0<0. On the other hand, choose a point D0(h1,yD0) on N1 satisfying y~<yD0<y~+ε (ε is a sufficiently small positive number). The orbit Π(D0,t) starting from D0 hits the impulsive set M1 at point D1(h1,yD1), and D1 jumps to point D11(h1,yD11), where yD11=yD1+λ. As D0 sufficiently closed to P, D1 is sufficiently close to P, then yD11=yD1+λ>yD0. Thus, the successor function of D0 satisfiesf(D0)=yD1-yD0>0.
From Lemma 6, system (6) exhibits a 1-periodic solution.
To sum up the above discussed, we get the following.
Theorem 13.
If the initial point C0(h1,yc0) of the orbit of the system (6) is on N1 with yC0>y~, then the system exhibits a 1-periodic solution.
Next, we will discuss the stability of the above periodic solutions.
Theorem 14.
Let (ξ(t),η(t)) be the T-periodic solution of the system (6) with the initial point C0(h1,η0); the closed orbit corresponding to the periodic solution is the curve C0C1C11⋯C1n, if
(20)|μ|=|κexp{-∫0T(ξ(t)-ξ(t)η(t)(ξ(t)+a)2+βη(t)ξ(t))dt}|<1,
where
(21)κ=η0-nλη0∏j=1n1-h1-(1/(h1+a))(η0-(n-j)λ)1-h1-(1/(h1+a))(η0-(n-j+1)λ),
then the periodic solution (ξ(t),η(t)) is orbitally asymptotically stable.
Proof.
Let the orbit Π(C0,t) with the initial point C0(h1,η0) hit the impulsive set M1 at C1(ξ(T),η(T)), and then C1 jumps to the point C11(ξ(τ1),η(τ1)). The C11 continued to jump to point C12(ξ(τ2),η(τ2)), and at last, the image point C12 reaches point C1n(ξ(τn),η(τn)) with n times pulses. Here, η(τ1)=η(T)+λ,η(τ2)=η(T)+2λ,…,η(τn)=η(T)+nλ. For the jth time impulse, obviously ξ(τj+)=ξ(τj+1),η(τj+)=η(τj+1). For the system (6), let P(x,y)=x(1-x)-(x/(a+x))y, Q(x,y)=y(δ-β(y/x)), α(x,y)=0, β(x,y)=λ, Φ(x,y)=x-h1, and ξ(T)=h1, η(T)=η0-nλ; therefore we have
(22)∂P∂x=1-2x-a(a+x)2y,∂Q∂y=δ-2βyx,∂α∂x=∂α∂y=0,∂β∂x=∂β∂y=0,∂Φ∂x=1,∂Φ∂y=0.
According to Lemma 7, we get
(23)κj=((∂β∂y·∂Φ∂x-∂β∂x·∂Φ∂y+∂Φ∂x)P+hhh+(∂α∂x·∂Φ∂y-∂α∂y·∂Φ∂x+∂Φ∂y)Q+)×(∂Φ∂xP+∂Φ∂yQ)-1=P(ξ(τj+),η(τj+))P(ξ(τj),η(τj))=P(h1,η(T)+jλ)P(h1,η(T)+(j-1)λ)=1-h1-(1/(h1+a))(η0-(n-j)λ)1-h1-(1/(h1+a))(η0-(n-j+1)λ)=1-λ/(h1+a)1-h1-(1/(h1+a))(η0-(n-j+1)λ),μ=∏j=1nκjexp[∫0T(∂P∂x+∂Q∂y)dt]=∏j=1nκjexp[∫0T(1-2x-a(a+x)2y+δ-2βyx)dt]=∏j=1nκjexp{∫0T[-x+xy(a+x)2-βyx](1-x-yx+a)+(δ-βyx)hhhhhhhhhhhhhh-x+xy(a+x)2-βyx]dt}=∏j=1nκjexp{∫0Tdxx+∫0Tdyyhhhhhhhhhh-∫0T(x-xy(a+x)2+βyx)dt}=∏j=1nκjη0-nλη0exp[-∫0T(x-xy(a+x)2+βyx)dt]=η0-nλη0×∏j=1n1-h1-(1/(h1+a))(η0-(n-j)λ)1-h1-(1/(h1+a))(η0-(n-j+1)λ)×exp[-∫0T(x-xy(a+x)2+βyx)dt]=κexp{-∫0T(ξ(t)-ξ(t)η(t)(ξ(t)+a)2+βη(t)ξ(t))dt}.
