Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Dynamics of a predator-prey model with generalized Holling type functional response and mutual interference

Dynamics of a predator-prey model with generalized Holling type functional response and mutual... DYNAMICS OF A PREDATOR-PREY MODEL WITH GENERALIZED HOLLING TYPE FUNCTIONAL RESPONSE AND MUTUAL INTERFERENCE 1 2 3 Kwadwo Antwi-Fordjour , Rana D. Parshad , Matthew A. Beauregard 1) Department of Mathematics and Computer Science, Samford University, Birmingham, AL 35229, USA 2) Department of Mathematics, Iowa State University, Ames, IA 50011, USA 3) Department of Mathematics and Statistics, Stephen F. Austin State University, Nacogdoches, TX 75962, USA Abstract. Mutual interference and prey refuge are important drivers of predator- prey dynamics. The \exponent" or degree of mutual interference has been un- der much debate in theoretical ecology. In the present work, we investigate the interplay of the mutual interference exponent, on the behavior of a predator- prey model with a generalized Holling type functional response. We investigate stability properties of the system and derive conditions for the occurrence of saddle-node and Hopf-bifurcations. A sucient condition for extinction of the prey species has also been derived for the model. In addition, we investigate the e ect of a prey refuge on the population dynamics of the model and derive conditions for the prey refuge that would yield persistence of populations. We provide additional veri cation our analytical results via numerical simulations. Our ndings are in accordance with classical experimental results in ecology [23], that show that extinction of predator and prey populations is possible in a nite time period - but that bringing in refuge can e ectively cause persistence. 1. Introduction Predator-prey dynamics form the corner stone of ecosystems. Mathematical models for such interactions goes back to the work of Lokta, Volterra, Holling and Gause [21, 23, 24, 26]. Holling's classical work proposes that a predators feeding rate depends solely on the prey density, and is modeled essentially by a saturating f(x) function called the functional response, described via p(x) = , where h is the 1+hf(x) handling time of one prey item and f (x) is a function of prey density x. Typically f is smooth, making the response p smooth, and depending on the form of f we f(x) have Holling type II, III, IV responses [16, 24]. The response, p(x) = , can 1+hf(x) (f(x)) be considered a special form of 2 di erent response types. p(x) = , or 1+h(f(x)) f(x) p(x) = . When p = 1, or m = 1, we recover the classical response 1+hf(x) posed earlier. 2010 Mathematics Subject Classi cation. 34D45; 37C10; 37C75; 37G15; 92B05. Key words and phrases. generalized interference; stability analysis; nite time extinction; Hopf-bifurcation; prey refuge. arXiv:2003.02712v1 [math.DS] 5 Mar 2020 2 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD An interesting subclass of these general responses are the cases when 0 < p < 1 or 0 < m < 1. In these cases p(x) is non-smooth, causing various diculties in the mathematical analysis of these systems. For example, linearization about the trivial steady state is no longer possible [32]. Such responses were considered by Sugie [32, 33] and more recently by Braza [27]. However these works miss a key dynamic inherent in such models - that of nite time extinction. Note, in these cases one might ask what real ecological scenarios do these models represent. The work by Sugie, proposed the (f (x)) term, as indicative of a predator which is highly ecient and with a high attack rate. Mutual interference is de ned as the behavioral interactions among feeding or- ganisms, that reduce the time that each individual spends obtaining food, or the amount of food each individual consumes [1, 2, 22, 30]. Some of the earliest work on mutual/predator interference, was initiated by Erbe [25], in which the mutual in- f(x) terference is modeled as y , where y is the predator density and 0 < m < 1. 1+hf(x) The exact value of the exponent m has been under much debate in ecology [18]. f(x) Various authors describe the response p(x) = in terms of mutual 1+hf(x) interference. This direction was rst considered by Upadhyay and Rao [31]. How- ever, an ecological motivation, to the best of our knowledge is not provided. We are motivated by certain theoretical ecology directions [12, 13, 14], and interpret f(x) p(x) = , 0 < m < 1, as a predator with a greater feeding rate, or a 1+hf(x) more aggressive predator, than one which is modeled in the classical scenario - that is when m = 1. This is clear from simple comparison, p(x)j < p(x)j , 1 m =1 0<m <1 1 1 8x > 0. Prey refuge, and its role in predator-prey communities has also been extremely well investigated, since the seminal work of Kar [17]. Refuge is de ned as any strategy taken by prey to avoid predation, such as shelter, dispersal, mimicry and camou age [34]. It can have strong in uence on predator-prey communities [20, 19] - often stabilizing systems, which are otherwise doomed for extinction. It is thus an important ingredient in ecosystem balance and diversity [28]. However, the e ect of refuge on non-smooth systems such as the a ore mentioned ones, remains less investigated [28]. The well known experiments of Gause nd in contradiction to the predictions of classical predator-prey models, that there is a distinct chance for the predator and prey populations to die out - unless the prey is provided with refuge [23]. Non-smooth systems such as when 0 < m; m < 1, in the a ore mentioned models, enable the dynamic of nite time predator-prey extinction (such as seen in the experiments of Gause [23, 19]) - however, to the best of our knowledge, the e ect of prey refuge on these systems has not been investigated. For the purposes of this manuscript we consider the functional response p(x) = f(x) , and de ne the parameter regimes m > 1 as super-critical, that is 1+hf(x) the regime where p(x) 2 C ;8k, and f (x) is a polynomial function. We de ne m = 1 as critical, recovering the classical case from the literature. Lastly we de ne 0 < m < 1 as sub-critical, that is the regime where p(x) looses smoothness, and is not even Lipschitz. Thus the goals of the current manuscript are: (1) To consider a generalized model of interference, in the sub-critical regime; therein to investigate the phenomenon of nite time extinction, that can occur in this regime. MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 3 (2) To investigate this model dynamically, including the various bifurcations that might occur; (3) To investigate the e ect of prey refuge on the dynamics of this generalized model. We nd that there is a critical amount of refuge that prevents nite time extinction of the prey. This is seen via theorem 5.1. The rest of the paper is organized as follows. The mathematical formulation of the problem and mathematical preliminaries such as nonnegativity, boundedness and dissipativeness are presented in Section 2. Existence of equilibria, stability analysis and various local bifurcation analysis are considered in Section 3. In Section 4, we analyze the possibility of nite time extinction of the prey population. We investigate the e ect of prey refuge in Section 5. Additionally, stability analysis and various local bifurcation analysis are carried out. Numerical simulations are performed in Section 6 to correlate with some of our key analytical ndings. In the last section, we present our discussions and conclusions. 2. Model Formulation First, we consider a general predator-prey model with mutual interference among predators of the form dx = x f (x ) w g(x )x ; < 1 1 0 1 dt (1) dx : 2 m = a x + w g(x )x ; 2 2 1 1 dt where f (x ) and g(x ) are the logistic growth and the functional response of the 1 1 predator towards the prey respectively. Assume that 0 < m  1, as per literature on mutual interference [1, 2]. In this paper, we consider the general logistic growth and the generalized Holling type functional response, see [3, 4]: (2) f (x) = a b x; g(x) = : 1 1 x + d Assume that 0 < m  1. The assumptions placed on the functions f and g in (2) are: (I) g is continuous for x  0 and g(0) = 0; (II) g is smooth for x > 0 and g (x ) > 0 for x > 0; 1 1 1 (III) f is smooth for x  0; a a 1 1 (IV) There exists > 0 such that x f (x ) < 0 for x  0, x 6= ; 1 1 1 1 b b 1 1 g(x ) 0 + (V) For 0 < m < 1, g (0 ) := lim + = +1; x !0 (VI) g is not smooth for x = 0 when 0 < m < 1; 1 1 dx (VII) The integral lim converges for xed > 0. !0 g(x ) Thus the predator-prey model with mutual interference and the generalized Holling type functional response becomes dx > 1 x m 2 1 2 < = a x b x w x ; 1 1 1 0 1 2 x +d dt (3) > 1 dx > 2 x m 1 2 = a x + w x ; 2 2 1 x +d dt 4 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD The variables and parameters used in the model are de ned in Table 1. Table 1. List of parameters used in the model (3). All parameters considered are positive constants. Variables/ Parameters Description x Prey population x Predator population t Time a Per capita rate of self-reproduction for the prey a Intrinsic death rate of the predator population w Maximum rate of per capita removal of prey w Measure eciency of biomass conversion from prey to predator b Death rate of prey population due to intra-species competition d Half saturation constant 1=m Predators feeding intensity m Mutual interference exponent 2.1. Mathematical Preliminaries. There are essential properties that a mathematical model must exhibit in order to obtain realistic solutions. In particular, it is important to guarantee positivity of the populations. Likewise, boundedness of the total population is another impor- tant feature of a realistic model. In this section, we present guarantee positivity, boundedness, and dissipativeness of the mathematical model (3). 2.1.1. Positivity and Boundedness. The nonnegativity of populations generated by the mathematical model (3) is clearly important to make biological sense. In addi- tion, positivity implies survival of the populations over the temporal domain. The boundedness of populations ensures that no population supercedes unrealistic val- ues in time. In particular, boundedness guarantees that the total population does not grow beyond an exponential rate for an unbounded interval. Guaranteeing both of these features makes strides to showing the feasibility of a mathematical model for describing population behavior. Lemma 2.1. Consider the following region R = f(x ; x ) : x  0; x  0g, then 1 2 1 2 all solutions (x (t); x (t)) of model (3) with initial conditions x (0) > 0; x (0) > 0 1 2 1 2 are nonnegative for all t  0. Proof. The proof of Lemma (2.1) follows from the proof of Theorem 3:1 in [3]. Lemma 2.2. All solutions (x (t); x (t)) of model (3) with initial conditions x (0) > 1 2 1 0; x (0) > 0 are bounded. 2 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 5 Proof. Let us de ne the function Q (x (t); x (t)) = x (t) + x (t): Then 1 2 1 2 dQ dx dx 1 2 = + dt dt dt m m 1 1 x x 1 m 1 m 2 2 2 = a x b x w x a x + w x : 1 1 1 0 2 2 1 1 2 2 x + d x + d 1 1 Let  be a positive constant such that   a and suppose w  w , then we obtain, 2 0 1 m m 1 1 dQ x x 1 m 1 m 2 2 2 + Q = a x b x w x a x + w x 1 1 1 0 2 2 1 2 2 dt x + d x + d 1 1 + (x + x ) 1 2 1 m 2 2 = a x b x (w w ) x (a )x + x 1 1 1 0 1 2 2 1 x + d (a + ) (a + )x b x  : 1 1 1 4b (a + ) Taking W = and applying the theory on di erential inequality, we obtain 4b W (1 e ) 0  Q (x (t); x (t))  + Q (x (0); x (0)) e ; 1 2 1 2 which implies (4) lim sup Q (x (t); x (t))  : 1 2 t!1 By (4) and Lemma (2.1), all solutions of (3) with initial conditions x (0) > 0; x (0) > 0 will be contained in the region = f(x ; x ) 2 R : Q (x (t); x (t))  + ; for any  > 0g: 1 2 1 2 The proof is complete. 2.1.2. Dissipativeness. In the previous section, it was shown that the total population remains positive and bounded for all time. Here, we showed that the individual populations are all bounded from above. In such a situation, we say that the model is dissipative. Lemma 2.3. The system (3) is dissipative. Proof. From the rst system of (3) dx x 1 1 2 2 = a x b x w x ; 1 1 1 0 1 2 dt x + d a x b x : 1 1 1 This implies that (5) lim sup x (t)  : t!1 1 6 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD The inequality (5) gives that for arbitrary small  > 0; there exist a real number T > 0 such that (6) x (t)  +  ; for all t  T : 1 1 1 Using (6), we obtain for all t  T , d w x dx w dx 0 2 1 0 2 x + = + dt w dt w dt 1 1 w a 0 2 = a x b x x 1 1 1 2 w a 0 2 a x x 1 1 2 w x 0 2 = (a + a )x a x + 1 2 1 2 1 w x 0 2 K a x + ; 1 2 1 where K = (a + a ) +  : Therefore, we obtain 1 1 2 1 w x K 0 2 2 (7) lim sup x +  : w a t!1 1 2 By (5) and (7), there exists a real number K such that lim sup x  K : 2 2 t!1 Thus, for arbitrary small  > 0, there exists T > T > 0, such that for all t  T 2 2 1 2 x  K +  : 2 2 2 Therefore the model (3) is dissipative. 3. Existence of Equilibria In this section, we determine and analyze equilibria for our mathematical model. Consider the solutions to the steady state equations: 2 m (8) a x b x w x = 0 1 1 1 0 1 2 x + d 1 m (9) a x + w x = 0 2 2 1 x + d The above equations, (8) and (9), have three types of non-negative equilibria: (i) The trivial equilibrium E (0; 0); (ii) The predator-free equilibrium E (a =b ; 0); 1 1 1 (iii) The interior equilibrium E (x ; x ) where x and x are related by 1 2 1 2 x = a x b x : 1 1 2 1 1 w a 0 2 We have that a b x  0 since x  0 and x  0. The possible existence 1 1 1 1 2 of a unique or multiple interior equilibria are shown in Fig. (1). MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 7 Figure 1. Figure (a) and (b) represent graphical illustration of the predator and prey non-trivial nullclines when m = m = 0:5. 