Electronic Journal of Differential Equations, Vol. 2012 (2012), No. 150, pp. 1–13. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu MULTIPLE POSITIVE PERIODIC SOLUTIONS FOR A FOOD-LIMITED TWO-SPECIES RATIO-DEPENDENT PREDATOR-PREY PATCH SYSTEM WITH DELAY AND HARVESTING HUI FANG Abstract. By using Mawhin’s coincidence degree theory, this paper establishes some sufficient conditions on the existence of four positive periodic solutions for a food-limited two-species ratio-dependent predator-prey patch system with delay and harvesting. Some novel techniques are employed to obtain the appropriate a priori estimates. An example is given to illustrate our results. 1. Introduction Since the exploitation of biological resources and the harvest of population species are related to the optimal management of renewable resources (see [5]), many researchers pay much attention to the study of population dynamics with harvesting in mathematical bioeconomics. For example, Brauer and Soudack [1, 2] analyzed the global behaviour of some predator-prey systems under constant rate harvesting and/or stocking of both species. Xiao and Jennings [15] studied the Bogdanov-Takens bifurcations in predator-prey systems with constant rate harvesting. But all the coefficients in the system they studied are constants. Recently, some researchers studied the existence of multiple positive periodic solutions for some predator-prey systems under the assumption of periodicity of the parameters by using Mawhin’s coincidence degree theory (see [4, 7, 14, 17]). These papers only focused on predator-prey systems without diffusion. However, due to the spatial heterogeneity and unbalanced food resources, the migration phenomena of biological species can often occur between heterogeneous spatial environments or patches. On the other hand, as pointed out by Huusko and Hyvarinen in [12]: ”the dynamics of exploited populations are clearly affected by recruitment and harvesting, and the changes in harvesting induced a tendency to generation cycling in the dynamics of a freshwater fish population”. So, it is very important to describe the potential role of harvesting as an external factor in producing and maintaining the 2000 Mathematics Subject Classification. 34C25, 92D25. Key words and phrases. Coincidence degree; periodic solution; food-limited supply; predator-prey patch system; harvesting rate. c 2012 Texas State University - San Marcos. Submitted April 18, 2012. Published August 29, 2012. Supported by grants 10971085 and 11061016 from the National Natural Science Foundation of China. 1 2 H. FANG EJDE-2012/150 periodic fluctuation of population of species. Mathematically, this is equivalent to the investigation of harvesting induced periodic solutions. In this paper, we consider the following food-limited two-species ratio-dependent predator-prey patch system with delay and harvesting: h a13 (t)x1 (t)y(t) i x1 (t) a1 (t) − a11 (t)x1 (t) − k1 (t) + c1 (t)x1 (t) m(t)y 2 (t) + x21 (t) + D1 (t)[x2 (t) − x1 (t)] − H1 (t), x2 (t) a2 (t) − a22 (t)x2 (t) + D2 (t)[x1 (t) − x2 (t)] − H2 (t), x02 (t) = k2 (t) + c2 (t)x2 (t) h i a31 (t)x21 (t − τ ) y 0 (t) = y(t) − a3 (t) + , m(t)y 2 (t − τ ) + x21 (t − τ ) (1.1) where x1 and y are the population numbers of prey species x and predator species y in patch 1, and x2 is the population number of species x in patch 2. Prey dispersion is included in the model to allow for prey