Chaos, Solitons and Fractals 191 (2025) 115790

Contents lists available at ScienceDirect

Chaos, Solitons and Fractals

journal homepage: www.elsevier.com/locate/chaos

Evolution of Turing patterns of a predator–prey system with variable

carrying capacity and harvesting
Lakpa Thendup Bhutia, Samir Biswas ∗, Esita Das, Tapan Kumar Kar, Bidhan Bhunia
Department of Mathematics, Indian Institute of Engineering Science and Technology, Shibpur Howrah 711103, West Bengal, India

ARTICLE INFO ABSTRACT

Keywords: In the present study, we propose a modified form of the Rosenzweig–MacArthur model by incorporating
Variable carrying capacity prey harvesting and variable carrying capacity. These modifications are addressed both from a mathematical
Enrichment and ecological perspective. By considering harvesting effort as a bifurcating parameter, sufficient criteria are
Hopf bifurcation
identified for the stability and occurrence of Hopf bifurcation. We also observe that the coexisting equilibrium
Turing patterns
stabilizes as harvesting effort is increased. On the contrary, increasing the influence parameter tends to
Spatio-temporal chaos
destabilize the equilibrium point via limit cycles. As species live in both space and time, we extend our study by
incorporating spatial dynamics into the system. The conditions for Turing instability are established, and it is
demonstrated that the system cannot produce heterogeneous spatial patterns without cross-diffusion. Extensive
numerical simulations are conducted, and various Turing patterns are observed as harvesting effort, influence
parameter, and cross-diffusion coefficient are varied. Spatio-temporal chaotic patterns are also detected by
choosing appropriate parametric values from the bifurcation plane. As a result, the population’s oscillatory
behavior is desynchronized, potentially lowering the risk of extinction.

1. Introduction 𝑉 and 𝑈 are the predator and prey density, respectively. 𝛼 is the preda-
tion coefficient, 𝛽 is the food conversion efficiency and 𝛿 is the natural
Ecologists have been able to better comprehend the intricate dy- death rate of predator species. Detailed analysis of the above system
namics of an ecosystem by developing and investigating various math- is presented in [11]. Kar [12] examined the above system with refuge
ematical models [1–8]. How species interact and coexist sustainably has among the prey species and ascertained the presence and dependence
been the main focus of their study. Verhulst was the first to formulate of the system on the refuge parameter along with the existence of limit
population dynamics using a mathematical model, cycles. Peng et al. [13] investigated the effect of constant harvesting
( ) in the above model and obtained numerous bifurcations. Barman and
𝑑𝑈 𝑈
= 𝑟𝑈 1 − , (1)
𝑑𝑡 𝐾 Ghosh [14] studied the effect of time delay in the above prey-predator
and the equation is referred to as the logistic equation [9]. 𝑈 is the system. Their findings show that time delay usually destabilizes the
biomass density, 𝑟 is the intrinsic growth rate of the species, and 𝐾 system via Hopf bifurcation. The integer-order system (2) has also
is called the environment’s carrying capacity. Although often criti- been modified to a fractional-order system to take into account the
cized for oversimplicity, it remains a central theorem for constructing memory effect. Rihan et al. [15] investigated the effect of time delay in
a single-species growth model. Many researchers have extended the a fractional-order Rosenzweig–MacArthur model. The system’s stability
above system to study the interaction between prey and predator improved considerably with the introduction of fractional order. Javidi
species. Rosenzweig and MacArthur [10] formulated a simple two- and Nyamoradi [16] and Moustafa et al. [17] also investigated the
species model where they used Holling type II functional response to impact of harvesting and prey refuge in a fractional-order system,
employ the interaction among species. When the predators are absent, respectively. Recently, Wyse et al. [18] performed structural sensitivity
the prey species has a logistic growth, and the model is given as, of the Rosenzweig-MacArthur model by considering various functional
( )
𝑑𝑈 𝑈 𝛼𝑈 𝑉 responses. They observed that introducing stochasticity and fitting
= 𝑟𝑈 1 − − ,
𝑑𝑡 𝐾 𝑎+𝑈 functional response over a steady prey density state tends to decrease
(2)
𝑑𝑉 𝛼 𝛽 𝑈 𝑉 structural sensitivity. The majority of the studies described above solely
= − 𝛿𝑉 .
𝑑𝑡 𝑎+𝑈