1 2 Figure (c) represents graphical illustration of the predator and prey non-trivial nullclines when m = 1; m = 0:5. 1 2 3.1. Stability Analysis of the Interior Equilibrium. The variational matrix J of the model (3) around the interior equilibrium E (x ; x ) is 1 2 a a 11 12 J = ; a a 21 22 where m m 1 2  1 dm w x x 1 0 2 1 a = a 2b x ; 11 1 1 m +1 (x + d) m 1 m 2  1 m w x x 2 0 2 1 a = < 0; (x + d) m 1 m 1 a = m dw x > 0; 21 1 1 m +1 (x + d) m 1 2 1 a = a + m w x : 22 2 2 1 x + d The characteristic equation corresponding to J evaluated at E (x ; x ) is given by 1 2 (10)  tr (J ) + det (J ) = 0; 8 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD where tr (J ) = a + a 11 22 1 m 1 x x m 1 m 2 1  2 1 = a a 2b x + m w x dm w x 1 2 1 2 1 1 0 1 2 2 m +1 x + d (x + d) 1 1 and det (J ) = a a a a 11 22 12 21 m m  1 2 1 dm w x x x 1 0 m 1 2 1  2 1 = a 2b x a + m w x 1 1 2 2 1 1 2 m +1 x (x + d) x + d 1 1 1 m 1 m m 1 2  1  1 m w x x x 2 0 m 2 1  2 1 m dw x : 1 1 m m +1 1 1 (x + d) (x + d) 1 1 Here, tr (J ) and det (J ) represents the trace and determinant of the variational matrix. Hence the stability of E (x ; x ) is determined by the sign of det (J ) and 1 2 tr (J ). The above results are encapsulated in the following theorem. Theorem 3.1. The interior equilibrium E (x ; x ) is locally asymptotically stable 1 2 if tr (J ) < 0 and det (J ) > 0 by Routh-Hurwitz stability criteria. Proof. The proof follows directly from the above discussion and hence omitted for brevity. 3.2. Global Asymptotic Stability. Considering the case where m = 1 and assumptions (V)-(VII) hold for our response functions f and g, then E is not a saddle point and the argument using the Poincare-Bendixson theorem cannot be applied. We will describe the global behavior of the system (3) by considering the relative position of the stable and unstable separatrix of the saddle point E . We denote the stable separatrix by s u W (E ) and the unstable separatrix by W (E ). 0 1 The predator nullcline is the vertical line x = x determined by the equation a + w g(x ) = 0. We assume 2 1 1 a d a (11) w > a ; > x := : 1 2 1 1 1 b m m 1 1 1 w a 1 2 The prey nullcline is the graph of the function y = (x ) x f (x ) 1 1 (12) (x ) = w g(x ) 0 1 where f (x ) and g(x ) are de ned in (2). Clearly ( ) = 0 and (x ) > 0 for 1 1 1 0 < x < . The unique interior equilibrium E (x ; x ) is the intersection of the 1 2 1 2 predator and prey nullclines and it can be stable or unstable depending on the sign 0  0  0 of (x ). For (x ) > 0, E is a repeller (unstable) and for (x ) < 0, E is an 2 2 1 1 1 attractor (or locally asymptotically stable). Remark 1. The predator-free equilibrium E turns into a stable node with the loss of the unique interior equilibrium E . 2 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 9 Based on the non-uniqueness of the solution of model (3) when m < 1 and m = 1, we have the following results. s u Lemma 3.2 ( Proposition 3.1 in [6]). Assume that W (E ) is above W (E ). If 0 1 E is a repeller, then it is surrounded by at least one limit cycle. If the system can have at most one cycle, then E is surrounded by at least a unique limit cycle which is orbitally asymptotically stable. This limit cycle is not globally orbitally asymptotically stable, even if it is unique. If E is an attractor and if the system has no cycles, then all orbits under W (E ) converge towards E . E is not globally 0 2 2 asymptotically stable, even if it is not surrounded by any unstable limit cycle. 0  s u Proposition 3.3. Assume (x )  0 and (11) hold, then W (E ) is above W (E ) 0 1 and the basin of attraction of E is the positive region of the plane located under W (E ). Hence E is not globally asymptotically stable. 0 2 Proof. The proof is similar to the proof of Proposition 4.2 in [6] and omitted for brevity. 3.3. Saddle-Node Bifurcation Analysis. We investigate the possibility of saddle- node bifurcation of the positive interior equilibrium E by using the intrinsic death rate of the predator population as a bifurcation parameter. The following theorem states the restrictions for occurrence of a saddle-node bifurcation for model (3). Theorem 3.4. The model (3) undergoes a saddle-node bifurcation around E at a when the system parameters satisfy the restriction det (J ) = 0 along with the condition tr (J ) < 0. Proof. To validate the restriction for the occurrence of saddle-node bifurcation, we apply Sotomayor's theorem [9] at a = a . At a = a , it can be seen that 2 2 2 2 det (J ) = 0 and tr (J ) < 0 which indicates that the Jacobian (J ) admits a zero eigenvalue. Let U and V be the eigenvectors corresponding to the zero eigenvalue T T of the matrix (J ) and (J ) respectively. We obtain that U = (u ; u ) and 1 2 a u a v T 2 2 12 21 V = (v ; v ) , where u = , v = and u ; v 2 Rnf0g. 1 2 1 1 2 2 a a 11 11 T   T Furthermore, let F = (F ; F ) and X = (x ; x ) , where F ; F are given by 1 2 1 2 1 2 2 m F = a x b x w x 1 1 1 1 0 1 2 x + d F = a x + w x : 2 2 2 1 x + d Now T  T V F (X; a ) = (v ; v )(0;x ) = v x 6= 0; a 1 2 2 2 2 2 2 and T 2 V D F (X; a )(U; U ) 6= 0: Hence, from Sotomayor's theorem the model undergoes a saddle-node bifurcation around E at a = a . 2 2 Theorem 3.5. The model (3) undergoes a saddle-node bifurcation around E at w when the system parameters satisfy the restriction det (J ) = 0 along with the condition tr (J ) < 0. 10 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD Theorem 3.6. The model (3) undergoes a saddle-node bifurcation around E at w when the system parameters satisfy the restriction det (J ) = 0 along with the condition tr (J ) < 0. Theorem 3.7. The model (3) undergoes a saddle-node bifurcation around E at b when the system parameters satisfy the restriction det (J ) = 0 along with the condition tr (J ) < 0. Theorem 3.8. The model (3) undergoes a saddle-node bifurcation around E at a when the system parameters satisfy the restriction det (J ) = 0 along with the condition tr (J ) < 0. Proof. The proof of Theorem 3.5, Theorem 3.6, Theorem 3.7 and Theorem 3.8 are similar to proof in Theorem 3.4 and omitted for brevity. 3.4. Hopf-Bifurcation Analysis. We investigate the possibility of Hopf-bifurcation of the positive interior equilibrium E by using the per capita rate of self-reproduction for the prey, a as a bifurcation parameter. Then, the characteristic equation cor- responding to model (3) at E is given by (13)  + A(a ) + B(a ) = 0; 1 1 where A = tr (J ) = (a + a ) and B = det (J ) = a a a a : 11 22 11 22 12 21 The instability of model (3) is demonstrated via the following theorem by con- sidering a as a bifurcation parameter. Theorem 3.9 (Hopf-Bifurcation Theorem [8]). If A(a ) and B(a ) are the smooth 1 1 functions of a in an open interval about a 2 R such that the characteristic equa- tion (13) has a pair of imaginary eigenvalues  =  (a )  i (a ) with  and 1 1 2 R so that they become purely imaginary at a = a and j 6= 0, then a 1 a =a 1 1 da 1 Hopf-bifurcation occurs around E (x ; x ) at a = a (i.e. a stability changes of 2 1 1 2 1 E (x ; x ) accompanied by the creation of a limit cycle at a = a ). 2 1 1 2 1 Theorem 3.10. The model (3) undergoes a Hopf-bifurcation around E (x ; x ) 1 2 when a crosses some critical value of parameter a , where 1 m 1 x x m 1 m 2 1  2 1 a = a + 2b x m w x + dm w x 2 1 2 1 1 0 1 1 2 2 m +1 x + d (x + d) 1 1 provided: (i) A(a ) = 0, (ii) B(a ) > 0, (iii) Re (a )j 6= 0 at a = a ; i = 1; 2. i 1 1 a =a 1 da Proof. Clearly A(a ) and B(a ) are the smooth functions of a . The roots of the 1 1 1 equation (13) are of the form  =  (a ) + i (a ) and  =  (a ) i (a ) where 1 1 1 2 1 1 (a ) and (a ) are real functions. 1 1 At a = a , the characteristic equation (13) reduces to (14)  + B(a ) = 0 p p By solving for the roots of equation (14), we obtain  = i B and  = i B. 1 2 Hence a pair of purely imaginary eigenvalues. Furthermore, we validate the transver- sality condition: Re (a )j 6= 0; i = 1; 2: i 1 a =a da 1 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 11 Substituting (a ) =  (a ) + i (a ) into equation (13), we obtain 1 1 1 (15) ( (a ) + i (a )) + A(a )( (a ) + i (a )) + B(a ) = 0: 1 1 1 1 1 1 Now, taking the derivative with respect to a , we get _ _ 2( (a ) + i (a ))( (a ) + i _ (a )) + A(a )( (a ) + i _ (a )) 1 1 1 1 1 1 1 _ _ +A(a )( (a ) + i (a )) + B(a ) = 0: 1 1 1 1 Separating the real and imaginary parts, we have _ _ _ _ (a )(2 (a ) + A(a )) + (a )(2 (a )) + A(a ) (a ) + B(a ) = 0; 1 1 1 1 1 1 1 1 which implies (16)  (a )Z (a ) _ (a )Z (a ) + Z (a ) = 0; 1 1 1 1 2 1 3 1 and _ _ (a )(2 (a )) + _ (a )(2 (a ) + A(a )) + A(a ) (a ) = 0; 1 1 1 1 1 1 1 which implies (17)  (a )Z (a ) + _ (a )Z (a ) + Z (a ) = 0; 1 2 1 1 1 1 4 1 _ _ where Z (a ) = 2 (a ) +A(a ), Z (a ) = 2 (a ), Z (a ) = A(a ) (a ) +B(a ) and 1 1 1 1 2 1 1 3 1 1 1 1 Z (a ) = A(a ) (a ). 4 1 1 1 Multiplying equation (41) by Z (a ) and equation (42) by Z (a ) and then 1 1 2 1 adding them, we obtain 2 2 (18) (Z (a ) + Z (a )) (a ) + Z (a )Z (a ) + Z (a )Z (a ) = 0; 1 1 1 1 1 3 1 2 1 4 1 1 2 thus solving for  (a ) from equation (43) and at a = a , 1 1 d [Z (a )Z (a ) + Z (a )Z (a )] 1 3 2 4 1 1 1 1 Re (a )j =  (a ) = : i 1 a =a 1 1 2  2 da Z (a ) + Z (a ) 1 1 2 1 2  2 It is easy to verify that Z (a )Z (a ) + Z (a )Z (a ) 6= 0 and Z (a ) + Z (a ) 6= 0 1 3 2 4 1 1 1 1 1 1 2 1 which implies Re (a )j 6= 0. Hence, a Hopf-bifurcation occurs around i 1 a =a da 1 E (x ; x ) at a = a . 2 1 1 2 1 4. Finite Time Extinction An interesting property of (3) is that the population of prey may go extinct in nite time for carefully chosen initial conditions. Therefore, while solutions do remain nonnegative, it is clear that the populations may become extinct and no longer persist. Theorem 4.1. Consider the predator-prey system given by (3). The solution x (t) to the prey equation x (t) with initial conditions x (0) > 0; x (0) > 0 can go extinct 1 1 2 in nite time, for suciently chosen initial conditions. Proof. Consider the substitution x = 1=u in the prey equation of (3). This yields the following system: 1 1 dx 1 du 1 1 2 2 > u < = = a b ( ) w x ; 1 1 0 1 2 +d dt u dt u u (19) 1 m > 1 dx : m u 2 = a x + w x : 2 2 1 1 +d dt u 12 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD This system can be simpli ed into the system in u; x : du u > m = a u + b + w x ; < 1 1 0 dt (1 + du) (20) >dx 1 > m = a x + w x ; 2 2 1 dt (1 + du) Note, via positivity dx (21)  a x : 2 2 dt Thus, a t (22) x  x (0)e : 2 2 Also, via positivity we have the inequality, 2 2 du u u m a t m 2 2 2 (23)  a u + w x  a u + w (x (0)e ) : 1 0 1 0 2 m m 1 1 dt (1 + du) (1 + du) Note that the solution to the di erential equation du ~ u ~ (24)  a u ~ + w 1 0 dt (1 + du ~) will blow up in a nite time, T (u ) < 1, as long as the initial data u ~(0) = u 0 0 satis es, m 2 (25) a u (1 + du )  w u : 1 0 0 0 Now if we choose x (0)  1, such that a t m 2 2 (26) (x (0)e ) > 1; t 2 [0; T ]; then u  u ~ on [0; T ], and must blow-up in nite time, at some T < T , by comparison, if u is chosen to satisfy (25) . Therefore, lim u ! 1 t!T <1 which implies 1 1 lim x = lim = ! 0; t!T <1 t!T <1 u lim u t!T <1 but that implies x (t) goes extinct in nite time for x (0) chosen large enough and 1 2 1m m 1 1 (x (0)) (x (0) + d)  : 1 1 5. The effect of prey refuge In the previous section it was shown that the prey population may go extinct in nite time. Therefore, we seek to investigate the e ect of protecting the prey from predation with their habitat. The aim is to provided avenues for which the prey population will persist. Here, using a similar ideas from [20], we introduce a prey refuge. A discussion of how a habitat controller may create a prey refuge is provided in section 7 of [20]. Essentially, one must protect a constant proportion of prey by replacing the predation term g(x ) by g(rx ), where 0  r  1. Here, r is a refuge parameter, 1 1 such that if r = 0 then complete protection of the prey is provided while r = 1 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 13 implies no protection and the original system (1) is recovered. Thus, we write the following system that models prey refuge as 8 m dx rx 1 1 > 2 2 > = a x b x w x ; 1 1 1 0 < 1 2 dt rx + d (27) dx rx 2 1 > m = a x + w x ; 2 2 1 dt rx + d We now state our rst result concerning prey refuge, Theorem 5.1. Consider the predator-prey system given by (27) for r = 1, that is no refuge. For any initial conditions > x (0) > 0; x (0) > 0 s.t. the 1 2 solution x (t) to the prey equation goes extinct in nite time, there exists a refuge r (x (0); a ; b ; d; w ; m ; m ; K ) > 0 s.t. for any r < r the solution x (t) to the 1 1 1 0 1 2 2 1 prey equation does not go extinct in nite time. Proof. The proof follows in similar fashion to the proof of Theorem (4.1). Hence, we make the substitution x = 1=u in the prey equation in (27). This results in: du u > m m 1 2 = a u + b + w r x ; 1 1 0 dt (r + du) (28) dx r > m = a x + w x ; 2 2 1 dt (r + du) Via comparison we have du dt m m 1 2 = a u + b + w r x 1 1 0 (r + du) w r 0 m 2m 2 a u + b + u x 1 1 w r 2m m 1 2 (29)  a u + b + u K 1 1 Recall that x and x are bounded above. Let the upper bounds for x and x be 1 2 1 2 du constants K and K , respectively. Now, we desire that  0, 8t, as this will 1 2 dt ensure that u cannot blow-up in nite time, or x cannot go extinct in nite time. lets compare to du w r 2m m 1 2 (30) = a u + b + u K 1 1 dt d let us set u = + v, this changes (30) to dv w r b 0 1 2m m 1 2 (31) = a v + ( + v) K dt d a du du Now, for =  0 one requires that dt dt w r b 0 1 2m m 1 2 (32) a v + ( + v) K  0 d a 1 14 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD This is possible if we choose r such that, ! 