refuge from predation. So, the predator species y is confined to patch 1, while the prey species x can diffuse between two patches. For biological relevance of allowing for prey refuge from predation, see Yakubu [16]. a1 (t)(a2 (t)) is the natural growth rate of prey species x in patch 1 (patch 2) in the absence of predation, a3 (t) is the natural death rate of predator species y in the absence of food, a13 (t) is the death rate per encounter of prey species x due to predation, a31 (t) is the efficiency of turning predated prey species x into predator species y. ki (t) (i = 1, 2) are the population numbers of prey species in patch 1 (patch 2) at saturation, respectively. ai (t) When ci (t) 6= 0(i = 1, 2), ki (t)c (i = 1, 2) are the rate of replacement of mass i (t) in the population at saturation (including the replacement of metabolic loss and of dead organisms). The effect of delay τ refers to the dynamics of a predator being related to the predation in the past. m(t) is the half capturing saturation coefficient. aii (t) (i = 1, 2) are the density-dependent coefficients. Di (t) (i = 1, 2) are diffusion coefficients of species x. Hi (t) (i = 1, 2) denote the harvesting rates. x01 (t) = a13 (t)x1 (t)y(t) , m(t)y 2 (t) + x21 (t) a31 (t)x21 (t − τ ) m(t)y 2 (t − τ ) + x21 (t − τ ) are ratio-dependent Holling type III functional responses. When ci (t) ≡ 0 (i = 1, 2), Hi (t) ≡ 0 (i = 1, 2), Dong and Ge in [6] showed that (1.1) has at least one positive periodic solution under the appropriate conditions. When ci (t) 6= 0 (i = 1, 2), system (1.1) is a food-limited population model. For other food-limited population models, we refer to [9, 10, 11, 13] and the references cited therein. To our knowledge, a few papers have been published on the existence of multiple periodic solutions for (1.1). In this paper, we study the existence of multiple positive periodic solutions of (1.1) by using Mawhin’s coincidence degree. Since system (1.1) involves the diffusion terms and the rates of replacement, the methods used in [4, 7, 14, 17] are not available to system (1.1). Our method of defining the operator N (u, λ) facilitates the computation of Brouwer degree degB (JQN (·, 0)|ker L , Ω ∩ ker L, 0). EJDE-2012/150 MULTIPLE POSITIVE PERIODIC SOLUTIONS 3 2. Existence of four positive periodic solutions To give the proof of the main results, we first make the following preparations (see [8]). Let X, Z be normed vector spaces, L : dom L ⊂ X → Z a linear mapping, N : X × [0, 1] → Z is a continuous mapping. The mapping L will be called a Fredholm mapping of index zero if dim ker L = codim Im L < +∞ and Im L is closed in Z. If L is a Fredholm mapping of index zero, there then exist continuous projectors P : X → X and Q : Z → Z such that ImP = ker L, Im L = ker Q = Im(I − Q). If we define LP : dom L ∩ ker P → Im L as the restriction L|dom L∩ker P of L to dom L ∩ ker P , then LP is invertible. We denote the inverse of that map by KP . If Ω is an open bounded subset of X, the mapping N will be called L-compact on Ω̄ × [0, 1] if QN (Ω̄ × [0, 1]) is bounded and KP (I − Q)N : Ω̄ × [0, 1] → X is compact; i.e., continuous and such that KP (I − Q)N (Ω̄ × [0, 1]) is relatively compact. Since Im Q is isomorphic to ker L, there exists isomorphism J : Im Q → ker L. For convenience, we introduce Mawhin’s continuation theorem [8, p.29] as follows. Lemma 2.1. Let L be a Fredholm mapping of index zero and let N : Ω̄ × [0, 1] → Z be L-compact on Ω̄ × [0, 1]. Suppose (a) Lu 6= λN (u, λ) for every u ∈ dom L ∩ ∂Ω and every λ ∈ (0, 1); (b) QN (u, 0) 6= 0 for every u ∈ ∂Ω ∩ ker L; (c) Brouwer degree degB (JQN (·, 0)|ker L , Ω ∩ ker L, 0) 6= 0. Then Lu = N (u, 1) has at least one solution in dom L ∩ Ω̄. For the sake of convenience and simplicity, we denote Z 1 T g(t)dt, g l = min g(t), g u = max g(t), ḡ = T 0 t∈[0,T ] t∈[0,T ] where g is a nonnegative continuous T -periodic function. Set ( ak13 )u 1 b1 = √ , b2 = 0. 