From Lemma 7, if |μ|=|κexp{-∫0T(ξ(t)-(ξ(t)η(t)/(ξ(t)+a)2)+(βη(t)/ξ(t)))dt}|<1, then the periodic solution of the system (6) is orbitally asymptotically stable. This completes the proof.
Remark 15.
If 1-h1-(η0/(h1+a))>0 and β≥1, then the periodic solution with initial point C0(h1,η0) (where η0>y~) is orbitally asymptotically stable.
4.2. The Initial Point on N2
Based on existence and uniqueness theorem of differential equations, there exists a unique point Q0((1-p)h2,yQ0) on N2 such that the orbit Π(Q0,t) starting from Q0 is tangent to N1 at point P(h1,y~). Assume that C0((1-p)h2,yC0) is the initial point of the orbit Π(C0,t) of system (6). Next, we will investigate the existence of periodic solution of the system with different positions of C0 and Q0. Three cases should be discussed.
Case I (yC0=yQ0; see Figure 4). The initial point C0 is exactly Q0.
The initial point C0 on N2 (Case I: yC0=yQ0).
The orbit Π(Q0,t) starting from Q0 is tangent to N1 at the point P, and through N2 hit M2 at the point Q1(h2,yQ1), and then Q1 jumps to Q1+(xQ1+,yQ1+) on N2. According to (6), the following is obtained:
(24)xQ1+=(1-p)h2,yQ1+=(1-q)yQ1+τ.
About the points Q1+ and Q0, there are the following three positional relations.
If Q1+ coincides with Q0:yQ1+=yQ0 (see Figure 4(a)), then the curve Q0PQ1Q1+ is closed orbit.
If Q1+ is below Q0:yQ1+<yQ0 (see Figure 4(b)), then the successor function of Q0 satisfies f(Q0)=yQ1+-yQ0<0. In the meantime, take a point S0((1-p)h2,yS0) on N2 satisfying 0<yS0<ε (ε>0 small enough). The orbit Π(S0,t) starting from S0 hits the impulsive M2 at the point S1(h2,yS1), and then S1 jumps to the point S1+(xS1+,yS1+), where xS1+=(1-p)h2,yS1+=(1-q)yS1+τ. Obviously, the successor function of S0 is f(S0)=yS1+-yS0>0. From Lemma 6, the system (6) has an order one periodic solution, where the initial point of the periodic solution is between Q0 and S0.
If Q1+ is above Q0:yQ1+>yQ0 (see Figure 4(c)), the system (6) does not have closed orbit in the area Ω1={(x,y)∣h1<x<h2} in this time.
Based on the discussion above, we get the following.
Theorem 16.
Assume that the orbit Π(Q0,t) starting from Q0((1-p)h2,yQ0) is tangent to N1 at the point P(h1,y~) and hits the impulsive set M2 at the point Q1(h2,yQ1), the image point of Q1 is Q1+(xQ1+,yQ1+) on N2. If yQ1+≤yQ0, the system (6) has 1-periodic solution in the area Ω1={(x,y)∣h1<x<h2}.
Case II (yC0<yQ0). The initial point C0 is below Q0.
The isocline L1:y=(1-x)(x+a) intersects with the phase set N2 at the point Q((1-p)h2,yQ), and Q is below Q0. In this case, we discuss the existence of periodic solution of the system (6) with the example yC0=yQ (see Figure 5).