∗ Corresponding author.
E-mail addresses: [email protected] (L.T. Bhutia), [email protected] (S. Biswas), [email protected] (E. Das), [email protected] (T.K. Kar),
[email protected] (B. Bhunia).

https://doi.org/10.1016/j.chaos.2024.115790
Received 6 April 2024; Received in revised form 30 August 2024; Accepted 17 November 2024
Available online 5 December 2024
0960-0779/© 2024 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

looked at the system’s temporal aspects. However, in recent years, the prioritized scenarios in which the solution exhibited severe occurrences
significance of the spatial effect has drawn a lot of attention. such as finite-time death and finite-time singularity. Their proposed
Due to constant environmental fluctuations, competition, predation, system has recently been expanded to a two-species model by Pati and
limited availability of resources, etc., the spatial distribution of the Ghosh [42] using linear functional response. They have quantitatively
species in their habitat is very heterogeneous. Many researchers have demonstrated the emergence of delay-induced degenerate dynamics
studied the reaction–diffusion system to incorporate and analyze such and the existence of supercritical and subcritical Hopf bifurcations.
spatial distribution of species. Various patterns have been reported in With recent developments, researchers have been able to model and
ecological systems with the aid of mathematical tools and simulations. analyze more realistic systems by including various ecological phenom-
Yang et al. [19] have investigated the impact of cross-diffusion in ena. However, the notion of constant environmental carrying capacity
a predator–prey system with herd behavior. They concluded that a still needs to be adequately studied. The concept of variable carrying
larger cross-diffusion coefficient is beneficial for the survival of both capacity due to environmental fluctuations still needs to be appropri-
species. Using weak non-linear analysis, Zhang et al. [20] determined ately incorporated. Ganguli et al. [35] explored a prey-predator system
the stability of various patterns in a Beddington–DeAngelis model. where the carrying capacity varies with time. They have conducted a
Wang et al. [21] analyzed the spatial effect on the transient behavior comparative study of constant and variable carrying capacity. Recently,
of the reaction–diffusion system. From their rigorous calculation, they Ang and Safuan [34] discussed a harvested intraguild system where
observed that for a two-species model, random dispersal had no effect both the species’ logistic carrying capacity varied based on the shared
on the reactivity of the system. Ning et al. [22] investigated the biotic resource. Hopf bifurcation is obtained by taking harvesting as a
spatial dynamics of a predator–prey system with the Allee effect on bifurcating parameter. These studies have taken the carrying capacity
predator species. They obtained multiple Turing patterns under varying as a function of some shared biotic resources. However, the carrying
intensities of the Allee effect. Abid et al. [23] studied a modified form of capacity may depend on the species itself. As discussed by Yukalov
the Leslie–Gower model with cross-diffusion. Using bifurcation theory, et al. [40,41], the carrying capacity of the human species is influenced
they proved the existence of Hopf and Turing bifurcation. Ghorai by human activity. With the development in technology, we have been
and Poria [24] investigated the impact of habitat complexity in a able to increase the environmental carrying capacity to a great extent.
predator–prey model with Holling type II functional response. Recently, However, at the same time, due to deforestation, climate change, and
reaction–diffusion systems have also been used to understand vege- habitat destruction, human species have led to a massive reduction
tation patterns [25–27], which serve as a key indicator of ecosystem of ecosystems and eradication of their carrying capacity. They have
robustness. modeled this constructive and destructive phenomenon by considering
The management of fisheries, forests, and wildlife frequently in- the carrying capacity to be a linear function of species density. Pati and
volves exploiting biological resources and harvesting populations. A Ghosh [42] also incorporated this concept of variable carrying capacity
broad spectrum of researchers is interested in using bioeconomic mod- in their study.