1 (a d )v(0) (33) r < : b m 1 2m 2 w ( + v(0)) K This is seen simply by looking in the phase for the v equation, and shows, v cant blow up in nite time. Thus neither can u = + v, thus making the nite time extinction of x an impossibility. 5.1. Existence of Equilibria and Stability Analysis. Similar to Section 3, we investigate and analyze the equilibrium solutions of our mathematical model with prey refuge. Consider the steady state equations of (27): rx 2 m (34) a x b x w x = 0 1 1 1 0 rx + d rx (35) a x + w x = 0 2 2 1 rx + d The above equations (34) and (35) have three types of non-negative equilibria: (i) The trivial equilibrium E (0; 0). (ii) The predator-free equilibrium E (a =b ; 0). 1 1 1 (iii) The interior equilibrium E (x ; x ) where x and x are related by 1 2 1 2 x = a x b x : 1 1 2 1 1 w a 0 2 The variational matrix J of the model (27) around the interior equilibrium E (x ; x ) is 1 2 b b 11 12 J = : b b 21 22 where m 1 r r x rx m 1 1 b = a 2b x m w x ; 11 1 1 1 0 1 2 d + rx (rx + d) rx + d 1 1 1 m 1 m 2  1 m w x (rx ) 2 0 2 1 b = ; (rx + d) m 1 m 1 m  2 b = m dw r x ; 21 1 1 m +1 (rx + d) rx m 1 2 1 b = a + m w x : 22 2 2 1 rx + d The characteristic equation corresponding to J is given by tr (J ) + det (J ) = 0; r r where tr (J ) = b + b 11 22 m 1 r r x rx m 1 1 = a a 2b x m w x 1 2 1 1 0 1 2 d + rx (rx + d) rx + d 1 1 1 rx m 1 2 1 + m w x ; 2 1 rx + d 1 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 15 and det (J ) = b b b b 11 22 12 21 m 1 r r x rx 2 1 1 = a 2b x m w x 1 1 1 0 1 2 d + rx (rx + d) rx + d 1 1 1 rx m 1 2 1 a + m w x 2 2 1 rx + d m 1 m m 1 2  1  1 m w x (rx ) x 2 0 2 1 m m 1 1 2 m dw r x : 1 1 m  m +1 1 1 (rx + d) (rx + d) 1 1 Here, tr (J ) and det (J ) represents the trace and determinant of the variational r r matrix. Hence the stability of E (x ; x ) is determined by the sign of det (J ) and 1 2 r tr (J ). The above results are summarized in the following theorem. Theorem 5.2. The interior equilibrium E (x ; x ) of system (27) is locally asymp- 1 2 totically stable if tr (J ) < 0 and det (J ) > 0 by Routh-Hurwitz stability criteria. r r Proof. The proof follows directly from the above discussion and hence omitted for brevity. 5.2. Bifurcation Analysis. In this subsection, we analyze the qualitative changes in the dynamical behavior of model (27) under the e ect of varying a speci c parameter. The conditions and restrictions for the occurrence of saddle-node, Hopf, and transcritical bifurcations are derived. The classi cation are of codimension one bifurcations. 5.2.1. Saddle-Node Bifurcation. We investigate the possibility of saddle-node bifur- cation of the positive interior equilibrium E by using the intrinsic death rate of the predator population as a bifurcation parameter. The following theorem states the restrictions for occurrence of a saddle-node bifurcation for model (27). Theorem 5.3. The model (27) undergoes a saddle-node bifurcation around E at a when the system parameters satisfy the restriction det (J ) = 0 along with the 2 r condition tr (J ) < 0. Proof. To validate the restriction for the occurrence of saddle-node bifurcation, we apply Sotomayor's theorem [9] at a = a . At a = a , it can be seen that 2 2 2 2 det (J ) = 0 and tr (J ) < 0 which indicates that the Jacobian (J ) admits a zero r r r eigenvalue. Let U and V be the eigenvectors corresponding to the zero eigenvalue T T of the matrix (J ) and (J ) respectively. We obtain that U = (u ; u ) and 1 2 r r b u b v T 2 2 12 21 V = (v ; v ) , where u = , v = and u ; v 2 Rnf0g. 1 2 1 1 2 2 b b 11 11 T   T Let G = (G ; G ) and X = (x ; x ) , where G ; G are given by 1 2 1 2 1 2 rx 2 m (36) G = a x b x w x ; 1 1 1 1 0 1 2 rx + d rx 1 m (37) G = a x + w x : 2 2 2 1 rx + d 1 16 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD Now T  T V G (X; a ) = (v ; v )(0;x ) = v x 6= 0; a 1 2 2 2 2 2 2 and T 2 V D G(X; a )(U; U ) 6= 0: Hence from Sotomayor's theorem the model (27) undergoes a saddle-node bifurca- tion around E at a = a . 2 2 Theorem 5.4. The model (27) undergoes a saddle-node bifurcation around E at w when the system parameters satisfy the restriction det (J ) = 0 along with the 1 r condition tr (J ) < 0. Theorem 5.5. The model (27) undergoes a saddle-node bifurcation around E at w when the system parameters satisfy the restriction det (J ) = 0 along with the 0 r condition tr (J ) < 0. Theorem 5.6. The model (27) undergoes a saddle-node bifurcation around E at b when the system parameters satisfy the restriction det (J ) = 0 along with the 1 r condition tr (J ) < 0. Theorem 5.7. The model (27) undergoes a saddle-node bifurcation around E at a when the system parameters satisfy the restriction det (J ) = 0 along with the 1 r condition tr (J ) < 0. Proof. The proof of Theorem 5.4, Theorem 5.5, Theorem 5.6 and Theorem 5.7 are similar to proof in Theorem 5.3 and omitted for brevity. 5.2.2. Transcritical Bifurcation. Here, we investigate the possibility of the existence of a transcritical bifurcation for the model (27). Transcritical bifurcation occurs when an equilibrium point interchanges its stability when it collides with another equilibrium point as a parameter is varied. The prey refuge parameter r is used as a bifurcation parameter. The variational matrix J of the model (27) for 0 < m < 1 and m = 1 r 1 2 evaluated at 0 1 b d a @ A r = r = 1 1 a m m 1 1 w a 1 2 around the predator-free equilibrium E (a =b ; 0) is given by 1 1 1 2   3 r a a w 1 0 4  5 J = : r r a + b d 1 1 1 1 0 0 At r = r , the matrix J has a negative eigenvalue and a zero eigenvalue. Let U and V be the eigenvectors corresponding to the zero eigenvalue of the matrix (J  ) and (J ) respectively. Then r r a b d 1 1 U = 1; 1 + ; V = (0; 1) : w r a 0 1 1 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 17 T T Let G = (G ; G ) and X = (a =b ; 0) , where G ; G are de ned in (36) and (37). 1 2 1 1 1 2 Now we have T  T V G (X; r ) = (0; 1)(0; 0) = 0; additionally V [DG (X; r )U ] 6= 0 and T 2 V D G(X; r )(U; U ) 6= 0: Hence using Sotomayor's theorem the model (27) undergoes a transcritical bifur- cation around E when the refuge r crosses the critical value of the parameter r . The above results are summarized in the following theorem. Theorem 5.8. The model (27) undergoes a transcritical bifurcation around E (a =b ; 0) 1 1 1 when the refuge r crosses the critical value of parameter r , where 0 1 b d a @ A r = . 1 1 1 a m m 1 1 w a 1 2 5.2.3. Hopf-Bifurcation. We investigate the possibility of Hopf-bifurcation of the positive interior equilibrium E by using the per capita rate of self-reproduction for the prey, a as a bifurcation parameter. Then, the characteristic equation corresponding to model (27) at E is given by (38)  + A (a ) + B (a ) = 0; 1 1 1 1 where A = tr (J ) = (b + b ) and B = det (J ) = b b b b . 1 11 22 1 11 22 12 21 r r The instability of model (27) is demonstrated via the following theorem by con- sidering a as a bifurcation parameter. Theorem 5.9. The model (27) undergoes a Hopf-bifurcation around E (x ; x ) 1 2 when a crosses some critical value of parameter a , where m 1 r r x rx 2 1 1 a = a + 2b x + m w x 2 1 1 0 1 1 2 d + rx (rx + d) rx + d 1 1 1 rx m 1 1 m w x ; 2 1 rx + d provided (i) A (a ) = 0, 1 1 (ii) B (a ) > 0, 1 1 (iii) Re (a )j 6= 0 at a = a ; i = 1; 2. i 1 1 a =a 1 da Proof. Clearly A (a ) and B (a ) are the smooth functions of a . The roots of the 1 1 1 1 1 equation (38) are of the form  = #(a ) + i$(a ) and  = #(a ) i$(a ) where 1 1 1 2 1 1 #(a ) and $(a ) are real functions. 1 1 At a = a , the characteristic equation (38) reduces to (39)  + B (a ) = 0 1 1 18 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD p p By solving for the roots of equation (39), we obtain  = i B and  = i B . 1 1 2 1 Therefore, we have purely imaginary eigenvalues. Hence, we are left with validating the transversality condition. Namely, Re (a )j 6= 0; i = 1; 2: i 1 a =a da Substituting (a ) = #(a ) + i$(a ) into equation (38), we obtain 1 1 1 (40) (#(a ) + i$(a )) + A (a )(#(a ) + i$(a )) + B (a ) = 0: 1 1 1 1 1 1 1 1 Upon taking the derivative with respect to a we obtain: _ _ 2(#(a ) + i$(a ))(#(a ) + i$ _ (a )) + A (a )(#(a ) + i$ _ (a )) 1 1 1 1 1 1 1 1 _ _ +A (a )(#(a ) + i$(a )) + B (a ) = 0: 1 1 1 1 1 1 Separating the real and imaginary parts, we have _ _ _ #(a )(2#(a ) + A (a )) + $(a )(2$(a )) + A (a )#(a ) + B (a ) = 0; 1 1 1 1 1 1 1 1 1 1 1 which implies (41) #(a )Z (a ) $ _ (a )Z (a ) + Z (a ) = 0; 1 1 1 1 2 1 3 1 and _ _ #(a )(2$(a )) + $ _ (a )(2#(a ) + A (a )) + A (a )$(a ) = 0; 1 1 1 1 1 1 1 1 1 which implies (42) #(a )Z (a ) + $ _ (a )Z (a ) + Z (a ) = 0; 1 2 1 1 1 1 4 1 _ _ where Z (a ) = 2#(a ) +A (a ), Z (a ) = 2$(a ), Z (a ) = A (a )#(a ) +B (a ) 1 1 1 1 1 2 1 1 3 1 1 1 1 1 1 and Z (a ) = A (a )$(a ). Multiplying equation (41) by Z (a ) and equation (42) 4 1 1 1 1 1 1 by Z (a ) and then adding them, we obtain 2 1 2 2 (43) (Z (a ) + Z (a ))#(a ) + Z (a )Z (a ) + Z (a )Z (a ) = 0; 1 1 1 1 1 3 1 2 1 4 1 1 2 thus solving for #(a ) from equation (43) and at a = a , 1 1 d [Z (a )Z (a ) + Z (a )Z (a )] 1 3 2 4 1 1 1 1 Re (a )j = #(a ) = : i 1 a =a 1 1 1 2  2 da Z (a ) + Z (a ) 1 1 2 1 2  2 It is easy to verify that Z (a )Z (a )+Z (a )Z (a ) 6= 0 and Z (a )+Z (a ) 6= 0 1 3 2 4 1 1 1 1 1 1 2 1 which implies Re (a )j 6= 0. Hence, a Hopf-bifurcation occurs around i 1 a =a da 1 E (x ; x ) at a = a . 2 1 1 2 1 Theorem 5.10. The model (27) undergoes a Hopf-bifurcation around E (x ; x ) 1 2 when the refuge r crosses some critical value of parameter r provided (i) A (r) = 0, (ii) B (r) > 0, (iii) Re (r)j 6= 0 at r = r ; i = 1; 2. r=r 1 dr Proof. The proof of Theorem 5.10 is similar to proof in Theorem 5.9 and omitted for brevity.  MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 19 6. Numerical Simulations Numerical simulations of model (3) are performed in this section to correlate with some of our key analytical ndings. The numerical simulations and gures have been developed using MATLABr R2019b, MATCONT[7], and XPPAUT[5]. For convenience, the parameters used in simulations are given in Table 2. Table 2. Parameters used in the simulations of Figs. 2, 3, 4. 7, 8, and 9. a = 0:6 a = 1 b = 0:063 w = 1 d = 2 1 2 1 0 w = 2 m = 0:8 m = 1 1 1 2 For the parameter values in Table 2, the predator-free equilibrium point E (9:52381; 0) is a saddle and E (1:45094; 1:47587) is a repeller, see Fig. 2(a). In Fig. 2(a), s u W (E ) is above W (E ) and E (1:45094; 1:47587) is surrounded by a limit cycle. 0 1 2 It is seen that in Fig. 2(b), when a = 2, b = 0:21 and all the other parameter 1 1 s u sets are given in Table 2, then W (E ) is under W (E ), where E (9:52381; 0) 0 1 1 and E (1:45094; 4:91957). So the predator-free equilibrium point E (9:52381; 0) is 2 1 a saddle and E (1:45094; 4:91957) is a repeller. Here all positive solutions converge towards E . In Fig. 3, when a = 1:5, b = 0:18, a = 1:4 and all the other 0 1 1 2 parameter sets are given in Table 2, we observe that E (3:55994; 4:36963) is an s u attractor and E (8:33333; 0) is a saddle. Also W (E ) is above W (E ) and the 1 0 1 model (3) has a unique limit cylce whose basin of attraction is the region under W (E ). However, in n Fig. 3, the numerical simulations illustrate that E is not 0 2 globally asymptotically stable for the given parameter set. (a) (b) Figure 2. The predator and prey nullclines for model (3) are represented by turquoise and red respectively. (a) W (E ) is above u u s W (E ): E is unstable (b) W (E ) is above W (E ), here a = 2 1 2 1 0 1 and b = 0:21: E is unstable. Other parameter sets are given in 1 2 Table 2. 20 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD s u Figure 3. E is an attractor and W (E ) is above W (E ). Here 2 0 1 a = 1:5, b = 0:18 and a = 1:4. Other parameter sets are given 1 1 2 in Table 2. Figure 4. Bifurcation diagrams of the model (3), as a changes. The stable and unstable interior equilibriums are given by the lines in red and black, respectively. The solid circles (green) represent stable limit cycles and the open circles (blue) represent unstable limit cycles. (a) prey (x ) (b) predator (x ). Parameter set are 1 2 given in Table (2). Furthermore, for the parameter sets in Table 2, we employ AUTO as imple- mented in the continuation software XPPAUT to analyze the bifurcation dia- grams of the model (3) in Fig. 4. The model undergoes Hopf-bifurcation around E (1:45094; 0:49456) as the parameter a crosses its critical value a = 0:261835. 