2 ml For the rest of this article, we assume the following: (A1) τ > 0, ki (t) (i = 1, 2), ai (t) (i = 1, 2, 3), aii (t) (i = 1, 2), a13 (t), a31 (t), m(t), Di (t) (i = 1, 2), Hi (t) (i = 1, 2) are positive continuous T -periodic functions, ci (t) (i = 1, 2) are nonnegative continuous T -periodic functions. (A2) al31 − ā3 > 0. q a1 u a2 u ) ,( ) , N1 = max ( a11 a22 (A3) kil ( akii )l > Diu + bi kil +cu i N1 Hil > Diu N1 (i = 1, 2). +2 ( akiii )u Hiu (i = 1, 2). (A4) For further convenience, we introduce the 12 positive numbers: r 4kl ai u ( ki ) ± [( akii )u ]2 − kl +cui N1 ( akiii )l (Hil − Diu N1 ) i i li± = , 2kil aii l ( ) l u k ki +ci N1 i r l l k k [ kl +ciu N1 ( akii )l − Diu − bi ] ± [ kl +ciu N1 ( akii )l − Diu − bi ]2 − 4( akiii )u Hiu i i i i ± ui = , aii u 2( ki ) 4 H. FANG x± i = ( akii ) ± EJDE-2012/150 q [( akii )]2 − 4( akiii )H̄i 2( akiii ) , i = 1, 2. It is not difficult to show that − + + + li− < x− i < ui < ui < xi < li (i = 1, 2) (2.1) Now, we are ready to state the main result of this article. Theorem 2.2. Assume that (A1)-(A4) hold, then (1.1) has at least four positive T -periodic solutions. Proof. Since we are concerning with positive solutions of (1.1), we make the change of variables, xj (t) = euj (t) (j = 1, 2), y(t) = eu3 (t) . Then (1.1) is rewritten as h a13 (t)eu1 (t) eu3 (t) i 1 u1 (t) a (t) − a (t)e − u01 (t) = 1 11 k1 (t) + c1 (t)eu1 (t) m(t)e2u3 (t) + e2u1 (t) H1 (t) eu2 (t) − 1] − u (t) , u (t) 1 e e 1 (2.2) 1 eu1 (t) H2 (t) 0 u2 (t) u2 (t) = [a2 (t) − a22 (t)e ] + D2 (t)[ u (t) − 1] − u (t) , k2 (t) + c2 (t)eu2 (t) e 2 e 2 + D1 (t)[ u03 (t) = −a3 (t) + a31 (t)e2u1 (t−τ ) . m(t)e2u3 (t−τ ) + e2u1 (t−τ ) Take X = Z = {u = (u1 , u2 , u3 )T ∈ C(R, R3 ) : ui (t + T ) = ui (t), i = 1, 2, 3} and define kuk = max |u1 (t)| + max |u2 (t)| + max |u3 (t)|, t∈[0,T ] t∈[0,T ] t∈[0,T ] T for u = (u1 , u2 , u3 ) in X or in Z. Equipped with the above norm k · k, it is easy to verify that X and Z are Banach spaces. Set h k (t) + (1 − λ)c (t)eu1 (t) ih a (t) a (t)eu1 (t) 1 1 1 11 ∆1 (u, t, λ) = − k1 (t) k1 (t) k1 (t) + c1 (t)eu1 (t) −1 u1 (t) u3 (t) i H1 (t) λk1 (t) a13 (t)e e eu2 (t) − + λD1 (t) u (t) − 1 − u (t) , 2u (t) 2u (t) 3 1 1 m(t)e +e e e 1 h k (t) + (1 − λ)c (t)eu2 (t) ih a (t) a (t)eu2 (t) i 22 2 2 2 − ∆2 (u, t, λ) = k2 (t) k2 (t) k2 (t) + c2 (t)eu2 (t) u (t) 1 e H2 (t) + λD2 (t) u (t) − 1 − u (t) , e 2 e 2 a31 (t)e2u1 (t−τ ) ∆3 (u, t, λ) = −a3 (t) + . 