The initial point C0 on N2 (Case II: yC0<yQ0).
The orbit Π(Q,t) starting from Q moves to the point C1(h2,yC1) on the impulsive set M2, and C1 jumps onto C1+(xC1+,yC1+) on the image set N2, and then
(25)yC0=yQ=(1-(1-p)h2)(a+(1-p)h2),xC1+=(1-p)h2,yC1+=(1-q)yC1+τ.
(a) If yC1+=yQ:C1+ coincide with Q (see Figure 5(a)), the curve QC1C1+ is the closed orbit of the system (6).
(b) If yQ<yC1+≤yQ0:C1+ is between Q and Q0 (see Figure 5(b)), in this case the successor function of Q is f(Q)=yC1+-yQ>0. On the other hand, consider the orbit Π(Q0,t) starting from Q0,Π(Q0,t) hits the impulsive set M2 at Q1(h2,yQ1), and Q1 jumps onto Q1+(xQ1+,yQ1+) on N2. Based on the existence and uniqueness theorem of differential equations, Q1 must be below C1, and Q1+ must be below C1+. Thus yQ1+<yC1+≤yQ0, and the successor function of Q0 is f(Q0)=yQ1+-yQ0<0. From Lemma 6, the system (6) has 1-periodic solution, and the initial point of the periodic solution is between Q0 and Q.
If yC1+>yQ0:C1+ is above Q0, in this case the successor function of Q is f(Q)=yC1+-yQ>0. There are two different cases (case (c) and case (d)).
(c) If yC1+>yQ0 and yQ1+≤yQ0:Q1+ is below Q0 (see Figure 5(c)), f(Q0)=yQ1+-yQ0≤0, the system (6) has closed orbit.
(d) If yC1+>yQ0 and yQ1+>yQ0:Q1+ is above Q0 (see Figure 5(d)), the system (6) does not have closed orbit in the area Ω1={(x,y)∣h1<x<h2}.
(e) If yC1+<yQ:C1+ is below Q (see Figure 5(e)), in this case the successor function of Q is f(Q)=yC1+-yQ<0. On the other hand, take a point D0((1-p)h2,yD0) from N2 satisfying that yD0 is sufficiently small number, which is 0<yD0<ε (ε>0 small enough). The orbit Π(D0,t) starting from D0 moves to D1(h2,yD1) on the impulsive set M2, and D1 jumps onto D1+(xD1+,yD1+) on N2, where xD1+=(1-p)h2, yD1+=(1-q)yD1+τ. Obviously, the successor function of D0 is f(D0)=yD1+-yD0>0. From Lemma 6, the system (6) has closed orbit.
Based on the discussion above, we get the following.
Theorem 17.
Assume that the orbit Π(Q,t) starting from Q((1-p)h2,yQ) hits the impulsive set M2 at C1(h2,yC1), and C1 jumps onto C1+(xC1+,yC1+) on N2. The orbit Π(Q0,t) starting from the point Q0((1-p)h2,yQ0) is tangent to N1 at P(h1,y~), and hitting the impulsive set M2 at Q1(h2,yQ1), the image point of Q1 is Q1+(xQ1+,yQ1+). Then one has the following.
If yC1+≤yQ0, the system (6) has 1-periodic solution.
If yC1+>yQ0 and yQ1+≤yQ0, the system (6) has 1-periodic solution.
Case III (yC0>yQ0; see Figure 6). The initial point C0 is above Q0.
The initial point C0 on N2 (Case III: yC0>yQ0).
In this case, the orbit Π(C0,t) starting from C0 goes through the isocline L1 from the left of the line x=h1, hitting the impulsive set M1 at C1(h1,yC1). The same conclusion can be made as in Section 4.1.
Next, we discuss the stability of the periodic solution with the initial point on N2.