eling to understand the scientific management of renewable resources These investigations done by Yukalov et al. [40,41] and Pati and
for humanity’s long-term benefit. Guin et al. [28] studied a reaction– Ghosh [42] on such system each concentrated solely on the delay
diffusion system with linear prey harvesting and ratio-dependent func- parameter. However, we speculate that these constructive and destruc-
tional response. By varying prey refuge and harvesting parameters, tive aspects are more prevalent ecologically, which the above studies
complex patterns are simulated in a bounded domain. By taking har- have not adequately discussed. As species exist in time and space, it
vesting as a bifurcating parameter, Zhang et al. [29] analyzed a is essential to study their spatial distribution. Hence, we extend the
Michaelis–Menten prey-predator system. Upadhyay et al. [30] investi- above concept by incorporating a spatio-temporal model. To our best
gated a modified Leslie–Gower system with Holling type IV functional knowledge, such variable carrying capacity has yet to be discussed in
response. Along with determining the direction of Hopf bifurcation, the presence of spatial heterogeneity. We shall also consider harvesting
they used numerical simulation to understand the development of vari- of prey species since harvesting the lowest trophic level impacts the
ous Turing patterns due to the presence of non-linear harvesting. Thus, overall system most [43,44]. Moreover, harvesting is an external factor,
the effect of harvesting in a diffusive system has recently drawn a lot of so we have much freedom by regulating it as a control rather than other
attention from researchers. Motivated by this, in this work, a reaction– intrinsic parameters. Therefore, the primary focus of our study will be
diffusion system with prey harvesting is taken into consideration, and to investigate the joint impact of harvesting on a predator–prey system
our analysis places more focus on the variable carrying capacity, an with variable carrying capacity.
important ecological parameter. The paper is organized as follows. With the above motivations in
Species cannot grow unboundedly as the resources available to them mind, in Section 2, the model is formulated. Stability and bifurcation
are limited. Logistic equation incorporates this limited growth in a nat- analysis of the temporal part are conducted in Section 3. In Section 4,
ural way [3]. Carrying capacity (𝐾) is the maximum number of species the corresponding spatio-temporal model is formulated and studied,
that can sustain in an environment. From the above equation, it is clear followed by numerical analysis in Section 5. Finally, in Section 6 main
that the per capita growth rate falls to zero at the carrying capacity. results are discussed and compared with other relevant studies.
However, most of the aforementioned studies [13,14] have considered
the carrying capacity to be constant, but the actual environment is 2. Model formulation
constantly subjected to change. Climate change, seasonal harvesting,
We consider the following model,
and migration are a few of the phenomena that lead to the ecosystem ( )
being frequently disturbed. The carrying capacity may, therefore, not 𝑑 𝑈 (𝑡) 𝑈 (𝑡) 𝛼 𝑈 (𝑡)𝑉 (𝑡)
=𝑟𝑈 (𝑡) 1 − − − 𝐸 𝑈 (𝑡),
be constant and may change over time. Numerous carrying capaci- 𝑑𝑡 𝐾 + 𝑘1 𝑈 (𝑡) 𝑎 + 𝑈 (𝑡)
(3)
ties have been suggested to accommodate such transformation. Time 𝑑 𝑉 (𝑡) 𝛼 𝛽 𝑈 (𝑡)𝑉 (𝑡)
= − 𝛿 𝑉 (𝑡),
periodic carrying capacity [31,32], carrying capacity of prey species 𝑑𝑡 𝑎 + 𝑈 (𝑡)
depending on the predator species and vice versa [33–35], and sigmoid- where 𝑈 (𝑡) and 𝑉 (𝑡) are the prey and predator population densities at
shaped logistic carrying capacity [36]. The carrying capacity has also time 𝑡, respectively. 𝑟 is the intrinsic growth rate of the prey, and 𝛼 is
been treated as a state variable [37,38] and incorporated into the the predation coefficient of prey by the predator. 𝑎 is the half saturation
system. Recently, Al-Moqbali et al. [39] extended this concept to form a point and 𝛽 is the conversion efficiency of the predator. 𝛿 is the natural
three-state variable system, where the carrying capacity itself followed death rate of predator. Here we have assumed that only prey species
a logistic growth. A delayed carrying capacity was proposed by Yukalov are subjected to harvesting effort 𝐸. The harvesting is based on the
et al. [40,41]. Their work primarily focuses on single species. They CPUE (catch-per-unit-effort ) hypothesis with the catch function is 𝑞 𝐸 𝑈 ,