2u m(t)e 3 (t−τ ) + λe2u1 (t−τ ) For any u ∈ X, because of the periodicity, we can easily check that ∆i (u, t, λ) ∈ C(R2 , R) (i = 1, 2, 3) are T -periodic in t. Let L : dom L = {u ∈ X : u ∈ C(R, R3 )} 3 u 7→ u0 ∈ Z, EJDE-2012/150 MULTIPLE POSITIVE PERIODIC SOLUTIONS T 1 T Z 1 Q : Z 3 u 7→ T Z P : X 3 u 7→ 5 u(t)dt ∈ X, 0 T u(t)dt ∈ Z, 0 N : X × [0, 1] 3 (u, λ) 7→ (∆1 (u, t, λ), ∆2 (u, t, λ), ∆3 (u, t, λ))T ∈ Z. Here, for any k ∈ R3 , we also identify it as the constant function in X or Z with the constant value k. It is easy to see that Z T ker L = R3 , Im L = {u ∈ X : ui (t)dt = 0, i = 1, 2, 3} 0 is closed in Z, dim ker L = codim Im L = 3, and P, Q are continuous projectors such that ImP = ker L, Im L = ker Q = Im(I − Q). Therefore, L is a Fredholm mapping of index zero. On the other hand, Kp : Im L 7→ dom L ∩ KerP has the form Z Z Z t 1 T t u(s) ds dt. Kp (u) = u(s)ds − T 0 0 0 Thus, Z Z T 1 Z T 1 T 1 T ∆1 (u, t, λ)dt, ∆2 (u, t, λ)dt, ∆3 (u, t, λ)dt , QN (u, λ) = T 0 T 0 T 0 Kp (I − Q)N (u, λ) = (Φ1 (u, t, λ), Φ2 (u, t, λ), Φ3 (u, t, λ))T , where Z t ∆j (u, s, λ)ds − Φj (u, t, λ) = 0 t 1 −( − ) T 2 Z 1 T Z T Z t ∆j (u, s, λ) ds dt 0 0 T ∆j (u, s, λ)ds, j = 1, 2, 3. 0 Obviously, QN and Kp (I − Q)N are continuous. By the Arzela-Ascoli theorem, it is not difficult to show that the closure of Kp (I − Q)N (Ω × [0, 1]) is compact for any open bounded set Ω ⊂ X. Moreover, QN (Ω̄ × [0, 1]) is bounded. Thus N is L-compact on Ω̄ × [0, 1] with any open bounded set Ω ⊂ X. To apply Lemma 2.1, we need to find at least four appropriate open, bounded subsets Ω1 , Ω2 , Ω3 , Ω4 in X. Corresponding to the operator equation Lu = λN (u, λ), λ ∈ (0, 1), we have h k (t) + (1 − λ)c (t)eu1 (t) ih a (t) a (t)eu1 (t) 1 1 1 11 u01 (t) = λ − k1 (t) k1 (t) k1 (t) + c1 (t)eu1 (t) (2.3) −1 u1 (t) u3 (t) i λH1 (t) λk1 (t) a13 (t)e e eu2 (t) 2 + λ D1 (t) u (t) − 1 − u (t) , − m(t)e2u3 (t) + e2u1 (t) e 1 e 1 h k (t) + (1 − λ)c (t)eu2 (t) ih a (t) a (t)eu2 (t) i 2 2 2 22 u02 (t) = λ − k2 (t) k2 (t) k2 (t) + c2 (t)eu2 (t) (2.4) h eu1 (t) i λH (t) 2 2 + λ D2 (t) u (t) − 1 − u (t) , e 2 e 2 h i a31 (t)e2u1 (t−τ ) u03 (t) = λ − a3 (t) + . (2.5) m(t)e2u3 (t−τ ) + λe2u1 (t−τ ) 6 H. FANG EJDE-2012/150 Suppose that (u1 (t), u2 (t), u3 (t))T is a T -periodic solution of (2.3), (2.4) and m (2.5) for some λ ∈ (0, 1). Choose tM i , ti ∈ [0, T ], i = 1, 2, 3, such that ui (tM i ) = max ui (t), t∈[0,T ] ui (tm i ) = min ui (t), i = 1, 2, 3. t∈[0,T ] Then, it is clear that u0i (tM i ) = 0, u0i (tm i ) = 0, i = 1, 2, 3. From this and (2.3), (2.4), we obtain that ih a (tM ) − a (tM )eu1 (tM h k (tM ) + (1 − λ)c (tM )eu1 (tM 1 ) 1 ) 1 1 11 1 1 1 1 1 0= M u1 (tM 1 ) k1 (tM k1 (tM 1 ) 1 ) + c1 (t1 )e M M i M M −1 M u1 (t1 ) u3 (t1 ) H1 (tM eu2 (t1 ) λk1 (t1 ) a13 (t1 )e ) e M − − 1 − u (t1M ) , + λD (t ) 1 M M M 1 M 2u (t ) 2u (t ) u (t ) 3 1 1 1 1 1 1 1 m(t1 )e +e e e (2.6) h k (tM ) + (1 − λ)c (tM )eu2 (tM ih a (tM ) a (tM )eu2 (tM i 2 ) 2 ) 2 2 2 2 2 2 22 2 0= − M u2 (tM 2 ) k2 (tM k2 (tM k2 (tM 2 ) 2 ) 2 ) + c2 (t2 )e (2.7) M H2 (tM eu1 (t2 ) 2 ) M + λD2 (t2 ) u (tM ) − 1 − u (tM ) , e 2 2 e 2 2 h k (tm ) + (1 − λ)c (tm )eu1 (tm ih a (tm ) − a (tm )eu1 (tm 1 ) 1 ) 11 1 1 1 1 1 1 1 0= m) m m m u (t 1 k1 (t1 ) k1 (t1 ) + c1 (t1 )e 1 m i m m u1 (tm ) u (t ) m −1 3 u2 (t1 ) H1 (tm λk1 (t1 ) a13 (t1 )e 1 e 1 ) m e − + λD (t ) − 1 − u (t1m ) , 1 m m m 1 2u3 (t1 ) + e2u1 (t1 ) u1 (t1 ) 1 1 m(tm e e )e 1 (2.8) i ih a (tm ) a (tm )eu2 (tm h k (tm ) + (1 − λ)c (tm )eu2 (tm 2 ) 2 ) 2 2 2 2 2 2 22 2 0= m) − m )eu2 (tm 2 ) k (t k2 (tm ) + c (t k2 (tm 2 2 2 2 ) 2 2 (2.9) m m u1 (t2 ) H2 (t2 ) m e + λD2 (t2 ) u (tm ) − 1 − u (tm ) . e 2 2 e 2 2 For u(tM ) (i = 1, 2), there are two cases to consider. i M M ) ≥ u (t ), then u1 (tM Case 1. Assume that u1 (tM 2 2 1 1 ) ≥ u2 (t1 ). From this and (2.6), we have M u1 (tM 1 ) > 0, a1 (tM 1 ) − a11 (t1 )e which implies M a1 u a1 (tM 1 ) ) ≤ N1 . eu1 (t1 ) < ≤( M a a11 (t1 ) 11 That is a1 u M u2 (tM ) ≤ ln N1 . 