2 1 The branch of periodic orbits emitting from a are stable and the rst Lyapunov coeicient [9],  = 1:49929e < 0 (obtained with the aid of MATCONT), hence the Hopf-bifurcation is supercritical. MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 21 Table 3. Parameters used in the simulations of Figs. 5, 6, 10, and 11. a = 0:5 a = 0:7 b = 0:05 w = 0:2 d = 0:2 1 2 1 0 w = 4 m = 0:5 m = 0:5 1 1 2 Figure 5. Time series depicting the stability behavior of the in- terior equilibrium point (6:67563; 31:7032) for the parameter sets given in Table 3. Also, for the parameter sets in Table 3 of model (3), we obtain the following 1 2 two interior equilibrium points E (3:12437; 30:6886) and E (6:67563; 31:7032) and 2 2 the predator-free equilibrium point is E (10; 0). The eigenvalues associated with 1 1 E (3:12437; 30:6886) are 0:343043 and 0:170265, hence E is a saddle. The eigen- 2 2 2 2 values associated with E (6:67563; 31:7032) are 0:34517 and 0:17481, hence E 2 2 is locally asymptotically stable, see Fig. 5. We note here that, E (10; 0) cannot be analyzed using the linear stability method since m = m = 0:5 < 1. We observed 1 2 that the model (3) undergoes a saddle-node bifurcation around E (x ; x ) when 1 2 the following bifurcation parameters a ; a ; w ; w and b crosses their correspond- 1 2 0 1 1 ing critical values a = 0:46809; a = 0:61515; w = 0:22759; w = 4:55175; and 1 2 0 1 b = 0:05722 respectively. The saddle-node bifurcation diagrams are depicted in Fig. 6 Now, we perform numerical simulations of model (27) to verify some of our ana- lytical results. For r = 0:3 and all other parameter values given in Table 2, the predator-free equilibrium point E (9:52381; 0) is a saddle and E (4:83648; 9:52147) 1 2 is an attractor (stable), see Fig. 7. By introducing a prey refuge of r = 0:3, we 22 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD (a) (b) (c) (d) (e) Figure 6. Bifurcation diagrams illustrating (a) SN at a = a = 0:46809, (b) SN at a = a = 0:61515, (c) SN at w = w = 2 0 2 0 0:22759, (d) SN at w = w = 4:55175, (e) SN at b = b = 1 1 1 1 0:05722. Other parameter sets are given in Table 3. (SN: Saddle- node bifurcation.) MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 23 s u observed in Fig. 7(c) that W (E ) is above W (E ) as compared to Fig. 2(b) 0 1 s u where W (E ) is below W (E ). Thus the stability of the interior equilibrium is 0 1 altered and not all positive solutions tend toward E . (a) (b) (c) Figure 7. Stable dynamics (a) time series (b) phase diagram of the interior equilibrium point (4:83648; 9:52147) (c) E is an at- s u tractor and W (E ) is above W (E ). Here r = 0:4 and all other 0 1 parameter sets are given in Table 2 24 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD (a) (b) Figure 8. Bifurcation diagrams of the model (27), as a crosses its critical value a . The stable and unstable interior equilibriums are given by the lines in red and black, respectively. The solid circles (green) represent stable limit cycles and the open circles (blue) represent unstable limit cycles. (a) prey (x ) (b) predator (x ). Parameter set are given in Table (2). (a) (b) Figure 9. Bifurcation diagrams of the model (27), as r crosses its critical value r . The stable and unstable interior equilibriums are given by the lines in red and black, respectively. The solid circles (green) represent stable limit cycles. (a) prey (x ) (b) predator (x ). Parameter set are given in Table (2). Additionally, for r = 0:3 and all the other parameter sets provided in Table 2, we use AUTO as implemented in the continuation software XPPAUT to analyze the bifurcation diagrams of the model (27) in Fig. 8. The model undergoes Hopf- bifurcation around E (4:83648; 5:49507) as the parameter a crosses its critical value 2 1 a = 0:87278. The branch of periodic orbits emitting from a are stable and the 1 1 rst Lyapunov coeicient [9],  = 2:88256e < 0, hence the Hopf-bifurcation is supercritical. MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 25 Furthermore, the model undergoes Hopf-bifurcation around E (3:32486; 2:59694) as the parameter r crosses its critical value r = 0:43639, see Fig. 9. The branch of periodic orbits bifurcation from r are stable and the rst Lyapunov coecient is = 4:99384e , hence supercritical. Also, the model (27) undergoes transcritical bifurcation around E (9:52381; 0) when the parameter r crosses its threshold r = 0:15239, see Fig. 9(a) . Figure 10. Time series depicting the stability behavior of the interior equilibrium point (7:03041; 29:8249) for r = 0:3 and all other parameter sets are given in Table 3. Next, for r = 0:3 and the other parameter sets given in Table 3 of the model (27), we obtain the following two interior equilibrium points E (2:30292; 25:3225) and E (7:03041; 29:8249) and the predator-free equilibrium point is E (10; 0). The eigenvalues associated with E (2:30292; 25:3225) are 0:32246 and 0:19897, hence 1 2 E is a saddle. The eigenvalues associated with E (7:03041; 29:8249) are 0:33157 2 2 and 0:22790, hence E is locally asymptotically stable, see Fig. 10. We note here that, E (10; 0) cannot be analyzed using the linear stability method since m = m = 0:5 < 1. We observed that the model (27) undergoes a saddle-node bi- 1 2 furcation around E (x ; x ) when the following bifurcation parameters a ; a ; w ; w 2 1 2 0 1 1 2 and b crosses their corresponding critical values a = 0:44476; a = 0:5625; w = 1 2 0 0:24889; w = 4:97778; and b = 0:064498 respectively. The saddle-node bifurcation 1 1 diagrams are presented in Fig. 11. 7. Discussions and Conclusions In this work, we consider a predator-prey model, that allows us to model both the feeding intensity of the predator, as well as the e ect of mutual/predator in- terference. Through numerical simulations, it has been noticed that based on the non-uniqueness of the solutions of model (3), when m < 1 and m = 1, the interior 1 2 equilibrium E is not globally asymptotically stable when it is an attractor (see Fig. 3). We observe that the per capita rate of self-reproduction a plays an important role because the interior equilibrium point E changes stability at the bifurcation 2 26 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD point a (see Fig. 4). The limit cycle through the bifurcation point is stable hence a supercritical Hopf-bifurcation. Furthermore, the e ect of prey refuge is also considered in model (27) - thus one can see the interplay of all of these factors in this model. The model possesses a rich array of dynamical behavior. We have established analytically the occurrence of various local bifurcations including saddle-node, transcritical and Hopf bifurcations. The occurrence of these local bifurcations are well supplemented with one parameter bifurcation diagrams (see Figs. 8, 9, and 11). Prey extinction in nite time is also possible - for large enough initial predator density, and small enough initial prey s u density. Moreover, we observed that when W (E ) is above W (E ), all solutions 0 1 with initial conditions above W (E ) goes to prey extinction in nite time (see Fig. 2(a)). This is in line with the result in [6]. Thus, from a practical point of view increasing m or decreasing the feeding intensity of the predator, will maintain ecosystem balance, as this decreases the predator nullcline, decreasing predator numbers and increasing prey numbers. Stability in the system can also be maintained via provision of the prey with refuge. This is rigorosly established via theorem 5.1. The requisite condition for a critical refuge, for persistence, derived via the theorem sheds light on various ecological scenarios. The ecological validity of the prey extinction state (0; x ) is questionable. In the experiments of Gause [23], once the prey has gone extinct the predator population also crashes, as there is no alternative/additional food in the experimental system. In a real scenario however, such a state might be indicative of a predator having switched to another food source after its primary source has depleted or surviving on additional food, such as in a bio-control situation [29]. Conflict of Interest The authors declare there is no con ict of interest in this paper. References [1] Hassell, M. (1971) Mutual interference between searching insect parasites. J. Anim. Ecol. 40, 473-486. [2] Hassell, M. (1975) Density dependence in single species population. . J. Anim. Ecol. 44, 283-295. [3] Upadhyay, R.K., Agrawal, R. (2015) Modeling the e ect of mutual interference in a delay- induced predator-prey system, J. Appl. Math. Comput. 49, 13-39. [4] Upadhyay, R.K., Parshad, R., Antwi-Fordjour, K., Quansah, E., Kumari, S. (2019) Global dynamics of stochastic predator-prey with mutual interference and prey defense. J. Appl. Math. Comput. 60, 169-190. [5] Ermentrout, B. (2002) Simulating, analyzing, and animating dynamical systems: a guide to XPPAUT for researchers and students. vol 14, SIAM. [6] Beroual, N., Sari, T. (2019) A predator-prey system with Holling-type functional response. hal-02002894. [7] Dhooge, A., Govaerts, W., Kuznetsov, Yu. A., Meijer, H.G.E., Sautois, B. (2009) New features of the software MatCont for bifurcation analysis of dynamical systems, MCMDS, Vol. 14, No. 2, pp 147-175. [8] Murray, J.D. (1993) Mathematical biology, Springer, New York. [9] Perko, L. (2013) Di erential equations and dynamical systems. Vol. 7, Springer Science & Business Media. [10] Kivan, V. (1996). Optimal foraging and predatorprey dynamics. theoretical population bi- ology, 49(3), 265-290. [11] V. Krivan, Evolutionary stability of optimal foraging: partial preferences in the diet and patch models, J Theor Biol, 267:486-494, 2010. MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 27 [12] McKenzie, H. W., Merrill, E. H., Spiteri, R. J., & Lewis, M. A. How linear features alter predator movement and the functional response. Interface focus, 2(2), 205-216, 2012. [13] Mols, C. M., van Oers, K., Witjes, L. M., Lessells, C. M., Drent, P. J., & Visser, M. E. . Central assumptions of predatorprey models fail in a seminatural experimental system. Proceedings of the Royal Society of London B: Biological Sciences, 271(Suppl 3), S85-S87, [14] Ruxton, G. D. Increasing search rate over time may cause a slower than expected increase in prey encounter rate with increasing prey density. Biology letters, 1(2), 133-135, 2005. [15] Christos C. Ioannou, Graeme D. Ruxton, Jens Krause, Search rate, attack probability, and the relationship between prey density and prey encounter rate, Behavioral Ecology, Vol- ume 19, Issue 4, July-August 2008, Pages 842846, https://doi.org/10.1093/beheco/arn038 Bulletin of Mathematical Biology, 73(10):2249-2276, 2011. [16] Y. Kuang, Some mechanistically derived population models, Math. Biosci. Eng, 4(4), 1-11, [17] Kar, T. K. Stability analysis of a preypredator model incorporating a prey refuge, Commu- nications in Nonlinear Science and Numerical Simulation, 10(6), 681-691, 2005. [18] DeLong, J. P., & Vasseur, D. A. Mutual interference is common and mostly intermediate in magnitude, BMC ecology, 11(1), 2011. [19] Kivan, V. On the Gause predatorprey model with a refuge: a fresh look at the history. Journal of theoretical biology, 274(1), 67-73, 2011. [20] Parshad, R. D., Quansah, E., Black, K., & Beauregard, M. Biological control via ecological damping: an approach that attenuates non-target e ects, Mathematical biosciences, 273, 23-44. [21] J. R. Beddington, Mutual interference between parasites or predators and its e ect on searching eciency J. Animal Ecol., 44, 331-340, 1975. [22] DeAngelis D. L., R. A. Goldstein and R. V. ONeill A model for trophic interaction, Ecology, 56, 881-892, 1975. [23] Gause, G. F. The struggle for existence, Williams & Wilkins, Baltimore, Maryland, USA, [24] Holling, C.S. The components of predation as revealed by a study of small mammal predation of the European pine saw y, Canad. Entomol. 91, 293-320, 1959. [25] Erbe, L. H., & Freedman, H. I. Modeling persistence and mutual interference among sub- populations of ecological communities . Bulletin of mathematical biology, 47(2), 295-304, [26] Lotka A. J. Elements of physical biology. Williams and Wilkins, Baltimore. Reprinted as Elements of mathematical biology, Dover, New York, 1925. [27] Braza, P. A. Predatorprey dynamics with square root functional responses N onlinear Anal- ysis: Real World Applications, 13(4), 1837-1843, 2012. [28] Wang, K. and Zhu, Y. Periodic solutions, permanence and global attractivity of a delayed impulsive preypredator system with mutual interference, N onlinear Analysis: Real World Applications, 14(2), 1044-1054, 2013. [29] Parshad, R. D., Wickramsooriya, S., & Bailey, S. A remark on Biological control through provision of additional food to predators: A theoretical study[Theor. Popul. Biol. 72 (2007) 111120]. T heoretical Population Biology, 2019. [30] Freedman, H. I. Stability analysis of a predator-prey system with mutual interference and density-dependent death rates, B ulletin of Mathematical Biology, 41(1), 67-78, 1979. [31] Upadhyay, R. K., and Rao, V. S. H. Short-term recurrent chaos and role of Toxin Pro- ducing Phytoplankton (TPP) on chaotic dynamics in aquatic systems, C haos, Solitons and Fractals, 39(4), 1550-1564, 2009. [32] Sugie, J., Kohno, R., and Miyazaki, R. On a predator-prey system of Holling type, P roceedings of the American Mathematical Society, 125(7), 2041-2050, 1997. [33] Sugie, J. and Katayama, M., Global asymptotic stability of a predatorprey system of Holling type, Nonlinear Anal-Theor., Vol. 38, Iss. 1, 105-121, 1999. [34] Finke, D. L., and Denno, R. F. (2006). Spatial refuge from intraguild predation: implications for prey suppression and trophic cascades. O ecologia, 149(2), 265-275. 28 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD (a) (b) (c) (d) (e) Figure 11. Bifurcation diagrams of the model (27) illustrating (a) SN at a = a = 0:44476, (b) SN at a = a = 0:5625, (c) SN 1 2 1 2 at w = w = 0:24889, (d) SN at w = w = 4:97778, (e) SN at 0 1 0 1 b = b = 0:064498. Here r = 0:3 and other parameter sets are given in Table 3. (SN: Saddle-node bifurcation.) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Mathematics arXiv (Cornell University)

Dynamics of a predator-prey model with generalized Holling type functional response and mutual interference

Loading next page...