Theorem 18.
Assume that (ξ(t),η(t)) is the T-periodic solution of the system (6) with initial point C0((1-p)h2,η0); and if
(26)|μ|=|κexp{-∫0T(ξ(t)-ξ(t)η(t)(ξ(t)+a)2+βη(t)ξ(t))dt}|<1,
where
(27)κ=η0-τη0·1-(1-p)h2-(η0/((1-p)h2+a))1-h2-((η0-τ)/((1-q)(h2+a))),
the periodic solution (ξ(t),η(t)) is orbitally asymptotically stable.
Proof.
Assume that the periodic orbit Π(C0,t) starting from the point C0((1-p)h2,η0) moves to the point C1(ξ(T),η(T)) on impulsive set M2, and C1 jumps onto the point C1+(ξ(T+),η(T+)) on N2. Therefore, Π(C0,T)=C1, C1+=φ2(C1)=C0, ξ(T+)=(1-p)ξ(T), η(T+)=(1-q)η(T)+τ.
Compared with the system (6), we get
(28)P(x,y)=x(1-x)-xa+xy,Q(x,y)=y(δ-βyx),α(x,y)=-px,β(x,y)=-qy+τ,Φ(x,y)=x-h2,ξ(T)=h2,η(T)=η0-τ1-q,∂P∂x=1-2x-a(a+x)2y,∂Q∂y=δ-2βyx,∂α∂x=-p,∂α∂y=0,∂β∂x=0,∂β∂y=-q,∂Φ∂x=1,∂Φ∂y=0.
Thus,
(29)κ1=((∂β∂y·∂Φ∂x-∂β∂x·∂Φ∂y+∂Φ∂x)P+hhhh+(∂α∂x·∂Φ∂y-∂α∂y·∂Φ∂x+∂Φ∂y)Q+)×(∂Φ∂xP+∂Φ∂yQ)-1=(1-q)P(ξ(T+),η(T+))P(ξ(T),η(T))=(η0(1-p)h2+a(1-p)(1-q)hhh×(1-(1-p)h2-η0(1-p)h2+a))×(1-h2-η0-τ(1-q)(h2+a))-1,μ=κ1exp[∫0T(∂P∂x+∂Q∂y)dt]=κ1exp[∫0T(1-2x-a(a+x)2y+δ-2βyx)dt]=κ1exp{∫0T[-x+xy(a+x)2-βyx](1-x-yx+a)+(δ-βyx)HHHHHHHh-x+xy(a+x)2-βyx]dt}=κ1exp{∫0Tdxx+∫0Tdyy-∫0T(x-xy(a+x)2+βyx)dt}=(1-p)(1-q)(1-(1-p)h2-(η0/((1-p)h2+a)))1-h2-((η0-τ)/((1-q)(h2+a)))×exp{∫0Tdxx+∫0Tdyy-∫0T(x-xy(a+x)2+βyx)dt}=η0-τη0·1-(1-p)h2-(η0/((1-p)h2+a))1-h2-((η0-τ)/((1-q)(h2+a)))×exp[-∫0T(x-xy(a+x)2+βyx)dt]=κexp[-∫0T(x-xy(a+x)2+βyx)dt].
From Lemma 7, if |μ|=|κexp{-∫0T(ξ(t)-(ξ(t)η(t)/(ξ(t)+a)2)+(βη(t)/ξ(t)))dt}|<1, then the periodic solution of system (6) is orbitally asymptotically stable. This completes the proof.
Remark 19.
If |((1-(1-p)h2)-(η0/((1-p)h2+a)))/(1-h2-((η0-τ)/((1-q)(h2+a))))|≤1 and β≥1, the periodic solution of system (6) with initial point C0((1-p)h2,η0) is orbitally asymptotically stable.