2 ) ≤ u1 (t1 ) < ln( a11 M M M Case 2. Assume that u1 (tM 1 ) < u2 (t2 ), then u2 (t2 ) > u1 (t2 ). From this and (2.7), we have M u2 (tM 2 ) > 0, a2 (tM 2 ) − a22 (t2 )e which implies M a2 u a2 (tM 2 ) ≤( ) ≤ N1 . eu2 (t2 ) < a a22 (tM ) 22 2 That is, a2 u M u1 (tM ) ≤ ln N1 . 1 ) < u2 (t2 ) < ln( a22 EJDE-2012/150 MULTIPLE POSITIVE PERIODIC SOLUTIONS 7 Therefore, M max{u1 (tM 1 ), u2 (t2 )} < ln N1 . (2.10) It follows from (2.6) that ih a (tM ) a (tM )eu1 (tM h k (tM ) + (1 − λ)c (tM )eu1 (tM i H (tM ) 1 ) 1 ) 1 1 1 1 1 1 1 11 1 − u (t1M ) − M) M) M )eu1 (tM 1 1 1 ) k (t k (t k1 (tM ) + c (t e 1 1 1 1 1 1 1 M M h k (tM ) + (1 − λ)c (tM )eu1 (tM i M −1 M u (t ) ) u (t 1 k1 (t1 ) a13 (t1 )e 1 1 e 3 1 ) 1 1 1 1 =λ M M u1 (tM 2u3 (tM 1 ) 1 ) + e2u1 (t1 ) k1 (tM m(tM 1 ) + c1 (t1 )e 1 )e − u2 (tM 1 ) M e λD1 (t1 ) u (tM ) 1 e 1 + λD1 (tM 1 ). Therefore, h k (tM ) + (1 − λ)c (tM )eu1 (tM ih a i 1 ) a11 u u1 (tM H1u 1 1 1 1 1 l 1 ) − ) − ( ) e ( M M) M M u u (t k1 k1 k1 (t1 ) + c1 (t1 )e 1 1 e 1 (t1 ) h k (tM ) + (1 − λ)c (tM )eu1 (tM i 1 ) ) a13 (tM 1 1 1 1 p1 < + D1 (tM M) 1 ), M M u (t M 1 k1 (t1 ) + c1 (t1 )e 1 2k1 (t1 ) m(tM ) 1 which implies i ih a h k (tM ) + (1 − λ)c (tM )eu1 (tM 1 ) a11 u u1 (tM H1u 1 1 1 1 1 l 1 ) − ) − ( ) e ( M) M M M u (t u k1 k1 k1 (t1 ) + c1 (t1 )e 1 1 e 1 (t1 ) i ( a13 )u h k (tM ) + (1 − λ)c (tM )eu1 (tM 1 ) 1 1 1 1 k1 √ + D1u . < M )eu1 (tM l 1 ) ) + c (t k1 (tM 2 m 1 1 1 From this and noticing that M M u1 (t1 k1 (tM k1 (tM 1 ) 1 ) + (1 − λ)c1 (t1 )e ≤ M u1 (tM M u1 (tM 1 ) 1 ) k1 (tM k1 (tM 1 ) + c1 (t1 )e 1 ) + c1 (t1 )e ) ≤ 1, we have ( ak13 )u a11 u u1 (tM H1u k1 (tM a1 l 1 ) 1 1 ) − √ ) − ( ) e + D1u . ( < M )eu1 (tM u1 (tM l k 1 ) k1 1 ) k1 (tM ) + c (t e 1 2 m 1 1 1 Therefore, we have ( ak13 )u k1l a1 l a11 u u1 (tM H1u 1 1 ) − √ ( ) − ( ) e < + D1u ; M k1 k1l + cu1 N1 k1 eu1 (t1 ) 2 ml that is, ( h i M k1l a1 a11 u 2u1 (tM 1 ) − ) e ( )l − D1u − b1 eu1 (t1 ) + H1u > 0. l u k1 k1 + c1 N1 k1 From (A3) and the above inequality, we have + u1 (tM 1 ) > ln u1 − or u1 (tM 1 ) < ln u1 (2.11) Similarly, we have + − u1 (tm or u1 (tm 1 ) > ln u1 1 ) < ln u1 By a similar argument, from (2.7), it follows that h i M k2l a2 a22 u 2u2 (tM 2 ) − ) e ( )l − D2u − b2 eu2 (t2 ) + H2u > 0. ( l u k2 k2 + c2 N1 k2 (2.12) 8 H. FANG EJDE-2012/150 By (A3) and the above inequality, we have + u2 (tM 2 ) > ln u2 − or u2 (tM 2 ) < ln u2 . (2.13) + u2 (tm 2 ) > ln u2 − or u2 (tm 2 ) < ln u2 . (2.14) Similarly, we have Again, from (2.6) it follows that ih a (tM ) a (tM )eu1 (tM i h k (tM ) + (1 − λ)c (tM )eu1 (tM 1 ) 1 ) 1 1 1 1 11 1 1 1 − M M u1 (t1 ) k1 (tM k1 (tM k1 (tM 1 ) 1 ) 1 ) + c1 (t1 )e M u2 (t1 ) H1 (tM 1 ) M e − D (t ) 1 M M . 1 eu1 (t1 ) eu1 (t1 ) Hence, we have > k1l k1l a11 (tM a1 (tM 1 ) 2u1 (tM 1 ) u1 (tM M u2 (tM 1 ) − 1 ) < 0, e e 1 ) + H1 (tM 1 ) − D1 (t1 )e M M u k1 (t1 ) + c1 N1 k1 (t1 ) which implies k1l a1 u u1 (tM a11 l 2u1 (tM k1l 1 ) − ( )e ) e 1 ) + H1l − D1u N1 < 0. ( k1 + cu1 N1 k1 From (A3), (A4) and the above inequality, we have + ln l1− < u1 (tM 1 ) < ln l1 . (2.15) + ln l1− < u1 (tm 1 ) < ln l1 . (2.16) Similarly, we have By a similar argument, it follows from (2.7) that k2l a22 l 2u2 (tM a2 u u2 (tM 2 ) − ( ( )e ) e 2 ) + H2l − D2u N1 < 0. u l k k2 k2 + c2 N1 2 From (A3), (A4) and the above inequality, we have + ln l2− < u2 (tM 2 ) < ln l2 . (2.17) + ln l2− < u2 (tm 2 ) < ln l2 . (2.18) Similarly, we have It