2
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

where 𝑞 is the catchability coefficient of prey species. For simplicity, (1 + 𝛿 − 𝐸)2

0 < 𝑤(𝑢(𝑡), 𝑣(𝑡)) < (1 − 𝑒−𝛿𝑡 ) + 𝑤(𝑢(0), 𝑣(0))𝑒−𝛿 𝑡 ,
we consider 𝑞 to be unity. 4𝛿
The carrying capacity is taken to be a linear function of prey species 2
density. 𝐾 is the natural carrying capacity and 𝑘1 𝑈 is the capacity and 𝑡 → ∞ gives 0 < 𝑤 < (1+𝛿−𝐸)
4𝛿
. So, we have all the solutions of (4)
induced by the prey species (𝑈 ). Here, 𝑘1 quantifies the species’ impact that start in R2+ are confined in region 𝐵2 , where, 𝐵2 = {(𝑢, 𝑣) ∈ R2+ ∶
(1+𝛿−𝐸)2
on its carrying capacity. If 𝑘1 > 0, the impact is constructive. In such 𝑤= 4𝛿
+ 𝜖 for 𝜖 > 0} provided the given condition holds. □
cases, the species work together to ameliorate their carrying capacity.
If 𝑘1 < 0, the impact is destructive. When 𝑘1 = 0, the system reduces to
the Rosenzwig–MacAurthur model (2). 3. Temporal model
To reduce the number of parameters involved, we non-
dimensionalize the above system as follows: In this section, we first discuss the existence and stability of equi-
𝑈 𝑎 𝐸 𝛼𝑉 𝛿 𝛼𝛽 librium points of the system (4), followed by stability and bifurcation
𝑢 = , 𝑡̃ = 𝑟𝑡, 𝑎̃ = , 𝐸̃ = , 𝑣 = , 𝛿̃ = , 𝛽̃ = , analyses to understand the intricate dynamics of the system.
𝐾 𝐾 𝑟 𝑟𝐾 𝑟 𝑟
and removing the tildes, we get,
( ) 3.1. Existence and stability of equilibrium points
𝑑 𝑢(𝑡) 𝑢(𝑡) 𝑢(𝑡)𝑣(𝑡)
=𝑢(𝑡) 1 − − − 𝐸 𝑢(𝑡) ≡ 𝑓1 (𝑢, 𝑣),
𝑑𝑡 1 + 𝑘1 𝑢(𝑡) 𝑎 + 𝑢(𝑡)
(4) The model (4) possesses the following equilibrium points.
𝑑 𝑣(𝑡) 𝛽 𝑢(𝑡)𝑣(𝑡)
= − 𝛿 𝑣(𝑡) ≡ 𝑓2 (𝑢, 𝑣), (i) The trivial point 𝑃0 (0, 0).
𝑑𝑡 𝑎 + 𝑢(𝑡)
̄ 0), where 𝑢̄ = 1−𝑘1−𝐸
(ii) The predator-free equilibrium 𝑃1 (𝑢, (1−𝐸)
. If
with the initial conditions 𝑢(0) ≥ 0 and 𝑣(0) ≥ 0. All the parameters are 1