5. Example and Numerical Simulation
In this part, we use numerical simulation to confirm the conclusion obtained above. Let a=0.05, δ=0.5, β=0.7, h1=0.24, h2=0.35, λ=0.18, p=0.2, q=0.2. By calculation, we obtain y~=0.22, P(0.24,0.22), Q(0.28,0.2376) and R(0.35,0.26). Then, we have an example as follows:
(30)dx(t)dt=x(t)(1-x(t))-x(t)0.05+x(t)y(t),dy(t)dt=y(t)(0.5-0.7y(t)x(t)),x≠0.24,0.35orx=0.24,y>0.22,Δx(t)=0,Δy(t)=0.18,x=0.24,y≤0.22,Δx(t)=-0.2x(t),Δy(t)=-0.2y(t)+τ,x=0.35,x(0)>0,y(0)>0.
Case 1.
Let τ=0.15, and the initial point is (0.24,0.3). From Figure 7 corresponding to Figure 3, the system exhibits a 1-periodic solution.
The periodic solution corresponding to Figure 3.
Phase diagram of system (30)
Time series of system (30)
Time series of system (30)
Case 2.
Let τ=0.1, then (0.28,0.3056) is the initial point. From Figure 8, which corresponds to Figure 4(b), the system exhibits a 1-periodic solution in the area Ω1={(x,y)∣h1<x<h2}.
There exists a 1-periodic solution in the area Ω1={(x,y)∣h1<x<h2} corresponding to Figure 4(b).
Phase diagram of system (30)
Time series of system (30)
Time series of system (30)
Case 3.
Let τ=0.15, we get the initial point (0.28,0.3056). It is easy to find that the system has no 1-periodic solution in the area Ω1={(x,y)∣h1<x<h2} from Figure 9 which corresponds to Figure 4(c).
1-periodic solution in the area Ω1={(x,y)∣h1<x<h2} corresponding to Figure 4(c) does not exist.
Phase diagram of system (30)
Time series of system (30)
Time series of system (30)
Case 4.
Let τ=0.1, the initial point (0.28,0.2376) is obtained. From Figure 10 corresponding to Figure 5(b), in the area Ω1={(x,y)∣h1<x<h2}, the system exhibits a 1-periodic solution.
There exists a 1-periodic solution in the area Ω1={(x,y)∣h1<x<h2} corresponding to Figure 5(b).
Phase diagram of system (30)
Time series of system (30)
Time series of system (30)
Case 5.
Let τ=0.15, we can easily get the initial point (0.28,0.2376). From Figure 11 corresponding to Figure 5(d), clearly, there is no 1-periodic solution in the area Ω1={(x,y)∣h1<x<h2}.
1-periodic solution in the area Ω1={(x,y)∣h1<x<h2} corresponding to Figure 5(d) does not exist.
Phase diagram of system (30)
Time series of system (30)
Time series of system (30)
Case 6.
Let τ=0.15, and the initial point is (0.28,0.45). Obviously, we can find a 1-periodic solution from Figure 12 which corresponds to Figure 6.
The periodic solution corresponding to Figure 6.
Phase diagram of system (30)
Time series of system (30)
Time series of system (30)
All the simulations above show agreement with the results in Section 4.
