Due to its worldwide existence and domination, the progressive connection between predator and prey has been studied. Fear of predators on prey has been shown to have a significant impact on anti-predator defenses and lower prey reproduction in recent zoological studies on terrestrial vertebrates. Prey that is afraid forage less, which may lower fertility, and they live by famine [1] [2]. Furthermore, the fear effect has a negative impact on the physiological state of young prey, which has a negative impact on their adult survival. All organisms in taxonomy face the risk of predation, and they display a range of anti-predator behaviours, including as adjustments to habitat utilisation, feeding behaviour, alertness, and physiological changes [3] [4] [5]. However, there is clear evidence that predator fear is just as important as direct eating. Fear is induced by such behavioral changes, which can lead to stress-related physiology and have a poor impact on physical health [6] [7] [8], which can affect reproductive health and survival. As a result, predator fear has long-term implications for population dynamics and ecosystems. Fear of predators on prey causes people to spend more time being watchful instead of foraging and looking for better food and reduced habitats [9] [10] [11]. As a result, impact of fear on reproductive success and adult body mass growth are still present. Because people become more attentive when density declines, fear of predators may amplify Allee effects. Individual attentiveness rises when group size diminishes, according to field study on social animals. Predation is causing small groups of prey to become extinct [12]. In population demographics, group defense is a common phrase that describes an occurrence in which predation is reduced as prey’s capacity to protect or hide themselves increases when they are in large numbers or form a group.

The fractional order derivative, unlike the conventional derivative, has an essential trait known as the memory effect. For biological systems, fractional order derivatives are connected to the entire time domain, whereas integer order derivatives reflect a change or a specific attribute at a certain moment [13]. As a result, the fractional order derivative is more suited to simulating memory difficulties in biological systems [14]. Recent study areas including ecological modeling, epidemiology, financial mathematics, and the physical and mathematical sciences have seen an increase in popularity of the fractional order system. The difference between our conventional integer order system and fractional order systems are illustrated in a number of articles on fractional order platforms [15] - [20]. Some ideas have been offered to show our research into various stability criterions such as limit cycle, Hopf bifurcations, global stability, and so on. The effect of fear has been taken up recently in the study prey-predator models [21] [22]. The authors in [23] studied the global asymptotic stability and hopf bifurcation in a homogeneous diffusive predator-prey System with Holling type II functional response. Hopf bifurcation analysis of the repressilator model is proposed in [24]. Our major goal is to combine the fear effect with Caputo fractional order model.

The definitions and features of fractional derivatives that we offer the reader are both informative and practical.

Definition 1. [33] “The Caputo fractional derivative of order 0 < α ≤ 1 for the function u : C n [ 0 , ∞ ] → ℝ is defined as

D C t α ( u ( t ) ) = 1 Γ ( n − α ) ∫ 0 t 1 ( t − z ) α + 1 − n d n d z n u ( z ) d z ,

where C n [ 0 , ∞ ] is a n tines continuously differentiable function and the Gamma function is defined by Γ( ) such that n − 1 < α < n ”.

Theorem 1. [34] “If D C t α u ( t ) is piecewise continuous, then L ( D C t α ( u ( t ) ) ) = z α L ( u ( t ) ) − ∑ i = 0 l − 1 z α − i − 1 u ( i ) ( 0 ) , l − 1 < α ≤ l ∈ ℕ ,where the Laplace transform is denoted by L ( g ( t ) ) ”.

Theorem 2. [35] “One-parametric and two-parametric Mittag-Leffler functions are described as follows: E a 1 ( z ) = ∑ i = 0 ∞ z i Γ ( a 1 i + 1 ) and E a 1 , a 2 ( z ) = ∑ i = 0 ∞ z i Γ ( a 1 i + a 2 ) , where a 1 , a 2 ∈ ℝ + ”.

Lemma 1. [36] “Let 0 < α ≤ 1 , u ( t ) ∈ C [ p , q ] and if D C t α u ( t ) is continuous in [ p , q ] , then u ( x ) = u ( p ) + 1 Γ ( α ) ( x − p ) α ⋅ D C t α u ( z ) , where 0 ≤ z ≤ x , ∀ x ∈ ( p , q ] ”.

Note 1. “If D C t α u ( t ) ≥ 0 ( D C t α u ( t ) ≤ 0 ), t ∈ ( p , q ) , then u ( t ) is a non-decreasing (non-increasing) function for t ∈ [ p , q ] ”.