 
/lp/arxiv-cornell-university/dynamics-of-a-predator-prey-model-with-generalized-holling-type-xgri4YMms8
ISSN
0025-5564
eISSN
ARCH-3343
DOI
10.1016/j.mbs.2020.108407
Publisher site
See Article on Publisher Site

Abstract

DYNAMICS OF A PREDATOR-PREY MODEL WITH GENERALIZED HOLLING TYPE FUNCTIONAL RESPONSE AND MUTUAL INTERFERENCE 1 2 3 Kwadwo Antwi-Fordjour , Rana D. Parshad , Matthew A. Beauregard 1) Department of Mathematics and Computer Science, Samford University, Birmingham, AL 35229, USA 2) Department of Mathematics, Iowa State University, Ames, IA 50011, USA 3) Department of Mathematics and Statistics, Stephen F. Austin State University, Nacogdoches, TX 75962, USA Abstract. Mutual interference and prey refuge are important drivers of predator- prey dynamics. The \exponent" or degree of mutual interference has been un- der much debate in theoretical ecology. In the present work, we investigate the interplay of the mutual interference exponent, on the behavior of a predator- prey model with a generalized Holling type functional response. We investigate stability properties of the system and derive conditions for the occurrence of saddle-node and Hopf-bifurcations. A sucient condition for extinction of the prey species has also been derived for the model. In addition, we investigate the e ect of a prey refuge on the population dynamics of the model and derive conditions for the prey refuge that would yield persistence of populations. We provide additional veri cation our analytical results via numerical simulations. Our ndings are in accordance with classical experimental results in ecology [23], that show that extinction of predator and prey populations is possible in a nite time period - but that bringing in refuge can e ectively cause persistence. 1. Introduction Predator-prey dynamics form the corner stone of ecosystems. Mathematical models for such interactions goes back to the work of Lokta, Volterra, Holling and Gause [21, 23, 24, 26]. Holling's classical work proposes that a predators feeding rate depends solely on the prey density, and is modeled essentially by a saturating f(x) function called the functional response, described via p(x) = , where h is the 1+hf(x) handling time of one prey item and f (x) is a function of prey density x. Typically f is smooth, making the response p smooth, and depending on the form of f we f(x) have Holling type II, III, IV responses [16, 24]. The response, p(x) = , can 1+hf(x) (f(x)) be considered a special form of 2 di erent response types. p(x) = , or 1+h(f(x)) f(x) p(x) = . When p = 1, or m = 1, we recover the classical response 1+hf(x) posed earlier. 2010 Mathematics Subject Classi cation. 34D45; 37C10; 37C75; 37G15; 92B05. Key words and phrases. generalized interference; stability analysis; nite time extinction; Hopf-bifurcation; prey refuge. arXiv:2003.02712v1 [math.DS] 5 Mar 2020 2 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD An interesting subclass of these general responses are the cases when 0 < p < 1 or 0 < m < 1. In these cases p(x) is non-smooth, causing various diculties in the mathematical analysis of these systems. For example, linearization about the trivial steady state is no longer possible [32]. Such responses were considered by Sugie [32, 33] and more recently by Braza [27]. However these works miss a key dynamic inherent in such models - that of nite time extinction. Note, in these cases one might ask what real ecological scenarios do these models represent. The work by Sugie, proposed the (f (x)) term, as indicative of a predator which is highly ecient and with a high attack rate. Mutual interference is de ned as the behavioral interactions among feeding or- ganisms, that reduce the time that each individual spends obtaining food, or the amount of food each individual consumes [1, 2, 22, 30]. Some of the earliest work on mutual/predator interference, was initiated by Erbe [25], in which the mutual in- f(x) terference is modeled as y , where y is the predator density and 0 < m < 1. 1+hf(x) The exact value of the exponent m has been under much debate in ecology [18]. f(x) Various authors describe the response p(x) = in terms of mutual 1+hf(x) interference. This direction was rst considered by Upadhyay and Rao [31]. How- ever, an ecological motivation, to the best of our knowledge is not provided. We are motivated by certain theoretical ecology directions [12, 13, 14], and interpret f(x) p(x) = , 0 < m < 1, as a predator with a greater feeding rate, or a 1+hf(x) more aggressive predator, than one which is modeled in the classical scenario - that is when m = 1. This is clear from simple comparison, p(x)j < p(x)j , 1 m =1 0<m <1 1 1 8x > 0. Prey refuge, and its role in predator-prey communities has also been extremely well investigated, since the seminal work of Kar [17]. Refuge is de ned as any strategy taken by prey to avoid predation, such as shelter, dispersal, mimicry and camou age [34]. It can have strong in uence on predator-prey communities [20, 19] - often stabilizing systems, which are otherwise doomed for extinction. It is thus an important ingredient in ecosystem balance and diversity [28]. However, the e ect of refuge on non-smooth systems such as the a ore mentioned ones, remains less investigated [28]. The well known experiments of Gause nd in contradiction to the predictions of classical predator-prey models, that there is a distinct chance for the predator and prey populations to die out - unless the prey is provided with refuge [23]. Non-smooth systems such as when 0 < m; m < 1, in the a ore mentioned models, enable the dynamic of nite time predator-prey extinction (such as seen in the experiments of Gause [23, 19]) - however, to the best of our knowledge, the e ect of prey refuge on these systems has not been investigated. For the purposes of this manuscript we consider the functional response p(x) = f(x) , and de ne the parameter regimes m > 1 as super-critical, that is 1+hf(x) the regime where p(x) 2 C ;8k, and f (x) is a polynomial function. We de ne m = 1 as critical, recovering the classical case from the literature. Lastly we de ne 0 < m < 1 as sub-critical, that is the regime where p(x) looses smoothness, and is not even Lipschitz. Thus the goals of the current manuscript are: (1) To consider a generalized model of interference, in the sub-critical regime; therein to investigate the phenomenon of nite time extinction, that can occur in this regime. MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 3 (2) To investigate this model dynamically, including the various bifurcations that might occur; (3) To investigate the e ect of prey refuge on the dynamics of this generalized model. We nd that there is a critical amount of refuge that prevents nite time extinction of the prey. This is seen via theorem 5.1. The rest of the paper is organized as follows. The mathematical formulation of the problem and mathematical preliminaries such as nonnegativity, boundedness and dissipativeness are presented in Section 2. Existence of equilibria, stability analysis and various local bifurcation analysis are considered in Section 3. In Section 4, we analyze the possibility of nite time extinction of the prey population. We investigate the e ect of prey refuge in Section 5. Additionally, stability analysis and various local bifurcation analysis are carried out. Numerical simulations are performed in Section 6 to correlate with some of our key analytical ndings. In the last section, we present our discussions and conclusions. 2. Model Formulation First, we consider a general predator-prey model with mutual interference among predators of the form dx = x f (x ) w g(x )x ; < 1 1 0 1 dt (1) dx : 2 m = a x + w g(x )x ; 2 2 1 1 dt where f (x ) and g(x ) are the logistic growth and the functional response of the 1 1 predator towards the prey respectively. Assume that 0 < m  1, as per literature on mutual interference [1, 2]. In this paper, we consider the general logistic growth and the generalized Holling type functional response, see [3, 4]: (2) f (x) = a b x; g(x) = : 1 1 x + d Assume that 0 < m  1. The assumptions placed on the functions f and g in (2) are: (I) g is continuous for x  0 and g(0) = 0; (II) g is smooth for x > 0 and g (x ) > 0 for x > 0; 1 1 1 (III) f is smooth for x  0; a a 1 1 (IV) There exists > 0 such that x f (x ) < 0 for x  0, x 6= ; 1 1 1 1 b b 1 1 g(x ) 0 + (V) For 0 < m < 1, g (0 ) := lim + = +1; x !0 (VI) g is not smooth for x = 0 when 0 < m < 1; 1 1 dx (VII) The integral lim converges for xed > 0. !0 g(x ) Thus the predator-prey model with mutual interference and the generalized Holling type functional response becomes dx > 1 x m 2 1 2 < = a x b x w x ; 1 1 1 0 1 2 x +d dt (3) > 1 dx > 2 x m 1 2 = a x + w x ; 2 2 1 x +d dt 4 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD The variables and parameters used in the model are de ned in Table 1. Table 1. List of parameters used in the model (3). All parameters considered are positive constants. Variables/ Parameters Description x Prey population x Predator population t Time a Per capita rate of self-reproduction for the prey a Intrinsic death rate of the predator population w Maximum rate of per capita removal of prey w Measure eciency of biomass conversion from prey to predator b Death rate of prey population due to intra-species competition d Half saturation constant 1=m Predators feeding intensity m Mutual interference exponent 2.1. Mathematical Preliminaries. There are essential properties that a mathematical model must exhibit in order to obtain realistic solutions. In particular, it is important to guarantee positivity of the populations. Likewise, boundedness of the total population is another impor- tant feature of a realistic model. In this section, we present guarantee positivity, boundedness, and dissipativeness of the mathematical model (3). 2.1.1. Positivity and Boundedness. The nonnegativity of populations generated by the mathematical model (3) is clearly important to make biological sense. In addi- tion, positivity implies survival of the populations over the temporal domain. The boundedness of populations ensures that no population supercedes unrealistic val- ues in time. In particular, boundedness guarantees that the total population does not grow beyond an exponential rate for an unbounded interval. Guaranteeing both of these features makes strides to showing the feasibility of a mathematical model for describing population behavior. Lemma 2.1. Consider the following region R = f(x ; x ) : x  0; x  0g, then 1 2 1 2 all solutions (x (t); x (t)) of model (3) with initial conditions x (0) > 0; x (0) > 0 1 2 1 2 are nonnegative for all t  0. Proof. The proof of Lemma (2.1) follows from the proof of Theorem 3:1 in [3]. Lemma 2.2. All solutions (x (t); x (t)) of model (3) with initial conditions x (0) > 1 2 1 0; x (0) > 0 are bounded. 2 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 5 Proof. Let us de ne the function Q (x (t); x (t)) = x (t) + x (t): Then 1 2 1 2 dQ dx dx 1 2 = + dt dt dt m m 1 1 x x 1 m 1 m 2 2 2 = a x b x w x a x + w x : 1 1 1 0 2 2 1 1 2 2 x + d x + d 1 1 Let  be a positive constant such that   a and suppose w  w , then we obtain, 2 0 1 m m 1 1 dQ x x 1 m 1 m 2 2 2 + Q = a x b x w x a x + w x 1 1 1 0 2 2 1 2 2 dt x + d x + d 1 1 + (x + x ) 1 2 1 m 2 2 = a x b x (w w ) x (a )x + x 1 1 1 0 1 2 2 1 x + d (a + ) (a + )x b x  : 1 1 1 4b (a + ) Taking W = and applying the theory on di erential inequality, we obtain 4b W (1 e ) 0  Q (x (t); x (t))  + Q (x (0); x (0)) e ; 1 2 1 2 which implies (4) lim sup Q (x (t); x (t))  : 1 2 t!1 By (4) and Lemma (2.1), all solutions of (3) with initial conditions x (0) > 0; x (0) > 0 will be contained in the region = f(x ; x ) 2 R : Q (x (t); x (t))  + ; for any  > 0g: 1 2 1 2 The proof is complete. 2.1.2. Dissipativeness. In the previous section, it was shown that the total population remains positive and bounded for all time. Here, we showed that the individual populations are all bounded from above. In such a situation, we say that the model is dissipative. Lemma 2.3. The system (3) is dissipative. Proof. From the rst system of (3) dx x 1 1 2 2 = a x b x w x ; 1 1 1 0 1 2 dt x + d a x b x : 1 1 1 This implies that (5) lim sup x (t)  : t!1 1 6 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD The inequality (5) gives that for arbitrary small  > 0; there exist a real number T > 0 such that (6) x (t)  +  ; for all t  T : 1 1 1 Using (6), we obtain for all t  T , d w x dx w dx 0 2 1 0 2 x + = + dt w dt w dt 1 1 w a 0 2 = a x b x x 1 1 1 2 w a 0 2 a x x 1 1 2 w x 0 2 = (a + a )x a x + 1 2 1 2 1 w x 0 2 K a x + ; 1 2 1 where K = (a + a ) +  : Therefore, we obtain 1 1 2 1 w x K 0 2 2 (7) lim sup x +  : w a t!1 1 2 By (5) and (7), there exists a real number K such that lim sup x  K : 2 2 t!1 Thus, for arbitrary small  > 0, there exists T > T > 0, such that for all t  T 2 2 1 2 x  K +  : 2 2 2 Therefore the model (3) is dissipative. 3. Existence of Equilibria In this section, we determine and analyze equilibria for our mathematical model. Consider the solutions to the steady state equations: 2 m (8) a x b x w x = 0 1 1 1 0 1 2 x + d 1 m (9) a x + w x = 0 2 2 1 x + d The above equations, (8) and (9), have three types of non-negative equilibria: (i) The trivial equilibrium E (0; 0); (ii) The predator-free equilibrium E (a =b ; 0); 1 1 1 (iii) The interior equilibrium E (x ; x ) where x and x are related by 1 2 1 2 x = a x b x : 1 1 2 1 1 w a 0 2 We have that a b x  0 since x  0 and x  0. The possible existence 1 1 1 1 2 of a unique or multiple interior equilibria are shown in Fig. (1). MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 7 Figure 1. Figure (a) and (b) represent graphical illustration of the predator and prey non-trivial nullclines when m = m = 0:5. 