6. Conclusion
This paper establishes a class of integrated pest management model based on state impulse control. In the initial stage of the occurrence of crop pests, that is, the pest density satisfies x(t)≤h1, we use environment protection measures to control pests, such as releasing natural enemies. Once the pest density reaches a higher level x(t)=h2, we will adapt a combination of spraying insecticide and releasing natural enemies to control pests. With a short time to finish spraying insecticide and releasing natural enemies which bring out a sharp change in the number of pests and natural enemies, the state impulsive differential system (6) is obtained. Firstly, let the control parameters p, q, λ, τ be zero, we get Holling-Tanner ecosystem without impulsive effects. By constructing Dulac function, we discussed the stability of the positive equilibrium point E(x*,y*), and the globally asymptotically stable conditions are given for focal points and nodal point, respectively. If the control parameters p, q, λ, τ are larger than zero, the system (6) is semicontinuous pulse dynamic system. The existence, uniqueness, and stability of the periodic solutions are the research difficulties, and we need to consider all the pulse conditions (the value of h1, h2) and pulse function and its corresponding qualitative properties of the continuous dynamic system. By introducing the successor function, using impulsive differential geometry theory, we have discussed the existence of periodic solutions of the system (6) with a focus. According to the theory of impulsive differential multiplier Analogue of Poincare Criterion, the conditions of periodic solution with orbit asymptotically stable are given. Since (6) is a two-dimensional dynamical system, geometric method is intuitive and effective. How to study high-dimensional ecological dynamic systems by the geometric theory needs to be resolved in the future.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgments
Wencai Zhao, Tongqian Zhang, and Xinzhu Meng are financially supported by the National Natural Science Foundation of China (no. 11371230), the Shandong Provincial Natural Science Foundation, China (no. ZR2012AM012), and a Project of Shandong Province Higher Educational Science and Technology Program of China (no. J13LI05). Xinzhu Meng is financially supported by the SDUST Research Fund (no. 2011KYTD105).
Agricultural Information Network of Chinahttp://www.agri.gov.c/YangC. C.WangC. S.ZhengY. N.Sustained effects of Trichogramma dendrolimi on Ostrinia furnacali2011191139142BainovD. D.SimeonovP. S.1993Boca Raton, Fla, USACRC PressLakshmikanthamV.BaĭnovD. D.SimeonovP. S.19896SingaporeWorld ScientificSeries in Modern Applied MathematicsMR1082551SimeonovP. S.BaĭnovD. D.Orbital stability of periodic solutions of autonomous systems with impulse effect198819122561258510.1080/00207728808559591MR975426ZBL0669.34044BonottoE. M.LaSalles theorems in impulsive semidynamical systems20089157168BonottoE. M.FedersonM.Topological conjugation and asymptotic stability in impulsive semidynamical systems2007326286988110.1016/j.jmaa.2006.03.042MR2280949ZBL1162.37008NietoJ. J.O'ReganD.Variational approach to impulsive differential equations200910268069010.1016/j.nonrwa.2007.10.022MR2474254ZBL1167.34318NietoJ. J.Periodic boundary value problems for first-order impulsive ordinary differential equations20025171223123210.1016/S0362-546X(01)00889-6MR1926625ZBL1015.34010ChenL. S.Pest control and geometric theory of semi-continuous dynamical system201112119ChenL. S.Theory and application of semi-continuous dynamical system201334219SongX.HaoM.MengX.A stage-structured predator-prey model with disturbing pulse and time delays200933121122310.1016/j.apm.2007.10.020MR2458507ZBL1167.34372SunS.ChenL.Mathematical modelling to control a pest population by infected pests20093362864287310.1016/j.apm.2008.08.018MR2502149ZBL1205.34065TangS.ChekeR. A.Models for integrated pest control and their biological implications2008215111512510.1016/j.mbs.2008.06.008MR2459534ZBL1156.92046TangS.XiaoY.ChekeR. A.Multiple