Lemma 2. “Let us consider the fractional order system as

D C t α ( Y ( t ) ) = Ψ ( Y ) , Y t 0 = ( y t 0 1 , y t 0 2 , ⋯ , y t 0 n ) , y t 0 j , j = 1 , 2 , ⋯ , n ,

with 0 < α < 1 , Y ( t ) = ( y 1 ( t ) , y 2 ( t ) , ⋯ , y n ( t ) ) and Ψ ( Y ) : [ t 0 , ∞ ] → ℝ n × n . For calculate the equilibrium points, we have Ψ ( Y ) = 0 . These equilibrium points are locally asymptotically stable iff each eigenvalue λ j of the Jacobian matrix

J ( Y ) = ∂ ( Ψ 1 , Ψ 2 , ⋯ , Ψ n ) ∂ ( y 1 , y 2 , ⋯ , y n ) calculated at the equilibrium points satisfies | arg ( λ j ) | > α π 2 ”.

Lemma 3. “Assume that u ( t ) ∈ ℝ + is a differentiable function. Then, for any t > 0 , D C t α [ u ( t ) − u * − u * ln u ( t ) u * ] ≤ ( 1 − u * u ( t ) ) D C t α ( u ( t ) ) , u * ∈ ℝ + , ∀ α ∈ ( 0 , 1 ) ”.

Step-1: Particular species never occupy the entire space in nature; they are either prey or predators of other species. The dynamical prey-predator model makes an assumption about the functional response, which is quantified by the amount of prey taken per predator per unit time. Let u ( t ) , v ( t ) and w ( t ) represent the density of the prey population, intermediate predator, and top predator, respectively, at any timet. To formulate the model system, the following assumption is made:

1) In the absence of a predator, the prey population u ( t ) increases logistically at an inherent growth rate of r, with k representing the environment’s carrying capacity.

2) According to Holling Type-I functional response, the intermediate predator v ( t ) consumes the prey species of with the predation rate a 1 . With the predation b 1 , the top predator w ( t ) also consumes the prey according to the law of mass action (Holling Type-I functional response).

3) The intermediate predator’s energy conversion coefficient and natural date rate, respectively are a 2 and m 1 .

4) The energy conversion coefficient of predation and the natural date rate of the top predator, respectively are b 2 and m 2 .

5) γ 1 is the intra-specific predation rate of top predator.

Kar and Ghosh [37] suggested the model system in the following way depending on these assumptions:

d u d t = r u ( 1 − u k ) − a 1 u v − b 1 u w , d v d t = a 2 u v − m 1 v , d w d t = b 2 u w − m 2 w − γ 1 w 2 . (3.1)

Setp-2: We have now changed the model by include a fear element that the intermediate predator causes in the prey. Then, we write down the model system

with the fear factor g ( ρ , v ) = 1 1 + ρ v where ρ represents the fear level as follows:

d u d t = r 1 + ρ v u ( 1 − u k ) − a 1 u v − b 1 u w , d v d t = a 2 u v − m 1 v , d w d t = b 2 u w − m 2 w − γ 1 w 2 . (3.2)

The function g ( ρ , v ) = 1 1 + ρ v , due to fear (felt by prey) of predator, the birth

rate of prey species is reduced. In biological aspects of ρ , v , g ( ρ , v ) , it is appropriate to assume:

g ( 0 , v ) = 1 , g ( ρ , 0 ) = 1 , lim ρ → ∞ g ( ρ , v ) = 0 , lim v → ∞ g ( ρ , v ) = 0 , ∂ g ( ρ , v ) ∂ ρ < 0 , ∂ g ( ρ , v ) ∂ v < 0.

Step-3: We have analyzed the suggested model utilizing the Caputo derivative of order 0 < α ≤ 1 .

D C t α u = r ∨   α 1 + ρ ∨   α v u ( 1 − u k ∨   α ) − a 1 ∨   α u v − b 1 ∨   α u w , D C t α v = a 2 ∨   α u v − m 1 ∨   α v , D C t α w = b 2 ∨   α u w − m 2 ∨   α w − γ 1 ∨   α w 2 . (3.3)

Step-4: Now, we rewrite the parameters as follows for the sake of computation convenience: r ∨   α = r , ρ ∨   α = ρ , k ∨   α = k , a 1 ∨   α = a 1 , b 1 ∨   α = b 1 , a 2 ∨   α = a 2 , m 1 ∨   α = m 1 , b 2 ∨   α = b 2 , m 2 ∨   α = m 2 , γ 1 ∨   α = γ 1 .