1 2 Figure (c) represents graphical illustration of the predator and prey non-trivial nullclines when m = 1; m = 0:5. 1 2 3.1. Stability Analysis of the Interior Equilibrium. The variational matrix J of the model (3) around the interior equilibrium E (x ; x ) is 1 2 a a 11 12 J = ; a a 21 22 where m m 1 2  1 dm w x x 1 0 2 1 a = a 2b x ; 11 1 1 m +1 (x + d) m 1 m 2  1 m w x x 2 0 2 1 a = < 0; (x + d) m 1 m 1 a = m dw x > 0; 21 1 1 m +1 (x + d) m 1 2 1 a = a + m w x : 22 2 2 1 x + d The characteristic equation corresponding to J evaluated at E (x ; x ) is given by 1 2 (10)  tr (J ) + det (J ) = 0; 8 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD where tr (J ) = a + a 11 22 1 m 1 x x m 1 m 2 1  2 1 = a a 2b x + m w x dm w x 1 2 1 2 1 1 0 1 2 2 m +1 x + d (x + d) 1 1 and det (J ) = a a a a 11 22 12 21 m m  1 2 1 dm w x x x 1 0 m 1 2 1  2 1 = a 2b x a + m w x 1 1 2 2 1 1 2 m +1 x (x + d) x + d 1 1 1 m 1 m m 1 2  1  1 m w x x x 2 0 m 2 1  2 1 m dw x : 1 1 m m +1 1 1 (x + d) (x + d) 1 1 Here, tr (J ) and det (J ) represents the trace and determinant of the variational matrix. Hence the stability of E (x ; x ) is determined by the sign of det (J ) and 1 2 tr (J ). The above results are encapsulated in the following theorem. Theorem 3.1. The interior equilibrium E (x ; x ) is locally asymptotically stable 1 2 if tr (J ) < 0 and det (J ) > 0 by Routh-Hurwitz stability criteria. Proof. The proof follows directly from the above discussion and hence omitted for brevity. 3.2. Global Asymptotic Stability. Considering the case where m = 1 and assumptions (V)-(VII) hold for our response functions f and g, then E is not a saddle point and the argument using the Poincare-Bendixson theorem cannot be applied. We will describe the global behavior of the system (3) by considering the relative position of the stable and unstable separatrix of the saddle point E . We denote the stable separatrix by s u W (E ) and the unstable separatrix by W (E ). 0 1 The predator nullcline is the vertical line x = x determined by the equation a + w g(x ) = 0. We assume 2 1 1 a d a (11) w > a ; > x := : 1 2 1 1 1 b m m 1 1 1 w a 1 2 The prey nullcline is the graph of the function y = (x ) x f (x ) 1 1 (12) (x ) = w g(x ) 0 1 where f (x ) and g(x ) are de ned in (2). Clearly ( ) = 0 and (x ) > 0 for 1 1 1 0 < x < . The unique interior equilibrium E (x ; x ) is the intersection of the 1 2 1 2 predator and prey nullclines and it can be stable or unstable depending on the sign 0  0  0 of (x ). For (x ) > 0, E is a repeller (unstable) and for (x ) < 0, E is an 2 2 1 1 1 attractor (or locally asymptotically stable). Remark 1. The predator-free equilibrium E turns into a stable node with the loss of the unique interior equilibrium E . 2 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 9 Based on the non-uniqueness of the solution of model (3) when m < 1 and m = 1, we have the following results. s u Lemma 3.2 ( Proposition 3.1 in [6]). Assume that W (E ) is above W (E ). If 0 1 E is a repeller, then it is surrounded by at least one limit cycle. If the system can have at most one cycle, then E is surrounded by at least a unique limit cycle which is orbitally asymptotically stable. This limit cycle is not globally orbitally asymptotically stable, even if it is unique. If E is an attractor and if the system has no cycles, then all orbits under W (E ) converge towards E . E is not globally 0 2 2 asymptotically stable, even if it is not surrounded by any unstable limit cycle. 0  s u Proposition 3.3. Assume (x )  0 and (11) hold, then W (E ) is above W (E ) 0 1 and the basin of attraction of E is the positive region of the plane located under W (E ). Hence E is not globally asymptotically stable. 0 2 Proof. The proof is similar to the proof of Proposition 4.2 in [6] and omitted for brevity. 3.3. Saddle-Node Bifurcation Analysis. We investigate the possibility of saddle- node bifurcation of the positive interior equilibrium E by using the intrinsic death rate of the predator population as a bifurcation parameter. The following theorem states the restrictions for occurrence of a saddle-node bifurcation for model (3). Theorem 3.4. The model (3) undergoes a saddle-node bifurcation around E at a when the system parameters satisfy the restriction det (J ) = 0 along with the condition tr (J ) < 0. Proof. To validate the restriction for the occurrence of saddle-node bifurcation, we apply Sotomayor's theorem [9] at a = a . At a = a , it can be seen that 2 2 2 2 det (J ) = 0 and tr (J ) < 0 which indicates that the Jacobian (J ) admits a zero eigenvalue. Let U and V be the eigenvectors corresponding to the zero eigenvalue T T of the matrix (J ) and (J ) respectively. We obtain that U = (u ; u ) and 1 2 a u a v T 2 2 12 21 V = (v ; v ) , where u = , v = and u ; v 2 Rnf0g. 1 2 1 1 2 2 a a 11 11 T   T Furthermore, let F = (F ; F ) and X = (x ; x ) , where F ; F are given by 1 2 1 2 1 2 2 m F = a x b x w x 1 1 1 1 0 1 2 x + d F = a x + w x : 2 2 2 1 x + d Now T  T V F (X; a ) = (v ; v )(0;x ) = v x 6= 0; a 1 2 2 2 2 2 2 and T 2 V D F (X; a )(U; U ) 6= 0: Hence, from Sotomayor's theorem the model undergoes a saddle-node bifurcation around E at a = a . 2 2 Theorem 3.5. The model (3) undergoes a saddle-node bifurcation around E at w when the system parameters satisfy the restriction det (J ) = 0 along with the condition tr (J ) < 0. 10 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD Theorem 3.6. The model (3) undergoes a saddle-node bifurcation around E at w when the system parameters satisfy the restriction det (J ) = 0 along with the condition tr (J ) < 0. Theorem 3.7. The model (3) undergoes a saddle-node bifurcation around E at b when the system parameters satisfy the restriction det (J ) = 0 along with the condition tr (J ) < 0. Theorem 3.8. The model (3) undergoes a saddle-node bifurcation around E at a when the system parameters satisfy the restriction det (J ) = 0 along with the condition tr (J ) < 0. Proof. The proof of Theorem 3.5, Theorem 3.6, Theorem 3.7 and Theorem 3.8 are similar to proof in Theorem 3.4 and omitted for brevity. 3.4. Hopf-Bifurcation Analysis. We investigate the possibility of Hopf-bifurcation of the positive interior equilibrium E by using the per capita rate of self-reproduction for the prey, a as a bifurcation parameter. Then, the characteristic equation cor- responding to model (3) at E is given by (13)  + A(a ) + B(a ) = 0; 1 1 where A = tr (J ) = (a + a ) and B = det (J ) = a a a a : 11 22 11 22 12 21 The instability of model (3) is demonstrated via the following theorem by con- sidering a as a bifurcation parameter. Theorem 3.9 (Hopf-Bifurcation Theorem [8]). If A(a ) and B(a ) are the smooth 1 1 functions of a in an open interval about a 2 R such that the characteristic equa- tion (13) has a pair of imaginary eigenvalues  =  (a )  i (a ) with  and 1 1 2 R so that they become purely imaginary at a = a and j 6= 0, then a 1 a =a 1 1 da 1 Hopf-bifurcation occurs around E (x ; x ) at a = a (i.e. a stability changes of 2 1 1 2 1 E (x ; x ) accompanied by the creation of a limit cycle at a = a ). 2 1 1 2 1 Theorem 3.10. The model (3) undergoes a Hopf-bifurcation around E (x ; x ) 1 2 when a crosses some critical value of parameter a , where 1 m 1 x x m 1 m 2 1  2 1 a = a + 2b x m w x + dm w x 2 1 2 1 1 0 1 1 2 2 m +1 x + d (x + d) 1 1 provided: (i) A(a ) = 0, (ii) B(a ) > 0, (iii) Re (a )j 6= 0 at a = a ; i = 1; 2. i 1 1 a =a 1 da Proof. Clearly A(a ) and B(a ) are the smooth functions of a . The roots of the 1 1 1 equation (13) are of the form  =  (a ) + i (a ) and  =  (a ) i (a ) where 1 1 1 2 1 1 (a ) and (a ) are real functions. 1 1 At a = a , the characteristic equation (13) reduces to (14)  + B(a ) = 0 p p By solving for the roots of equation (14), we obtain  = i B and  = i B. 1 2 Hence a pair of purely imaginary eigenvalues. Furthermore, we validate the transver- sality condition: Re (a )j 6= 0; i = 1; 2: i 1 a =a da 1 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 11 Substituting (a ) =  (a ) + i (a ) into equation (13), we obtain 1 1 1 (15) ( (a ) + i (a )) + A(a )( (a ) + i (a )) + B(a ) = 0: 1 1 1 1 1 1 Now, taking the derivative with respect to a , we get _ _ 2( (a ) + i (a ))( (a ) + i _ (a )) + A(a )( (a ) + i _ (a )) 1 1 1 1 1 1 1 _ _ +A(a )( (a ) + i (a )) + B(a ) = 0: 1 1 1 1 Separating the real and imaginary parts, we have _ _ _ _ (a )(2 (a ) + A(a )) + (a )(2 (a )) + A(a ) (a ) + B(a ) = 0; 1 1 1 1 1 1 1 1 which implies (16)  (a )Z (a ) _ (a )Z (a ) + Z (a ) = 0; 1 1 1 1 2 1 3 1 and _ _ (a )(2 (a )) + _ (a )(2 (a ) + A(a )) + A(a ) (a ) = 0; 1 1 1 1 1 1 1 which implies (17)  (a )Z (a ) + _ (a )Z (a ) + Z (a ) = 0; 1 2 1 1 1 1 4 1 _ _ where Z (a ) = 2 (a ) +A(a ), Z (a ) = 2 (a ), Z (a ) = A(a ) (a ) +B(a ) and 1 1 1 1 2 1 1 3 1 1 1 1 Z (a ) = A(a ) (a ). 4 1 1 1 Multiplying equation (41) by Z (a ) and equation (42) by Z (a ) and then 1 1 2 1 adding them, we obtain 2 2 (18) (Z (a ) + Z (a )) (a ) + Z (a )Z (a ) + Z (a )Z (a ) = 0; 1 1 1 1 1 3 1 2 1 4 1 1 2 thus solving for  (a ) from equation (43) and at a = a , 1 1 d [Z (a )Z (a ) + Z (a )Z (a )] 1 3 2 4 1 1 1 1 Re (a )j =  (a ) = : i 1 a =a 1 1 2  2 da Z (a ) + Z (a ) 1 1 2 1 2  2 It is easy to verify that Z (a )Z (a ) + Z (a )Z (a ) 6= 0 and Z (a ) + Z (a ) 6= 0 1 3 2 4 1 1 1 1 1 1 2 1 which implies Re (a )j 6= 0. Hence, a Hopf-bifurcation occurs around i 1 a =a da 1 E (x ; x ) at a = a . 2 1 1 2 1 4. Finite Time Extinction An interesting property of (3) is that the population of prey may go extinct in nite time for carefully chosen initial conditions. Therefore, while solutions do remain nonnegative, it is clear that the populations may become extinct and no longer persist. Theorem 4.1. Consider the predator-prey system given by (3). The solution x (t) to the prey equation x (t) with initial conditions x (0) > 0; x (0) > 0 can go extinct 1 1 2 in nite time, for suciently chosen initial conditions. Proof. Consider the substitution x = 1=u in the prey equation of (3). This yields the following system: 1 1 dx 1 du 1 1 2 2 > u < = = a b ( ) w x ; 1 1 0 1 2 +d dt u dt u u (19) 1 m > 1 dx : m u 2 = a x + w x : 2 2 1 1 +d dt u 12 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD This system can be simpli ed into the system in u; x : du u > m = a u + b + w x ; < 1 1 0 dt (1 + du) (20) >dx 1 > m = a x + w x ; 2 2 1 dt (1 + du) Note, via positivity dx (21)  a x : 2 2 dt Thus, a t (22) x  x (0)e : 2 2 Also, via positivity we have the inequality, 2 2 du u u m a t m 2 2 2 (23)  a u + w x  a u + w (x (0)e ) : 1 0 1 0 2 m m 1 1 dt (1 + du) (1 + du) Note that the solution to the di erential equation du ~ u ~ (24)  a u ~ + w 1 0 dt (1 + du ~) will blow up in a nite time, T (u ) < 1, as long as the initial data u ~(0) = u 0 0 satis es, m 2 (25) a u (1 + du )  w u : 1 0 0 0 Now if we choose x (0)  1, such that a t m 2 2 (26) (x (0)e ) > 1; t 2 [0; T ]; then u  u ~ on [0; T ], and must blow-up in nite time, at some T < T , by comparison, if u is chosen to satisfy (25) . Therefore, lim u ! 1 t!T <1 which implies 1 1 lim x = lim = ! 0; t!T <1 t!T <1 u lim u t!T <1 but that implies x (t) goes extinct in nite time for x (0) chosen large enough and 1 2 1m m 1 1 (x (0)) (x (0) + d)  : 1 1 5. The effect of prey refuge In the previous section it was shown that the prey population may go extinct in nite time. Therefore, we seek to investigate the e ect of protecting the prey from predation with their habitat. The aim is to provided avenues for which the prey population will persist. Here, using a similar ideas from [20], we introduce a prey refuge. A discussion of how a habitat controller may create a prey refuge is provided in section 7 of [20]. Essentially, one must protect a constant proportion of prey by replacing the predation term g(x ) by g(rx ), where 0  r  1. Here, r is a refuge parameter, 1 1 such that if r = 0 then complete protection of the prey is provided while r = 1 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 13 implies no protection and the original system (1) is recovered. Thus, we write the following system that models prey refuge as 8 m dx rx 1 1 > 2 2 > = a x b x w x ; 1 1 1 0 < 1 2 dt rx + d (27) dx rx 2 1 > m = a x + w x ; 2 2 1 dt rx + d We now state our rst result concerning prey refuge, Theorem 5.1. Consider the predator-prey system given by (27) for r = 1, that is no refuge. For any initial conditions > x (0) > 0; x (0) > 0 s.t. the 1 2 solution x (t) to the prey equation goes extinct in nite time, there exists a refuge r (x (0); a ; b ; d; w ; m ; m ; K ) > 0 s.t. for any r < r the solution x (t) to the 1 1 1 0 1 2 2 1 prey equation does not go extinct in nite time. Proof. The proof follows in similar fashion to the proof of Theorem (4.1). Hence, we make the substitution x = 1=u in the prey equation in (27). This results in: du u > m m 1 2 = a u + b + w r x ; 1 1 0 dt (r + du) (28) dx r > m = a x + w x ; 2 2 1 dt (r + du) Via comparison we have du dt m m 1 2 = a u + b + w r x 1 1 0 (r + du) w r 0 m 2m 2 a u + b + u x 1 1 w r 2m m 1 2 (29)  a u + b + u K 1 1 Recall that x and x are bounded above. Let the upper bounds for x and x be 1 2 1 2 du constants K and K , respectively. Now, we desire that  0, 8t, as this will 1 2 dt ensure that u cannot blow-up in nite time, or x cannot go extinct in nite time. lets compare to du w r 2m m 1 2 (30) = a u + b + u K 1 1 dt d let us set u = + v, this changes (30) to dv w r b 0 1 2m m 1 2 (31) = a v + ( + v) K dt d a du du Now, for =  0 one requires that dt dt w r b 0 1 2m m 1 2 (32) a v + ( + v) K  0 d a 1 14 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD This is possible if we choose r such that, ! 