positive except 𝑘1 ∈ R. 𝑘1 > 0, the equilibrium exists provided, 1 − 𝑘1 < 𝐸 < 1. If 𝑘1 < 0, the
(1 )
equilibrium points is feasible provided 𝐸 ∉ 1, 1 − 𝑘1 .
1
Theorem 2.1. All the solutions of Eq. (4) with the given non-negative 𝑎𝛿
(iii) The interior equilibrium point 𝑃 ∗ (𝑢∗ , 𝑣∗ ), where 𝑢∗ = 𝛽−𝛿 and
initial conditions is non-negative ∀ 𝑡 > 0. Furthermore, the solutions are [ ]
∗ 𝑎𝛽 𝑎𝛿
also uniformly bounded provided 1 + 𝑘1 𝐸 ≥ 𝑘1 (1 + 𝛿). 𝑣 = 𝛽−𝛿 1 − 𝐸 − 𝛽−𝛿+𝑘 𝑎𝛿 . 𝛽 − 𝛿 > 0 is ecologically relevant. Let
1
𝛽−𝛿
𝑘1 > 1 − than the interior point 𝑃 ∗ is feasible provided,
𝑎𝛿
,
Proof. It is easy to verify that the solutions of the above system 𝑎𝛿
0≤𝐸 <1− = 𝐸 𝑚𝑎𝑥 . (5)
with given initial conditions are always non-negative. For uniform 𝛽 − 𝛿 + 𝑘1 𝑎𝛿
boundness, we consider the following.
(i) Case I: Let 𝑘1 > 0.
Let us assume 𝑤(𝑡) = 𝑢(𝑡) + 𝛽1 𝑣(𝑡). Differentiating we get, Remark 1. We note that the dimensionless prey equilibrium point 𝑢∗ is
𝑑 𝑤(𝑡) 𝑑 𝑢(𝑡) 1 𝑑 𝑣(𝑡) independent of parameters 𝐸 and 𝑘1 , while the dimensionless predator
= + ,
𝑑𝑡 𝑑𝑡 𝛽 𝑑𝑡 equilibrium point is a decreasing function of 𝐸 and is positive under
( )
𝑑 𝑤(𝑡) 𝑢(𝑡) 𝑢(𝑡)𝑣(𝑡) the condition (5). Moreover, it is also an increasing function of 𝑘1 but
=𝑢(𝑡) 1 − − − 𝐸 𝑢(𝑡)
𝑑𝑡 1 + 𝑘1 𝑢(𝑡) 𝑎 + 𝑢(𝑡) gets saturated in the limiting condition of 𝑘1 .
𝑢(𝑡)𝑣(𝑡) 𝛿 The Jacobian matrix of system (4) evaluated at the equilibria is
+ − 𝑣(𝑡),
𝑎 + 𝑢(𝑡) 𝛽 given by,
( )
𝑑 𝑤(𝑡) 𝑢(𝑡) ⎛1 − 2𝑢 𝑘1 𝑢2 𝑣 𝑢𝑣 𝑢 ⎞
+ 𝛿 𝑤(𝑡) =𝑢(𝑡) 1 + 𝛿 − 𝐸 − . + − 𝑎+𝑢 + −𝐸 − 𝑎+𝑢
𝑑𝑡 1 + 𝑘1 𝑢(𝑡) J0 = ⎜ 1+𝑘1 𝑢 (1+𝑘1 𝑢)2 (𝑎+𝑢)2 ⎟. (6)
( ) ⎜ 𝑎𝛽 𝑣 𝛽𝑢
− 𝛿 ⎟
Let 𝜓(𝑢) = 𝑢(𝑡) 1 + 𝛿 − 𝐸 − 1+𝑘𝑢(𝑡)𝑢(𝑡) . The maximum of this function is ⎝ (𝑎+𝑢)2 𝑎+𝑢 ⎠
1
attained at,
√
2 − 𝑘1 (1 + 𝛿 − 𝐸) − 2 1 − 𝑘1 (1 + 𝛿 − 𝐸) Theorem 3.1. For the system (4),
𝜓 𝑚𝑎𝑥 = .