With beginning circumstances, the improved model system (3.3) may eventually be stated as follows:

D C t α u = r 1 + ρ v u ( 1 − u k ) − a 1 u v − b 1 u w , D C t α v = a 2 u v − m 1 v , D C t α w = b 2 u w − m 2 w − γ 1 w 2 . (3.4)

This section investigates the existence, uniqueness, non-negativity, and boundedness of the proposed model.

In this part, we’ll identify all of system (3.4)’s trivial and non-trivial equilibrium points, as well as their existence conditions.

The main objective of this section is to investigate the impact of fear all populations. To look at this, we must first separate each component of the fear factor ρ at E * ( u * , v * , w * ) .

Now u * = m 1 a 2 , w * = m 2 γ 1 [ b 2 a 2 m 1 m 2 − 1 ] . Hence u * and w * is positive if b 2 a 2 m 1 m 2 > 1 , and v * is the solution of the equation

R v * 2 + S v * + T = 0 , (6.1)

where R = ρ ( a 1 ) , S = [ a 1 + ρ b 1 { m 2 γ 1 ( b 2 a 2 m 1 m 2 − 1 ) } ] and T = b 1 { m 2 γ 1 ( b 2 a 2 m 1 m 2 − 1 ) } − r ( 1 − m 1 a 2 ⋅ 1 k ) .

Differentiating (6.1) w.r. to ρ , we have d v * d ρ = − v * [ L v * + M ] 2 R v * + S , where

L = a 1 , M = b 1 { m 2 γ 1 ( b 2 a 2 m 1 m 2 − 1 ) }

and v * = 1 2 ρ a 1 ( − a 1 − ρ a 2 b 1 ( b 2 m 1 − a 2 m 2 ) γ 1 + X + Y ) , with

X = − 4 ρ a 1 { − ( 1 − a 2 m 1 k ) r + a 2 b 1 ( b 2 m 1 − a 2 m 2 ) γ 1 } ,

Y = { a 1 + ρ a 2 b 1 ( b 2 m 1 − a 2 m 2 ) γ 1 } 2 .

As a result, the sign of d v * d ρ is always negative, suggesting that as the fear level

ρ rises, v decreases. Differentiating equations (5.2) with respect to fear level ρ , we have

d u * d ρ = 0   and   d w * d ρ = 0.

As a result, we observed that as fear levels rise, the density of intermediate predators decreases.

Theorem 7.1. Let us consider D C t α u = h ( ρ , u ) , where 0 < α < 1 , u ∈ ℝ 3 . A fractional order Hopf bifurcation is proposed in which the model (3.4) undergoes a Hopf bifurcation through E * at the value ρ = ρ * if

● The Jacobian matrix has two complex-conjugate eigenvalues λ 1 , 2 .

● θ 1 , 2 ( α , ρ * ) = 0 .

● ∂ θ 1 , 2 ∂ ρ | ρ = ρ * ≠ 0 , where θ i ( α , ρ * ) = α π 2 − | arg ( λ i ( ρ ) ) | , i = 1 , 2 .

Proof: For ρ = ρ * , the characteristic Equation (5.6) becomes

λ 3 + φ 1 λ 2 + φ 2 λ + φ 1 φ 2 = 0 .

i.e., ( λ 2 + ϕ 2 ) ( λ + φ 1 ) = 0 .

i.e., λ = − ϕ 1 , ± i φ 2 .

Let λ 1 = i φ 2 , λ 2 = − i φ 2 , λ 3 = − φ 1 .

For ρ ∈ ( ρ * − δ , ρ * + δ ) , where δ > 0 , then we have

λ 1 ( ρ ) = σ 1 ( ρ ) + i σ 2 ( k ) ,

λ 2 ( ρ ) = σ 1 ( ρ ) − i σ 2 ( k ) ,

λ 3 ( ρ ) = − φ 1 .

Now the condition ∂ θ 1 , 2 ∂ ρ | ρ = ρ * ≠ 0 at ρ = ρ * , is verify below.