1 (a d )v(0) (33) r < : b m 1 2m 2 w ( + v(0)) K This is seen simply by looking in the phase for the v equation, and shows, v cant blow up in nite time. Thus neither can u = + v, thus making the nite time extinction of x an impossibility. 5.1. Existence of Equilibria and Stability Analysis. Similar to Section 3, we investigate and analyze the equilibrium solutions of our mathematical model with prey refuge. Consider the steady state equations of (27): rx 2 m (34) a x b x w x = 0 1 1 1 0 rx + d rx (35) a x + w x = 0 2 2 1 rx + d The above equations (34) and (35) have three types of non-negative equilibria: (i) The trivial equilibrium E (0; 0). (ii) The predator-free equilibrium E (a =b ; 0). 1 1 1 (iii) The interior equilibrium E (x ; x ) where x and x are related by 1 2 1 2 x = a x b x : 1 1 2 1 1 w a 0 2 The variational matrix J of the model (27) around the interior equilibrium E (x ; x ) is 1 2 b b 11 12 J = : b b 21 22 where m 1 r r x rx m 1 1 b = a 2b x m w x ; 11 1 1 1 0 1 2 d + rx (rx + d) rx + d 1 1 1 m 1 m 2  1 m w x (rx ) 2 0 2 1 b = ; (rx + d) m 1 m 1 m  2 b = m dw r x ; 21 1 1 m +1 (rx + d) rx m 1 2 1 b = a + m w x : 22 2 2 1 rx + d The characteristic equation corresponding to J is given by tr (J ) + det (J ) = 0; r r where tr (J ) = b + b 11 22 m 1 r r x rx m 1 1 = a a 2b x m w x 1 2 1 1 0 1 2 d + rx (rx + d) rx + d 1 1 1 rx m 1 2 1 + m w x ; 2 1 rx + d 1 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 15 and det (J ) = b b b b 11 22 12 21 m 1 r r x rx 2 1 1 = a 2b x m w x 1 1 1 0 1 2 d + rx (rx + d) rx + d 1 1 1 rx m 1 2 1 a + m w x 2 2 1 rx + d m 1 m m 1 2  1  1 m w x (rx ) x 2 0 2 1 m m 1 1 2 m dw r x : 1 1 m  m +1 1 1 (rx + d) (rx + d) 1 1 Here, tr (J ) and det (J ) represents the trace and determinant of the variational r r matrix. Hence the stability of E (x ; x ) is determined by the sign of det (J ) and 1 2 r tr (J ). The above results are summarized in the following theorem. Theorem 5.2. The interior equilibrium E (x ; x ) of system (27) is locally asymp- 1 2 totically stable if tr (J ) < 0 and det (J ) > 0 by Routh-Hurwitz stability criteria. r r Proof. The proof follows directly from the above discussion and hence omitted for brevity. 5.2. Bifurcation Analysis. In this subsection, we analyze the qualitative changes in the dynamical behavior of model (27) under the e ect of varying a speci c parameter. The conditions and restrictions for the occurrence of saddle-node, Hopf, and transcritical bifurcations are derived. The classi cation are of codimension one bifurcations. 5.2.1. Saddle-Node Bifurcation. We investigate the possibility of saddle-node bifur- cation of the positive interior equilibrium E by using the intrinsic death rate of the predator population as a bifurcation parameter. The following theorem states the restrictions for occurrence of a saddle-node bifurcation for model (27). Theorem 5.3. The model (27) undergoes a saddle-node bifurcation around E at a when the system parameters satisfy the restriction det (J ) = 0 along with the 2 r condition tr (J ) < 0. Proof. To validate the restriction for the occurrence of saddle-node bifurcation, we apply Sotomayor's theorem [9] at a = a . At a = a , it can be seen that 2 2 2 2 det (J ) = 0 and tr (J ) < 0 which indicates that the Jacobian (J ) admits a zero r r r eigenvalue. Let U and V be the eigenvectors corresponding to the zero eigenvalue T T of the matrix (J ) and (J ) respectively. We obtain that U = (u ; u ) and 1 2 r r b u b v T 2 2 12 21 V = (v ; v ) , where u = , v = and u ; v 2 Rnf0g. 1 2 1 1 2 2 b b 11 11 T   T Let G = (G ; G ) and X = (x ; x ) , where G ; G are given by 1 2 1 2 1 2 rx 2 m (36) G = a x b x w x ; 1 1 1 1 0 1 2 rx + d rx 1 m (37) G = a x + w x : 2 2 2 1 rx + d 1 16 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD Now T  T V G (X; a ) = (v ; v )(0;x ) = v x 6= 0; a 1 2 2 2 2 2 2 and T 2 V D G(X; a )(U; U ) 6= 0: Hence from Sotomayor's theorem the model (27) undergoes a saddle-node bifurca- tion around E at a = a . 2 2 Theorem 5.4. The model (27) undergoes a saddle-node bifurcation around E at w when the system parameters satisfy the restriction det (J ) = 0 along with the 1 r condition tr (J ) < 0. Theorem 5.5. The model (27) undergoes a saddle-node bifurcation around E at w when the system parameters satisfy the restriction det (J ) = 0 along with the 0 r condition tr (J ) < 0. Theorem 5.6. The model (27) undergoes a saddle-node bifurcation around E at b when the system parameters satisfy the restriction det (J ) = 0 along with the 1 r condition tr (J ) < 0. Theorem 5.7. The model (27) undergoes a saddle-node bifurcation around E at a when the system parameters satisfy the restriction det (J ) = 0 along with the 1 r condition tr (J ) < 0. Proof. The proof of Theorem 5.4, Theorem 5.5, Theorem 5.6 and Theorem 5.7 are similar to proof in Theorem 5.3 and omitted for brevity. 5.2.2. Transcritical Bifurcation. Here, we investigate the possibility of the existence of a transcritical bifurcation for the model (27). Transcritical bifurcation occurs when an equilibrium point interchanges its stability when it collides with another equilibrium point as a parameter is varied. The prey refuge parameter r is used as a bifurcation parameter. The variational matrix J of the model (27) for 0 < m < 1 and m = 1 r 1 2 evaluated at 0 1 b d a @ A r = r = 1 1 a m m 1 1 w a 1 2 around the predator-free equilibrium E (a =b ; 0) is given by 1 1 1 2   3 r a a w 1 0 4  5 J = : r r a + b d 1 1 1 1 0 0 At r = r , the matrix J has a negative eigenvalue and a zero eigenvalue. Let U and V be the eigenvectors corresponding to the zero eigenvalue of the matrix (J  ) and (J ) respectively. Then r r a b d 1 1 U = 1; 1 + ; V = (0; 1) : w r a 0 1 1 MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 17 T T Let G = (G ; G ) and X = (a =b ; 0) , where G ; G are de ned in (36) and (37). 1 2 1 1 1 2 Now we have T  T V G (X; r ) = (0; 1)(0; 0) = 0; additionally V [DG (X; r )U ] 6= 0 and T 2 V D G(X; r )(U; U ) 6= 0: Hence using Sotomayor's theorem the model (27) undergoes a transcritical bifur- cation around E when the refuge r crosses the critical value of the parameter r . The above results are summarized in the following theorem. Theorem 5.8. The model (27) undergoes a transcritical bifurcation around E (a =b ; 0) 1 1 1 when the refuge r crosses the critical value of parameter r , where 0 1 b d a @ A r = . 1 1 1 a m m 1 1 w a 1 2 5.2.3. Hopf-Bifurcation. We investigate the possibility of Hopf-bifurcation of the positive interior equilibrium E by using the per capita rate of self-reproduction for the prey, a as a bifurcation parameter. Then, the characteristic equation corresponding to model (27) at E is given by (38)  + A (a ) + B (a ) = 0; 1 1 1 1 where A = tr (J ) = (b + b ) and B = det (J ) = b b b b . 1 11 22 1 11 22 12 21 r r The instability of model (27) is demonstrated via the following theorem by con- sidering a as a bifurcation parameter. Theorem 5.9. The model (27) undergoes a Hopf-bifurcation around E (x ; x ) 1 2 when a crosses some critical value of parameter a , where m 1 r r x rx 2 1 1 a = a + 2b x + m w x 2 1 1 0 1 1 2 d + rx (rx + d) rx + d 1 1 1 rx m 1 1 m w x ; 2 1 rx + d provided (i) A (a ) = 0, 1 1 (ii) B (a ) > 0, 1 1 (iii) Re (a )j 6= 0 at a = a ; i = 1; 2. i 1 1 a =a 1 da Proof. Clearly A (a ) and B (a ) are the smooth functions of a . The roots of the 1 1 1 1 1 equation (38) are of the form  = #(a ) + i$(a ) and  = #(a ) i$(a ) where 1 1 1 2 1 1 #(a ) and $(a ) are real functions. 1 1 At a = a , the characteristic equation (38) reduces to (39)  + B (a ) = 0 1 1 18 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD p p By solving for the roots of equation (39), we obtain  = i B and  = i B . 1 1 2 1 Therefore, we have purely imaginary eigenvalues. Hence, we are left with validating the transversality condition. Namely, Re (a )j 6= 0; i = 1; 2: i 1 a =a da Substituting (a ) = #(a ) + i$(a ) into equation (38), we obtain 1 1 1 (40) (#(a ) + i$(a )) + A (a )(#(a ) + i$(a )) + B (a ) = 0: 1 1 1 1 1 1 1 1 Upon taking the derivative with respect to a we obtain: _ _ 2(#(a ) + i$(a ))(#(a ) + i$ _ (a )) + A (a )(#(a ) + i$ _ (a )) 1 1 1 1 1 1 1 1 _ _ +A (a )(#(a ) + i$(a )) + B (a ) = 0: 1 1 1 1 1 1 Separating the real and imaginary parts, we have _ _ _ #(a )(2#(a ) + A (a )) + $(a )(2$(a )) + A (a )#(a ) + B (a ) = 0; 1 1 1 1 1 1 1 1 1 1 1 which implies (41) #(a )Z (a ) $ _ (a )Z (a ) + Z (a ) = 0; 1 1 1 1 2 1 3 1 and _ _ #(a )(2$(a )) + $ _ (a )(2#(a ) + A (a )) + A (a )$(a ) = 0; 1 1 1 1 1 1 1 1 1 which implies (42) #(a )Z (a ) + $ _ (a )Z (a ) + Z (a ) = 0; 1 2 1 1 1 1 4 1 _ _ where Z (a ) = 2#(a ) +A (a ), Z (a ) = 2$(a ), Z (a ) = A (a )#(a ) +B (a ) 1 1 1 1 1 2 1 1 3 1 1 1 1 1 1 and Z (a ) = A (a )$(a ). Multiplying equation (41) by Z (a ) and equation (42) 4 1 1 1 1 1 1 by Z (a ) and then adding them, we obtain 2 1 2 2 (43) (Z (a ) + Z (a ))#(a ) + Z (a )Z (a ) + Z (a )Z (a ) = 0; 1 1 1 1 1 3 1 2 1 4 1 1 2 thus solving for #(a ) from equation (43) and at a = a , 1 1 d [Z (a )Z (a ) + Z (a )Z (a )] 1 3 2 4 1 1 1 1 Re (a )j = #(a ) = : i 1 a =a 1 1 1 2  2 da Z (a ) + Z (a ) 1 1 2 1 2  2 It is easy to verify that Z (a )Z (a )+Z (a )Z (a ) 6= 0 and Z (a )+Z (a ) 6= 0 1 3 2 4 1 1 1 1 1 1 2 1 which implies Re (a )j 6= 0. Hence, a Hopf-bifurcation occurs around i 1 a =a da 1 E (x ; x ) at a = a . 2 1 1 2 1 Theorem 5.10. The model (27) undergoes a Hopf-bifurcation around E (x ; x ) 1 2 when the refuge r crosses some critical value of parameter r provided (i) A (r) = 0, (ii) B (r) > 0, (iii) Re (r)j 6= 0 at r = r ; i = 1; 2. r=r 1 dr Proof. The proof of Theorem 5.10 is similar to proof in Theorem 5.9 and omitted for brevity.  MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 19 6. Numerical Simulations Numerical simulations of model (3) are performed in this section to correlate with some of our key analytical ndings. The numerical simulations and gures have been developed using MATLABr R2019b, MATCONT[7], and XPPAUT[5]. For convenience, the parameters used in simulations are given in Table 2. Table 2. Parameters used in the simulations of Figs. 2, 3, 4. 7, 8, and 9. a = 0:6 a = 1 b = 0:063 w = 1 d = 2 1 2 1 0 w = 2 m = 0:8 m = 1 1 1 2 For the parameter values in Table 2, the predator-free equilibrium point E (9:52381; 0) is a saddle and E (1:45094; 1:47587) is a repeller, see Fig. 2(a). In Fig. 2(a), s u W (E ) is above W (E ) and E (1:45094; 1:47587) is surrounded by a limit cycle. 0 1 2 It is seen that in Fig. 2(b), when a = 2, b = 0:21 and all the other parameter 1 1 s u sets are given in Table 2, then W (E ) is under W (E ), where E (9:52381; 0) 0 1 1 and E (1:45094; 4:91957). So the predator-free equilibrium point E (9:52381; 0) is 2 1 a saddle and E (1:45094; 4:91957) is a repeller. Here all positive solutions converge towards E . In Fig. 3, when a = 1:5, b = 0:18, a = 1:4 and all the other 0 1 1 2 parameter sets are given in Table 2, we observe that E (3:55994; 4:36963) is an s u attractor and E (8:33333; 0) is a saddle. Also W (E ) is above W (E ) and the 1 0 1 model (3) has a unique limit cylce whose basin of attraction is the region under W (E ). However, in n Fig. 3, the numerical simulations illustrate that E is not 0 2 globally asymptotically stable for the given parameter set. (a) (b) Figure 2. The predator and prey nullclines for model (3) are represented by turquoise and red respectively. (a) W (E ) is above u u s W (E ): E is unstable (b) W (E ) is above W (E ), here a = 2 1 2 1 0 1 and b = 0:21: E is unstable. Other parameter sets are given in 1 2 Table 2. 20 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD s u Figure 3. E is an attractor and W (E ) is above W (E ). Here 2 0 1 a = 1:5, b = 0:18 and a = 1:4. Other parameter sets are given 1 1 2 in Table 2. Figure 4. Bifurcation diagrams of the model (3), as a changes. The stable and unstable interior equilibriums are given by the lines in red and black, respectively. The solid circles (green) represent stable limit cycles and the open circles (blue) represent unstable limit cycles. (a) prey (x ) (b) predator (x ). Parameter set are 1 2 given in Table (2). Furthermore, for the parameter sets in Table 2, we employ AUTO as imple- mented in the continuation software XPPAUT to analyze the bifurcation dia- grams of the model (3) in Fig. 4. The model undergoes Hopf-bifurcation around E (1:45094; 0:49456) as the parameter a crosses its critical value a = 0:261835. 