follows from (2.11), (2.12), (2.15), (2.16) that − − + + u1 (tM 1 ) ∈ (ln l1 , ln u1 ) ∪ (ln u1 , ln l1 ), − − + + u1 (tm 1 ) ∈ (ln l1 , ln u1 ) ∪ (ln u1 , ln l1 ). It follows from (2.13), (2.14), (2.17), (2.18) that − − + + u2 (tM 2 ) ∈ (ln l2 , ln u2 ) ∪ (ln u2 , ln l2 ), − − + + u2 (tm 2 ) ∈ (ln l2 , ln u2 ) ∪ (ln u2 , ln l2 ). From (2.5), we have Z T Z a3 (t)dt = 0 0 T a31 (t)e2u1 (t−τ ) dt. m(t)e2u3 (t−τ ) + λe2u1 (t−τ ) (2.19) EJDE-2012/150 Therefore, Z T MULTIPLE POSITIVE PERIODIC SOLUTIONS |u03 (t)|dt < 0 T Z Z T a3 (t)dt + 0 0 T Z = Z 0 a31 (t)e2u1 (t−τ ) dt m(t)e2u3 (t−τ ) + λe2u1 (t−τ ) T a3 (t)dt + 9 a3 (t)dt (2.20) 0 < 2T ā3 := d3 . By (2.19), there must be η ∈ (0, T ) such that ā3 = a31 (η)e2u1 (η−τ ) . m(η)e2u3 (η−τ ) + λe2u1 (η−τ ) (2.21) From (2.21), we have e2u3 (η−τ ) = a31 (η) − λā3 2u1 (η−τ ) e . ā3 m(η) Therefore, au31 2 N := p, ā3 ml 1 al − ā3 > 31 u (l1− )2 := q. ā3 m e2u3 (η−τ ) < e2u3 (η−τ ) (2.22) (2.23) Since u3 (t) is a T -periodic function, there exists η1 ∈ [0, T ] such that u3 (η1 ) = u3 (η − τ ). Again, it follows from (2.20), (2.22), (2.23) that for any t ∈ [0, T ], we have Z t 1 u3 (t) = u3 (η1 ) + u03 (s)ds < ln p + d3 , 2 η1 Z t 1 u3 (t) = u3 (η1 ) + u03 (s)ds > ln q − d3 . 2 η1 Set 1 1 H = max{| ln p + d3 |, | ln q − d3 |}. 2 2 Then max |u3 (t)| < H. (2.24) t∈[0,T ] Clearly, li± , u± i (i = 1, 2), H are independent of λ. Now, let us consider QN (u, 0) with u = (u1 , u2 , u3 )T ∈ R3 . Note that H̄1 u1 ( ak11 ) − ( ak11 )e − u1 e 1 QN (u, 0) = ( ak22 ) − ( ak22 )eu2 − eH̄u22 . 2 2u1 31 e −ā3 + ( am ) e2u3 Letting QN (u, 0) = 0, we have a1 a11 u1 H̄1 )−( )e − u1 = 0, k1 k1 e a2 a22 u2 H̄2 ( )−( )e − u2 = 0, k2 k2 e ( (2.25) (2.26) 10 H. FANG −ā3 + ( EJDE-2012/150 a31 e2u1 ) = 0. m e2u3 (2.27) From (2.25), we have a11 2u1 a1 )e − ( )eu1 + H̄1 = 0, (2.28) k1 k1 which implies that (2.28) has two distinct roots ln x± 1 . From (2.26), we have a22 2u2 a2 u2 ( )e − ( )e + H̄2 = 0, (2.29) k2 k2 ( which implies that (2.29) has two distinct roots ln x± 2 . Again, it follows from (2.27) that s 31 ( am ) ± x . x± = 3 ā3 1 Therefore, QN (u, 0) = 0 has four distinct solutions. + + T ũ1 = (ln x+ 1 , ln x2 , ln x3 ) , − + T ũ2 = (ln x+ 1 , ln x2 , ln x3 ) , + − T ũ3 = (ln x− 1 , ln x2 , ln x3 ) , − − T ũ4 = (ln x− 1 , ln x2 , ln x3 ) . Choose C > 0 such that s n C > max | ln 31 ( am ) + x |, | ln ā3 1 s 31 ( am ) −o x | . ā3 1 Let n + Ω1 = u = (u1 , u2 , u3 )T ∈ X : max u1 (t) ∈ (ln u+ 1 , ln l1 ), t∈[0,T ] + max u2 (t) ∈ (ln u+ 2 , ln l2 ), o + min u2 (t) ∈ (ln u+ 2 , ln l2 ), max |u3 (t)| < H + C , min u1 (t) ∈ t∈[0,T ] + (ln u+ 1 , ln l1 ), t∈[0,T ] t∈[0,T ] t∈[0,T ] n + Ω2 = u = (u1 , u2 , u3 )T ∈ X : max u1 (t) ∈ (ln u+ 1 , ln l1 ), t∈[0,T ] max u2 (t) ∈ (ln l2− , ln u− 2 ), o min u2 (t) ∈ (ln l2− , ln u− 2 ), max |u3 (t)| < H + C , min u1 (t) ∈ t∈[0,T ] + (ln u+ 1 , ln l1 ), t∈[0,T ] t∈[0,T ] t∈[0,T ] n Ω3 = u = (u1 , u2 , u3 )T ∈ X : max u1 (t) ∈ (ln l1− , ln u− 1 ), t∈[0,T ] (ln l1− , ln u− 1 ), + max u2 (t) ∈ (ln u+ 2 , ln l2 ), o + min u2 (t) ∈ (ln u+ , ln l ), max |u (t)| < H + C , 3 2 2 min u1 (t) ∈ t∈[0,T ] t∈[0,T ] t∈[0,T ] t∈[0,T ] n Ω4 = u = (u1 , u2 , u3 )T ∈ X : max u1 (t) ∈ (ln l1− , ln u− 1 ), t∈[0,T ] max u2 (t) ∈ (ln l2− , ln u− 2 ), o min u2 (t) ∈ (ln l2− , ln u− 2 ), max |u3 (t)| < H + C . min u1 (t) ∈ t∈[0,T ] t∈[0,T ] (ln l1− , ln u− 1 ), t∈[0,T ] t∈[0,T ] (2.30) EJDE-2012/150 MULTIPLE POSITIVE PERIODIC SOLUTIONS 11 Then Ω1 , . . . , Ω4 are bounded open subsets of X. It follows from (2.1) and (2.30) that ũi ∈ Ωi (i = 1, 2, 3, 4). It is easy to see that Ω̄i ∩ Ω̄j = ∅ (i, j = 1, 2, 3, 4, i 6= j) and Ωi satisfies (a) in Lemma 2.1 for i = 1, 2, 3, 4. Moreover, QN (u, 0) 6= 0 for u ∈ ∂Ωi ∩ ker L. Set c h(x) = b − ax − , x ∈ (0, +∞), x √ where a, b, c are positive constants. When b > 2 ac, it is easy to see that h(x) = 0 has exactly two positive solutions √ √ b − b2 − 4ac b + b2 − 4ac − + x = , x = 2a 2a 0 − 0 + such that h (x ) > 0, h (x ) < 0. From this, a direct computation gives deg{JQN (·, 0), Ω1 ∩ ker L, 0} = −1, deg{JQN (·, 0), Ω3 ∩ ker L, 0} = 1, deg{JQN (·, 0), Ω2 ∩ ker L, 0} = 1, deg{JQN (·, 0), Ω4 ∩ ker L, 0} = −1. Here, J is taken as the identity mapping since Im Q = ker L. So far we have proved that Ωi satisfies all the assumptions in Lemma 2.1. Hence, (2.2) has at least four T periodic solutions (ui1 (t), ui2 (t), ui3 (t))T (i = 1, 2, 3, 4) and (ui1 , ui2 , ui3 )T ∈ dom L∩Ω̄i . i Obviously, (ui1 , ui2 , ui3 )T (i = 1, 2, 3, 4) are different. Let xij (t) = euj (t) (j = 1, 2), i y i (t) = eu3 (t) (i = 1, 2, 3, 4). Then (xi1 (t), xi2 (t), y i (t))T (i = 1, 2, 3, 4) are four different positive T -periodic solutions of (1.1). The proof is complete. Theorem 2.3. In addition to (A1), (A2), (A4), assume further that (1.1) satisfies kl 1 i (A3)* Hiu < NN ( ai )l − ( akiii )u − bi (i = 1, 2). u l 1 +1 k +c N1 ki i i Then (1.1) has at least four positive T -periodic solutions. Proof. It suffices to verify (A3) in Theorem 2.2. Indeed, it follows from (A3)* and (A4) that r ai aii Hl aii kl Diu + bi + 2 ( )u Hiu < i + bi + ( )u + Hiu < l iu ( )l , ki N1 ki ki + ci N1 ki for i = 1, 2. Finally, we describe the biological meaning of (A1)–(A4) and Theorem 2.2. In realistic world, the environment is always varying periodically with time. This motivates us to consider system (1.1) under the condition (A1). (A2) indicates that the efficiency of turning predated prey species x into predator species y is higher than the natural death rate of predator species y,qwhich is the necessary condition for the survival of predator species y. Note that ( akiii )u Hiu is the geometric mean of ( akiii )u and Hiu (i = 1, 2), which describes the mean effect of the intra-species competition and harvesting. Hence, (A3) indicates that the natural growth rate of prey species x with a food-limited supply in patch 1 is higher than the decay rate due to the intra-species competition, predation, dispersion and harvesting, and that the natural growth rate of prey species x with a limited food supply in patch 2 is higher than the decay rate due to the intra-species competition, the dispersion and harvesting. Since N1 is the environmental carrying capacity of prey species x, (A4) indicates that the effect of harvesting on populations is stronger than the effect of the dispersion during each time period. From the ecological viewpoint, Theorem 12 H. FANG EJDE-2012/150 2.2 indicates that a food-limited two-species ratio-dependent predator-prey patch systems with delay and harvesting can lead to four different periodic fluctuations in a periodic environment if the regulated harvesting on prey populations is made according to (A1)–(A4). 