Substitute λ 1 ( ρ ) = σ 1 ( ρ ) + i σ 2 ( ρ ) into the characteristic equation and taking derivative with respect to ρ , we get

A ( ρ ) σ ′ 1 ( ρ ) − B ( ρ ) σ ′ 2 ( ρ ) + S 1 ( ρ ) = 0 ,

B ( ρ ) σ ′ 1 ( ρ ) + A ( ρ ) σ ′ 2 ( ρ ) + S 2 ( ρ ) = 0 .

where

A ( ρ ) = 3 σ 1 2 ( ρ ) + 2 φ 1 ( ρ ) σ 1 ( ρ ) + φ 2 ( ρ ) − 3 σ 2 2 ( ρ ) ,

B ( ρ ) = 6 σ 1 ( ρ ) σ 2 ( ρ ) + 2 φ 1 ( ρ ) σ 2 ( ρ ) ,

S 1 ( ρ ) = σ 1 2 ( ρ ) φ ′ 1 ( ρ ) + φ ′ 2 ( ρ ) σ 1 ( ρ ) + φ ′ 3 ( ρ ) − φ ′ 1 ( ρ ) σ 1 2 ( ρ ) ,

S 2 ( ρ ) = 2 σ 1 ( ρ ) σ 2 ( ρ ) φ ′ 1 ( ρ ) + φ ′ 2 ( ρ ) σ 2 ( ρ ) .

Noticing that, σ 1 ( ρ * ) = 0 , σ 2 ( ρ * ) = φ 2 ( ρ * ) .

We have, A ( ρ * ) = − 2 φ 2 ( ρ * ) ,

B ( ρ * ) = 2 φ 1 ( ρ * ) φ 2 ( ρ * ) ,

S 1 ( ρ ) = φ ′ 3 ( ρ * ) − φ ′ 1 ( ρ * ) φ 2 ( ρ * ) ,

S 2 ( ρ ) = − φ ′ 2 ( ρ * ) φ 2 ( ρ * ) .

Therefore,

d d ρ ( R e ( λ ( ρ ) ) )   at   ρ = ρ * = − B ( ρ * ) S 2 ( ρ * ) + A ( ρ * ) S 1 ( ρ * ) [ A ( ρ * ) ] 2 + [ B ( ρ * ) ] 2 = − 2 φ 1 ( ρ * ) φ ′ 2 ( ρ * ) φ 2 ( ρ * ) + 2 φ ′ 1 ( ρ * ) φ 2 ( ρ * ) φ 2 ( ρ * ) − 2 φ 2 ( ρ * ) φ ′ 3 ( ρ * ) 4 [ φ 2 2 ( ρ * ) + φ 1 2 ( ρ * ) φ 2 ( ρ * ) ] = − φ 1 ( ρ * ) φ ′ 2 ( ρ * ) − φ ′ 3 ( ρ * ) + φ ′ 1 ( ρ * ) φ 2 ( ρ * ) 2 [ φ 2 ( ρ * ) + ( φ 1 ( ρ * ) ) 2 ] ≠ 0.

when, φ 1 ( ρ * ) φ ′ 2 ( ρ * ) − φ ′ 3 ( ρ * ) + φ ′ 1 ( ρ * ) φ 2 ( ρ * ) ≠ 0 , and λ 3 ( ρ * ) = − φ 1 ( ρ * ) ≠ 0 .

To validate the theoretical findings presented in the earlier portions of this study, numerical simulations have been carried out using the modified Predictor-corrector approach [39] [40] in the Matlab framework. The Adams-Bashforth-Moulton scheme is employed for the numerical study.

D C t α F j ( t ) = g j ( t , F j ( t ) ) , F j r ( 0 ) = F j 0 r , (8.1)

r = 0 , 1 , 2 , ⋯ , ⌈ α ⌉ , j ∈ ℕ

where F j 0 r ∈ ℝ , α > 0 and in the Caputo interpretation, D t α is equivalent to the popular Volterra integral equation.

F j ( t ) = ∑ n = 0 ⌈ α ⌉ − 1 F j 0 r t n n ! + 1 Γ ( α ) ∫ 0 t ( t − u ) α − 1 g j ( u , F j ( u ) ) d u , j ∈ ℕ . (8.2)

We established model parameter values for numerical simulation purposes based on information from appropriate journal articles (see Table 1). This section is divided into four parts. The stability of our proposed model is discussed at E 2 , E 3 and E * in Part 1. Part 2 delves into the dynamical behavior of all population of various fractional orders. Part 3 is to explore the Hopf bifurcation of the model system (3.4). Graph of mean density of all population with respect to the variation of α is discussed in Part 4.