attractors of host-parasitoid models with integrated pest management strategies: eradication, persistence and outbreak20087321811972-s2.0-3914910143510.1016/j.tpb.2007.12.001LiuB.ZhangY.ChenL.The dynamical behaviors of a Lotka-Volterra predator-prey model concerning integrated pest management20056222724310.1016/j.nonrwa.2004.08.001MR2111652MailleretL.GrognardF.Global stability and optimisation of a general impulsive biological control model200922129110010.1016/j.mbs.2009.07.002MR2561136ZBL1175.92070ShiR.ChenL.The study of a ratio-dependent predator-prey model with stage structure in the prey2009581-244345110.1007/s11071-009-9491-2MR2550826ZBL1183.92083MengX.JiaoJ.ChenL.The dynamics of an age structured predator-prey model with disturbing pulse and time delays20089254756110.1016/j.nonrwa.2006.12.001MR2382398ZBL1142.34054MengX.LiZ.NietoJ. J.Dynamic analysis of Michaelis-Menten chemostat-type competition models with time delay and pulse in a polluted environment201047112314410.1007/s10910-009-9536-2MR2576641ZBL1194.92075JiaoJ.LongW.ChenL.A single stage-structured population model with mature individuals in a polluted environment and pulse input of environmental toxin20091053073308110.1016/j.nonrwa.2008.10.007MR2523269ZBL1162.92330ZhangT.MengX.SongY.The dynamics of a high-dimensional delayed pest management model with impulsive pesticide input and harvesting prey at different fixed moments2011641-211210.1007/s11071-010-9840-1MR2782953ZBL06251524ZhangT. Q.MengX. Z.SongY.ZhangT. H.A stage-structured predator-prey SI model with disease in the prey and impulsive effects2013184505528ZhangH.JiaoJ.ChenL.Pest management through continuous and impulsive control strategies20079023503612-s2.0-3454839814910.1016/j.biosystems.2006.09.038ZhangH.ChenL.NietoJ. J.A delayed epidemic model with stage-structure and pulses for pest management strategy2008941714172610.1016/j.nonrwa.2007.05.004MR2422575ZBL1154.34394JiaoJ.-J.ChenL.-S.NietoJ. J.TorresA.Permanence and global attractivity of stage-structured predator-prey model with continuous harvesting on predator and impulsive stocking on prey200829565366310.1007/s10483-008-0509-xMR2414687ZBL1231.34021TangS.XiaoY.ChenL.ChekeR. A.Integrated pest management models and their dynamical behaviour200567111513510.1016/j.bulm.2004.06.005MR2204249TangS.ChekeR. A.State-dependent impulsive models of integrated pest management (IPM) strategies and their dynamic consequences200550325729210.1007/s00285-004-0290-6MR2135823ZBL1080.92067ZengG.ChenL.SunL.Existence of periodic solution of order one of planar impulsive autonomous system2006186246648110.1016/j.cam.2005.03.003MR2176360ZBL1088.34040ZhaoL.ChenL.ZhangQ.The geometrical analysis of a predator-prey model with two state impulses20122382556410.1016/j.mbs.2012.03.011MR2947084ZBL1250.92047NieL.TengZ.HuL.PengJ.Existence and stability of periodic solution of a predator-prey model with state-dependent impulsive effects20097972122213410.1016/j.matcom.2008.11.015MR2504088ZBL1185.34123NieL.TengZ.HuL.PengJ.Qualitative analysis of a modified Leslie-Gower and Holling-type II predator-prey model with state dependent impulsive effects20101131364137310.1016/j.nonrwa.2009.02.026MR2646552ZBL1228.37058LiuB.TianY.KangB.Dynamics on a Holling II predator-prey model with state-dependent impulsive control201253126000610.1142/S1793524512600066MR2922585ZBL06254987DaiC.ZhaoM.ChenL.Homoclinic bifurcation in semi-continuous dynamic systems201256125005910.1142/S1793524512500593MR2969658FuJ. B.WangY. Z.The mathematical study of pest management strategy201220121925194210.1155/2012/251942WeiC.ChenL.Periodic solution of prey-predator model with Beddington-DeAngelis functional response and impulsive state feedback control2012201217607105MR2991600ZBL1263.9205310.1155/2012/607105WeiC.ChenL.Heteroclinic bifurcations of a prey-predator fishery model with impulsive harvesting2013651350031MR3123176ChenL. S.MengX. Z.JiaoJ. J.2009Beijing, ChinaScience PressHsuS. B.HuangT. W.Global stability for a class of predator-prey systems199555376378310.1137/S0036139993253201MR1331585ZBL0832.34035