2 1 The branch of periodic orbits emitting from a are stable and the rst Lyapunov coeicient [9],  = 1:49929e < 0 (obtained with the aid of MATCONT), hence the Hopf-bifurcation is supercritical. MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 21 Table 3. Parameters used in the simulations of Figs. 5, 6, 10, and 11. a = 0:5 a = 0:7 b = 0:05 w = 0:2 d = 0:2 1 2 1 0 w = 4 m = 0:5 m = 0:5 1 1 2 Figure 5. Time series depicting the stability behavior of the in- terior equilibrium point (6:67563; 31:7032) for the parameter sets given in Table 3. Also, for the parameter sets in Table 3 of model (3), we obtain the following 1 2 two interior equilibrium points E (3:12437; 30:6886) and E (6:67563; 31:7032) and 2 2 the predator-free equilibrium point is E (10; 0). The eigenvalues associated with 1 1 E (3:12437; 30:6886) are 0:343043 and 0:170265, hence E is a saddle. The eigen- 2 2 2 2 values associated with E (6:67563; 31:7032) are 0:34517 and 0:17481, hence E 2 2 is locally asymptotically stable, see Fig. 5. We note here that, E (10; 0) cannot be analyzed using the linear stability method since m = m = 0:5 < 1. We observed 1 2 that the model (3) undergoes a saddle-node bifurcation around E (x ; x ) when 1 2 the following bifurcation parameters a ; a ; w ; w and b crosses their correspond- 1 2 0 1 1 ing critical values a = 0:46809; a = 0:61515; w = 0:22759; w = 4:55175; and 1 2 0 1 b = 0:05722 respectively. The saddle-node bifurcation diagrams are depicted in Fig. 6 Now, we perform numerical simulations of model (27) to verify some of our ana- lytical results. For r = 0:3 and all other parameter values given in Table 2, the predator-free equilibrium point E (9:52381; 0) is a saddle and E (4:83648; 9:52147) 1 2 is an attractor (stable), see Fig. 7. By introducing a prey refuge of r = 0:3, we 22 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD (a) (b) (c) (d) (e) Figure 6. Bifurcation diagrams illustrating (a) SN at a = a = 0:46809, (b) SN at a = a = 0:61515, (c) SN at w = w = 2 0 2 0 0:22759, (d) SN at w = w = 4:55175, (e) SN at b = b = 1 1 1 1 0:05722. Other parameter sets are given in Table 3. (SN: Saddle- node bifurcation.) MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 23 s u observed in Fig. 7(c) that W (E ) is above W (E ) as compared to Fig. 2(b) 0 1 s u where W (E ) is below W (E ). Thus the stability of the interior equilibrium is 0 1 altered and not all positive solutions tend toward E . (a) (b) (c) Figure 7. Stable dynamics (a) time series (b) phase diagram of the interior equilibrium point (4:83648; 9:52147) (c) E is an at- s u tractor and W (E ) is above W (E ). Here r = 0:4 and all other 0 1 parameter sets are given in Table 2 24 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD (a) (b) Figure 8. Bifurcation diagrams of the model (27), as a crosses its critical value a . The stable and unstable interior equilibriums are given by the lines in red and black, respectively. The solid circles (green) represent stable limit cycles and the open circles (blue) represent unstable limit cycles. (a) prey (x ) (b) predator (x ). Parameter set are given in Table (2). (a) (b) Figure 9. Bifurcation diagrams of the model (27), as r crosses its critical value r . The stable and unstable interior equilibriums are given by the lines in red and black, respectively. The solid circles (green) represent stable limit cycles. (a) prey (x ) (b) predator (x ). Parameter set are given in Table (2). Additionally, for r = 0:3 and all the other parameter sets provided in Table 2, we use AUTO as implemented in the continuation software XPPAUT to analyze the bifurcation diagrams of the model (27) in Fig. 8. The model undergoes Hopf- bifurcation around E (4:83648; 5:49507) as the parameter a crosses its critical value 2 1 a = 0:87278. The branch of periodic orbits emitting from a are stable and the 1 1 rst Lyapunov coeicient [9],  = 2:88256e < 0, hence the Hopf-bifurcation is supercritical. MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 25 Furthermore, the model undergoes Hopf-bifurcation around E (3:32486; 2:59694) as the parameter r crosses its critical value r = 0:43639, see Fig. 9. The branch of periodic orbits bifurcation from r are stable and the rst Lyapunov coecient is = 4:99384e , hence supercritical. Also, the model (27) undergoes transcritical bifurcation around E (9:52381; 0) when the parameter r crosses its threshold r = 0:15239, see Fig. 9(a) . Figure 10. Time series depicting the stability behavior of the interior equilibrium point (7:03041; 29:8249) for r = 0:3 and all other parameter sets are given in Table 3. Next, for r = 0:3 and the other parameter sets given in Table 3 of the model (27), we obtain the following two interior equilibrium points E (2:30292; 25:3225) and E (7:03041; 29:8249) and the predator-free equilibrium point is E (10; 0). The eigenvalues associated with E (2:30292; 25:3225) are 0:32246 and 0:19897, hence 1 2 E is a saddle. The eigenvalues associated with E (7:03041; 29:8249) are 0:33157 2 2 and 0:22790, hence E is locally asymptotically stable, see Fig. 10. We note here that, E (10; 0) cannot be analyzed using the linear stability method since m = m = 0:5 < 1. We observed that the model (27) undergoes a saddle-node bi- 1 2 furcation around E (x ; x ) when the following bifurcation parameters a ; a ; w ; w 2 1 2 0 1 1 2 and b crosses their corresponding critical values a = 0:44476; a = 0:5625; w = 1 2 0 0:24889; w = 4:97778; and b = 0:064498 respectively. The saddle-node bifurcation 1 1 diagrams are presented in Fig. 11. 7. Discussions and Conclusions In this work, we consider a predator-prey model, that allows us to model both the feeding intensity of the predator, as well as the e ect of mutual/predator in- terference. Through numerical simulations, it has been noticed that based on the non-uniqueness of the solutions of model (3), when m < 1 and m = 1, the interior 1 2 equilibrium E is not globally asymptotically stable when it is an attractor (see Fig. 3). We observe that the per capita rate of self-reproduction a plays an important role because the interior equilibrium point E changes stability at the bifurcation 2 26 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD point a (see Fig. 4). The limit cycle through the bifurcation point is stable hence a supercritical Hopf-bifurcation. Furthermore, the e ect of prey refuge is also considered in model (27) - thus one can see the interplay of all of these factors in this model. The model possesses a rich array of dynamical behavior. We have established analytically the occurrence of various local bifurcations including saddle-node, transcritical and Hopf bifurcations. The occurrence of these local bifurcations are well supplemented with one parameter bifurcation diagrams (see Figs. 8, 9, and 11). Prey extinction in nite time is also possible - for large enough initial predator density, and small enough initial prey s u density. Moreover, we observed that when W (E ) is above W (E ), all solutions 0 1 with initial conditions above W (E ) goes to prey extinction in nite time (see Fig. 2(a)). This is in line with the result in [6]. Thus, from a practical point of view increasing m or decreasing the feeding intensity of the predator, will maintain ecosystem balance, as this decreases the predator nullcline, decreasing predator numbers and increasing prey numbers. Stability in the system can also be maintained via provision of the prey with refuge. This is rigorosly established via theorem 5.1. The requisite condition for a critical refuge, for persistence, derived via the theorem sheds light on various ecological scenarios. The ecological validity of the prey extinction state (0; x ) is questionable. In the experiments of Gause [23], once the prey has gone extinct the predator population also crashes, as there is no alternative/additional food in the experimental system. In a real scenario however, such a state might be indicative of a predator having switched to another food source after its primary source has depleted or surviving on additional food, such as in a bio-control situation [29]. Conflict of Interest The authors declare there is no con ict of interest in this paper. References [1] Hassell, M. (1971) Mutual interference between searching insect parasites. J. Anim. Ecol. 40, 473-486. [2] Hassell, M. (1975) Density dependence in single species population. . J. Anim. Ecol. 44, 283-295. [3] Upadhyay, R.K., Agrawal, R. (2015) Modeling the e ect of mutual interference in a delay- induced predator-prey system, J. Appl. Math. Comput. 49, 13-39. [4] Upadhyay, R.K., Parshad, R., Antwi-Fordjour, K., Quansah, E., Kumari, S. (2019) Global dynamics of stochastic predator-prey with mutual interference and prey defense. J. Appl. Math. Comput. 60, 169-190. [5] Ermentrout, B. (2002) Simulating, analyzing, and animating dynamical systems: a guide to XPPAUT for researchers and students. vol 14, SIAM. [6] Beroual, N., Sari, T. (2019) A predator-prey system with Holling-type functional response. hal-02002894. [7] Dhooge, A., Govaerts, W., Kuznetsov, Yu. A., Meijer, H.G.E., Sautois, B. (2009) New features of the software MatCont for bifurcation analysis of dynamical systems, MCMDS, Vol. 14, No. 2, pp 147-175. [8] Murray, J.D. (1993) Mathematical biology, Springer, New York. [9] Perko, L. (2013) Di erential equations and dynamical systems. Vol. 7, Springer Science & Business Media. [10] Kivan, V. (1996). Optimal foraging and predatorprey dynamics. theoretical population bi- ology, 49(3), 265-290. [11] V. Krivan, Evolutionary stability of optimal foraging: partial preferences in the diet and patch models, J Theor Biol, 267:486-494, 2010. MUTUAL INTERFERENCE WITH GENERALIZED HOLLING TYPE RESPONSE 27 [12] McKenzie, H. W., Merrill, E. H., Spiteri, R. J., & Lewis, M. A. How linear features alter predator movement and the functional response. Interface focus, 2(2), 205-216, 2012. [13] Mols, C. M., van Oers, K., Witjes, L. M., Lessells, C. M., Drent, P. J., & Visser, M. E. . Central assumptions of predatorprey models fail in a seminatural experimental system. Proceedings of the Royal Society of London B: Biological Sciences, 271(Suppl 3), S85-S87, [14] Ruxton, G. D. Increasing search rate over time may cause a slower than expected increase in prey encounter rate with increasing prey density. Biology letters, 1(2), 133-135, 2005. [15] Christos C. Ioannou, Graeme D. Ruxton, Jens Krause, Search rate, attack probability, and the relationship between prey density and prey encounter rate, Behavioral Ecology, Vol- ume 19, Issue 4, July-August 2008, Pages 842846, https://doi.org/10.1093/beheco/arn038 Bulletin of Mathematical Biology, 73(10):2249-2276, 2011. [16] Y. Kuang, Some mechanistically derived population models, Math. Biosci. Eng, 4(4), 1-11, [17] Kar, T. K. Stability analysis of a preypredator model incorporating a prey refuge, Commu- nications in Nonlinear Science and Numerical Simulation, 10(6), 681-691, 2005. [18] DeLong, J. P., & Vasseur, D. A. Mutual interference is common and mostly intermediate in magnitude, BMC ecology, 11(1), 2011. [19] Kivan, V. On the Gause predatorprey model with a refuge: a fresh look at the history. Journal of theoretical biology, 274(1), 67-73, 2011. [20] Parshad, R. D., Quansah, E., Black, K., & Beauregard, M. Biological control via ecological damping: an approach that attenuates non-target e ects, Mathematical biosciences, 273, 23-44. [21] J. R. Beddington, Mutual interference between parasites or predators and its e ect on searching eciency J. Animal Ecol., 44, 331-340, 1975. [22] DeAngelis D. L., R. A. Goldstein and R. V. ONeill A model for trophic interaction, Ecology, 56, 881-892, 1975. [23] Gause, G. F. The struggle for existence, Williams & Wilkins, Baltimore, Maryland, USA, [24] Holling, C.S. The components of predation as revealed by a study of small mammal predation of the European pine saw y, Canad. Entomol. 91, 293-320, 1959. [25] Erbe, L. H., & Freedman, H. I. Modeling persistence and mutual interference among sub- populations of ecological communities . Bulletin of mathematical biology, 47(2), 295-304, [26] Lotka A. J. Elements of physical biology. Williams and Wilkins, Baltimore. Reprinted as Elements of mathematical biology, Dover, New York, 1925. [27] Braza, P. A. Predatorprey dynamics with square root functional responses N onlinear Anal- ysis: Real World Applications, 13(4), 1837-1843, 2012. [28] Wang, K. and Zhu, Y. Periodic solutions, permanence and global attractivity of a delayed impulsive preypredator system with mutual interference, N onlinear Analysis: Real World Applications, 14(2), 1044-1054, 2013. [29] Parshad, R. D., Wickramsooriya, S., & Bailey, S. A remark on Biological control through provision of additional food to predators: A theoretical study[Theor. Popul. Biol. 72 (2007) 111120]. T heoretical Population Biology, 2019. [30] Freedman, H. I. Stability analysis of a predator-prey system with mutual interference and density-dependent death rates, B ulletin of Mathematical Biology, 41(1), 67-78, 1979. [31] Upadhyay, R. K., and Rao, V. S. H. Short-term recurrent chaos and role of Toxin Pro- ducing Phytoplankton (TPP) on chaotic dynamics in aquatic systems, C haos, Solitons and Fractals, 39(4), 1550-1564, 2009. [32] Sugie, J., Kohno, R., and Miyazaki, R. On a predator-prey system of Holling type, P roceedings of the American Mathematical Society, 125(7), 2041-2050, 1997. [33] Sugie, J. and Katayama, M., Global asymptotic stability of a predatorprey system of Holling type, Nonlinear Anal-Theor., Vol. 38, Iss. 1, 105-121, 1999. [34] Finke, D. L., and Denno, R. F. (2006). Spatial refuge from intraguild predation: implications for prey suppression and trophic cascades. O ecologia, 149(2), 265-275. 28 ANTWI-FORDJOUR, PARSHAD, BEAUREGARD (a) (b) (c) (d) (e) Figure 11. Bifurcation diagrams of the model (27) illustrating (a) SN at a = a = 0:44476, (b) SN at a = a = 0:5625, (c) SN 1 2 1 2 at w = w = 0:24889, (d) SN at w = w = 4:97778, (e) SN at 0 1 0 1 b = b = 0:064498. Here r = 0:3 and other parameter sets are given in Table 3. (SN: Saddle-node bifurcation.)

Journal

MathematicsarXiv (Cornell University)

Published: Mar 5, 2020

References