𝑘21 (i) The trivial equilibrium point 𝑃0 (0, 0) is locally asymptotically stable
provided 𝐸 > 1.
So, we have,
𝑑 𝑤(𝑡) (ii) The predator-free equilibrium point 𝑃1 (𝑢,
̄ 0) is locally asymptotically
+ 𝛿 𝑤(𝑡) ≤ 𝜓 𝑚𝑎𝑥 . stable if 𝐸 𝑚𝑎𝑥 < 𝐸 else it is a saddle.
𝑑𝑡
Applying the theory of differential inequality [12,45], we obtain,
𝜓 𝑚𝑎𝑥 Proof. (i) At 𝑃0 (0, 0), the characteristic equation of Eq. (6) gives the
0 < 𝑤(𝑢(𝑡), 𝑣(𝑡)) < (1 − 𝑒−𝛿𝑡 ) + 𝑤(𝑢(0), 𝑣(0))𝑒−𝛿 𝑡 ,
𝛿 eigenvalues 𝜆1 = 1 − 𝐸 and 𝜆2 = −𝛿. So if 𝜆1 < 0, then the trivial
𝑚𝑎𝑥
and 𝑡 → ∞ gives 0 < 𝑤 < 𝜓 𝛿 . So, we have all the solutions of (4) that equilibrium point is locally asymptotically stable else it is s saddle.
start in R2+ are confined in region 𝐵1 , where, 𝐵1 = {(𝑢, 𝑣) ∈ R2+ ∶ 𝑤 = (ii) At 𝑃1 (𝑢,
̄ 0), the characteristic equation of Eq. (6) gives the eigenval-
𝛽 𝑢̄
𝜓 𝑚𝑎𝑥
+ 𝜖 for 𝜖 > 0} provided the given condition holds. ues 𝜆1 = − (1+𝑘𝑢̄ 𝑢)
̄2
(< 0) and 𝜆2 = 𝑎+ 𝑢̄
− 𝛿. So if 𝐸 𝑚𝑎𝑥 < 𝐸, then 𝜆2 < 0
𝛿 1
(ii) Case II: Let 𝑘1 < 0 and proceeding in similar manner as before we and hence 𝑃1 is locally asymptotically stable. If 𝐸 𝑚𝑎𝑥 > 𝐸 then 𝜆2 > 0
get, and 𝑃1 becomes a saddle point. This completes the proof. □
( )
𝑑 𝑤(𝑡) 𝑢(𝑡) 𝛿
=𝑢(𝑡) 1 − − 𝐸 𝑢(𝑡) − 𝑣(𝑡),
𝑑𝑡 1 + 𝑘1 𝑢(𝑡) 𝛽
𝑑 𝑤(𝑡) 𝛿 Remark 2. So when the predator-free equilibrium 𝑃1 (𝑢, ̄ 0) is stable, the
≤𝑢(𝑡)(1 − 𝑢(𝑡)) − 𝐸 𝑢(𝑡) − 𝑣(𝑡), interior equilibrium 𝑃 ∗ (𝑢∗ , 𝑣∗ ) does not exist. However, if it exists, then
𝑑𝑡 𝛽
𝑑 𝑤(𝑡) the former is a saddle point. This indicates that the system undergoes
+ 𝛿 𝑤(𝑡) ≤𝑢(𝑡)(1 + 𝛿 − 𝐸 − 𝑢(𝑡)), a transcritical bifurcation at 𝐸 = 𝐸 𝑚𝑎𝑥 .
𝑑𝑡
𝑑 𝑤(𝑡) (1 + 𝛿 − 𝐸)2
+ 𝛿 𝑤(𝑡) ≤ .
𝑑𝑡 4 Theorem 3.2. The system (4) undergoes a transcritical bifurcation around
Similarly, applying the theory of differential inequality [12,45], we ̄ 0) at 𝐸 𝑇 𝐶 = 𝐸 𝑚𝑎𝑥 .
𝑃1 (𝑢,
obtain,