3. An Example Take τ = 2, T = 2, k1 (t) = 4 + sin(πt), k2 (t) = 3 + sin(πt), ci (t) = 0.2 + 0.05 sin(πt) (i = 1, 2), 1 + sin2 (πt) 1 + sin2 (πt) , D1 (t) = D2 (t) = , 10 200 (4 + sin(πt))2 (4 + sin(πt))2 a1 (t) = (4 + sin(πt))2 , a11 (t) = , a13 (t) = , 4 4 (3 + sin(πt))2 , m(t) = 1 + sin2 (πt), a2 (t) = (3 + sin(πt))2 , a22 (t) = 4 a31 (t) = 4 + sin(πt), a3 (t) = 2 + sin(πt). H1 (t) = 1 + sin2 (πt) , 24 H2 (t) = Then we have k1l = 3, k2l = 2, cui = 0.25 (i = 1, 2), 1 1 1 1 1 H1u = , H2u = , H1l = , H2l = , D1u = D2u = , 12 5 24 10 100 a1 a11 u 5 a13 u 5 N1 = 4, ( )l = 3, ( ) = , ( ) = , ml = 1, k1 k1 4 k1 4 a22 u 5 a2 l ) = 1, al31 = 3, ā3 = 2. b1 = , ( ) = 2, ( 8 k2 k2 It is easy to see that the conditions in Theorem 2.3 are satisfied. By Theorem 2.3, system (1.1) has at least four positive 2-periodic solutions. References [1] F. Brauer, A. Soudack; Coexistence properties of some predator-prey systems under constant rate harvesting and stocking, J. Math. Biol. 12 (1981) 101-114. [2] F. Brauer, A. Soudack; On constant effort harvesting and stocking in a class of predator-prey systems, J. Theoret. Biol. 95 (1982) 247-252. [3] F. Chen, D. Sun, J. Shi; Periodicity in a food-limited population model with toxicants and state dependent delays, J. Math. Anal. Appl. 288 (2003) 136-146. [4] Y. Chen; Multiple periodic solutions of delayed predator-prey systems with type IV functional responses, Nonlinear Anal. Real World Appl. 5 (2004) 45-53. [5] C. Clark; Mathematical Bioeconomics, The optimal management of renewable resources, 2nd Ed. John Wiley & Sons, New York-Toronto, USA, 1990. [6] S. Dong, W. Ge; Periodic solutions for two-species ratio-dependent predator-prey patch systems with delay, Acta Math. Appl. Sin. 27 (2004) 132-141. [7] J. Feng, S. Chen; Four periodic solutions of a generalized delayed predator-prey system, Appl. Math. Comput. 181 (2006) 932-939. [8] R. Gaines, J. Mawhin; Coincidence Degree and Nonlinear Differetial Equitions, Springer Verlag, Berlin, 1977. [9] K. Gopalsamy, M. Kulenovic, G. Ladas; Environmental periodicity and time delays in a food-limited population model, J. Math. Anal. Appl. 147 (1990) 545-555. [10] S. Gourley, M. Chaplain; Travelling fronts in a food-limited population model with time delay, Proc. R. Soc. Lond. Ser. A Math. Phys. Eng. Sci. 132A (2002) 75-89. [11] S. Gourley, J. So; Dynamics of a food-limited population model incorporating non local delays on a finite domain, J. Math. Biol. 44 (2002) 49-78. [12] A. Huusko, P. Hyvarinen; A high harvest rate induces a tendency to generation cycling in a freshwater fish population, Journal of Animal Ecology, 74(3), 2005, 525-531. EJDE-2012/150 MULTIPLE POSITIVE PERIODIC SOLUTIONS 13 [13] Y. Li; Periodic solutions of periodic generalized food limited model, Int. J. Math. Math. Sci. 25 (2001) 265-271. [14] Y. Xia, J. Cao, S. Cheng; Multiple periodic solutions of a delayed stage-structured predatorprey model with non-monotone functional responses, Appl. Math. Modelling 31 (2007) 19471959. [15] D. Xiao and L. Jennings; Bifurcations of a ratio-dependent predator-prey system with constant rate harvesting, SIAM J. Appl. Math. 65 (2005) 737-753. [16] A. Yakubu; Searching predator and prey dominance in discrete predator-prey systems with dispersion, SIAM J. Appl. Math. 61 (2000) 870-888. [17] Z. Zhang, T. Tian; Multiple positive periodic solutions for a generalized predator-prey system with exploited terms, Nonlinear Anal. Real World Appl. 9 (2007) 26-39. Hui Fang Department of Mathematics, Kunming University of Science and Technology, Kunming, Yunnan 650500, China E-mail address: kmustfanghui@hotmail.com