Part 1:

The stability of our suggested model is discussed in this section. The parameter values used for the numerical simulations in Part 1 is provided in Table 1. Instead of the trivial E 0 and axial E 1 , we are more concerned in the stability of the coexistence equilibrium E 2 , E 3 and E * . System (3.4) is shown in Figures 1-3 to be asymptotically stable for α = 0.999 .

Part 2:

The parameter values in Table 1 are used to examine the dynamical behaviour

of the entire population. Figures 4(a)-(c) depict all populations’ behavior over time for different fractional orders α. Figure 4(a) depicts that the number of prey population increases when α changes from 0.85 to 0.95. We see in Figure 4(b) that number of intermediate predator population increases when α increases. Figure 4(c) depicts that the number of top predator population increases with time when α changes from 0.95 to 0.8.

Part-3:

In this part, it is explored whether the model system (3.4) exhibits a Hopf bifurcation with fractional order α = 1 . The relevance of Hopf bifurcation is discussed using the following set of parametric variables.

The bifurcation analysis is investigated using the values of the parameters in Table 2. The model system’s unique endemic equilibrium E * ( 2.5 , 6.91 , 6.33 ) is obtained using the parameters in Table 2. Figure 5 shows the Hopf bifurcation diagram of the model system (3.4) with respect to the parameter ρ taking α = 1 . Also Figure 5 depicts that the model remains stable until ρ crosses its threshold value ρ = ρ * = 0.05 , when ρ crosses its threshold value the model become unstable.

Part 4:

Figure 6 shows the change in mean density of prey, intermediate predator and top predator with respect to the variation of α . Figure 6(a) depicts that mean density of prey population increases for 0 ≤ α ≤ 0.1 , decreases for 0.1 < α ≤ 0.8 and oscillates for α > 0.8 . Figure 6(b) shows that mean density of intermediate predator population increases for 0 ≤ α ≤ 0.45 , decreases for 0.45 < α ≤ 0.7 and oscillates for α > 0.7 . Figure 6(c) depicts that mean density of top predator population decreases for 0 ≤ α ≤ 0.8 .

A fractional order prey-predator system with a fear impact on the prey population has been suggested. The non-negativity, boundedness, and uniqueness of the model system’s solutions have been demonstrated and investigated. The equilibrium points have been determined, and the stability of all of the model system’s possible equilibrium points has been studied both analytically and numerically. The system’s local and global stability are both established. Under constrained parametric circumstances, the system is found to be locally and globally asymptotically stable. The system is found to have Hopf bifurcations with respect to the parameter ρ . Local stability and Hopf bifurcation analysis have been the focus of our research. Fractional derivatives and integrals are a difficult idea to convey since they are derived from pure mathematics. The influence of predator anxiety on the population of prey is proposed in this research. A higher order indexing can be associated with weak memory, but a lower order classification can be associated with distant memory since the fractional order and memory are coupled. As a result, our research suggests that weak memories can help to improve the predator-prey system’s ability to cohabit peacefully, but powerful memories can actually make this situation worse. The discretization technique and FDE12 based on Adams-Bashforth-Moulton scheme are used to perform simulation studies. The concept of fractional calculus has nothing to do with any major geometrical meaning, such as function trend or convexity. Finally, we translate our mathematical findings into ecological terms as follows: a large number of prey refuges are produced in the system as a result of the prey species’ profound recollection of the exogenous effects that fear has on their life cycles. However, as we gradually lower the model system’s order, especially in the case of a low amount of predator-induced fear, the dynamics of the model system change away from its unstable behavior and toward stability. Consequently, our extensive mathematical findings show that weak memory might contribute to the stable existence of the predator-prey system whereas excessive memory of the species degrades the stability of the model system. In the suggested model (3.4), there is more work to be done, such as determining the species’ maturity to release harmful compounds into the environment. The model is now more realistic and intriguing as a result of these changes. We will leave this for further study.

The authors declare no conflicts of interest regarding the publication of this paper.

Paul, S., Mahata, A., Mukherjee, S., Mali, P.C. and Roy, B. (2022) Study of Fractional Order Tri-Tropic Prey-Predator Model with Fear Effect on Prey Population. Advances in Pure Mathematics, 12, 652-675. https://doi.org/10.4236/apm.2022.1211050