3
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

Proof. The Jacobian matrix of system (4) at 𝑃1 is, then 𝑡𝑟(J0 ) > 0 and if 𝐸 ∈ (𝐸 𝐻 𝑜𝑝𝑓 , 𝐸 𝑚𝑎𝑥 ) then 𝑡𝑟(J0 ) < 0 and 𝑡𝑟(J0 ) = 0
( )
− (1+𝑘𝑢̄ 𝑢) − 𝑎+𝑢̄ 𝑢̄ at = 𝐸 𝐻 𝑜𝑝𝑓 . Thus 𝑃 ∗ is unstable when 𝐸 < 𝐸 𝐻 𝑜𝑝𝑓 and stable when
̄2
J0 (𝑃1 ) = 1
𝛽 𝑢̄ . 𝐸 > 𝐸 𝐻 𝑜𝑝𝑓 . Moreover,
0 −𝛿
𝑑 𝑅𝑒(𝜆) || 1 𝑑 𝑡𝑟(J0 ) ||
𝑎+𝑢̄
𝛽−𝛿
Clearly at 𝐸 𝑇 𝐶 = 𝐸 𝑚𝑎𝑥 , one of the eigenvalue of J0 (𝑃1 ) is negative | = | =− ≠ 0,
𝑑 𝐸 || 𝐻 𝑜𝑝𝑓 2 𝑑 𝐸 || 𝐻 𝑜𝑝𝑓 2𝑎𝛽
𝐸=𝐸 𝐸=𝐸
and the other one is zero. Let 𝜉 = [𝜉1 , 𝜉2 ]𝑇 and 𝜁 = [𝜁1 , 𝜁2 ]𝑇 be the
eigenvectors corresponding to the zero eigenvalue of matrix J0 (𝑃1 ) and where 𝜆 is the eigenvalue of the Jacobian matrix at 𝑃 ∗ (𝑢∗ , 𝑣∗ ). There-
[ ]𝑇 fore, a Hopf bifurcation occurs at 𝐸 = 𝐸 𝐻 𝑜𝑝𝑓 and the proof is
𝑎+𝑢̄
J0 (𝑃1 )𝑇 , respectively. Then, 𝜉 = 1, − (1+𝑘 and 𝜁 = [0, 1]𝑇 .
̄2
𝑢) 1 complete. □
Let 𝐹 = [𝑓1 , 𝑓2 ]𝑇 then we get the following conditions,
(𝑖) 𝜁 𝑇 𝐹𝐸 (𝑃1 , 𝐸 = 𝐸 𝑚𝑎𝑥 ) = 0, 3.2.1. Stability of limit cycle
(𝑖𝑖) 𝜁 𝑇 [𝐷𝐹𝐸 (𝑃1 , 𝐸 = 𝐸 𝑚𝑎𝑥 )𝜉] = 0, Here we will discuss the stability of the limit cycle by computing the
2𝑎𝛽 first Lyapunov number [46,49] at the equilibrium point 𝑃 ∗ (𝑢∗ , 𝑣∗ ) of the
(𝑖𝑖𝑖) 𝜁 𝑇 [𝐷2 𝐹 (𝑃1 , 𝐸 = 𝐸 𝑚𝑎𝑥 )(𝜉 , 𝜉)] = − ≠ 0.
(𝑎 + 𝑢)(1
̄ + 𝑘1 𝑢)
̄ system (4). In order to do so, we make the equilibrium point 𝑃 ∗ (𝑢∗ , 𝑣∗ )
the origin by using the transformation, 𝑢̂ = 𝑢 − 𝑢∗ and 𝑣̂ = 𝑣 − 𝑣∗ .
Hence, all the conditions of Sotomayer’s theorem are satisfied (cf. [46]),
Expanding in Taylor series, we get,
except for the second condition, which should be nonzero for non-
degenerate transcritical bifurcation. As a result, the system experiences 𝑢̂ ′ =𝑚10 𝑢̂ + 𝑚01 𝑣̂ + 𝑚20 𝑢̂ 2 + 𝑚11 𝑢̂ 𝑣̂ + 𝑚02 𝑣̂ 2 + 𝑚30 𝑢̂ 3 + 𝑚21 𝑢̂ 2 𝑣̂ + 𝑚12 𝑢̂ 𝑣̂ 2
a degenerate transcritical bifurcation [47,48] around 𝑃1 (𝑢,̄ 0) at 𝐸 𝑇 𝐶 = + 𝑚03 𝑣̂ 3 + 𝑂(𝑢̂ 4 , 𝑣̂ 4 ),
(9)
𝐸 𝑚𝑎𝑥 . □ 𝑣̂ ′ =𝑛10 𝑢̂ + 𝑛01 𝑣̂ + 𝑛20 𝑢̂ 2 + 𝑛11 𝑢̂ 𝑣̂ + 𝑛02 𝑣̂ 2 + 𝑛30 𝑢̂ 3 + 𝑛21 𝑢̂ 2 𝑣̂ + 𝑛12 𝑢̂ 𝑣̂ 2
From an ecological point of view, it is interesting to study the + 𝑛03 𝑣̂ 3 + 𝑂(𝑢̂ 4 , 𝑣̂ 4 ),
dynamical behavior of the interior equilibrium point. Hence we will
consider 𝐸 ∈ [0, 𝐸 𝑚𝑎𝑥 ) for the remaining part of the paper. Corre- where 𝑚10 , 𝑚01 , 𝑛10 and 𝑛01 are the coefficient of the Jacobian matrix
𝜕 𝑖+𝑗 𝑓 (𝑢,𝑣) 𝜕 𝑖+𝑗 𝑓 (𝑢,𝑣)
spondingly, the Jacobian matrix at the interior point 𝑃 ∗ (𝑢∗ , 𝑣∗ ) is given (7) and 𝑚𝑖𝑗 = 𝜕𝑖 𝑢𝜕1 𝑗 𝑣 and 𝑛𝑖𝑗 = 𝜕 𝑖 𝑢𝜕2 𝑗 𝑣 for 𝑖, 𝑗 = 0, 1, 2, 3 evaluated
by, at 𝑃 ∗ (𝑢∗ , 𝑣∗ ) and 𝐸 = 𝐸 𝐻 𝑜𝑝𝑓 . Hence from Eq. (4) we get the coefficients
( ) as,
𝑎11 𝑎12
J0 (𝑃 ∗ ) = , (7) 𝑢∗ 𝑢 ∗ 𝑣∗ 𝑢∗
𝑎21 𝑎22 𝑚10 = − + , 𝑚01 = − ,
(1 + 𝑘1 𝑢∗ )2 (𝑎 + 𝑢∗ )2 𝑎 + 𝑢∗
∗ ∗ ∗ ∗ 𝑎𝛽 𝑣∗
𝑢 𝑣
where 𝑎11 = − (1+𝑘𝑢 𝑢∗ )2 + (𝑎+𝑢 𝑢
∗ )2 , 𝑎12 = − 𝑎+𝑢∗ , 𝑎21 = (𝑎+𝑢∗ )2
, and 𝑎22 = 0. 1 𝑎𝑣∗
1 𝑚20 = − + ,
Then we have, (1 + 𝑘1 𝑢∗ )3 (𝑎 + 𝑢∗ )3
𝑢∗ 𝑢 ∗ 𝑣∗ 𝑎 𝑘1 𝑎𝑣∗
𝑡𝑟(J0 ) =𝑎11 + 𝑎22 = − + , 𝑚11 = − , 𝑚30 = − ,
(1 + 𝑘1 𝑢 )∗ 2 (𝑎 + 𝑢∗ )2 (𝑎 + 𝑢∗ )2 (1 + 𝑘1 𝑢∗ )4 (𝑎 + 𝑢∗ )4
(8)
𝑎𝛽 𝑢∗ 𝑣∗ 𝑚21 =
𝑎
, 𝑚02 = 𝑚12 = 𝑚03 = 0,
𝑑 𝑒𝑡(J0 ) =𝑎11 𝑎22 − 𝑎12 𝑎21 = (> 0).
(𝑎 + 𝑢∗ )3 (𝑎 + 𝑢∗ )3
and,
𝑎𝛽 𝑣∗ 𝑎𝛽 𝑣∗ 𝑎𝛽
3.2. The stabilizing effect of 𝐸 𝑛10 = , 𝑛20 = − , 𝑛11 = ,
(𝑎 + 𝑢∗ )2 (𝑎 + 𝑢∗ )3 (𝑎 + 𝑢∗ )2
𝑎𝛽 𝑣 ∗ 𝑎𝛽
In this subsection, we investigate the effect of parameter 𝐸 on 𝑛30 = , 𝑛21 = − , 𝑛01 = 𝑛02 = 𝑛12 = 𝑛03 = 0.
(𝑎 + 𝑢∗ )4 (𝑎 + 𝑢∗ )3
the stability of interior equilibrium. We proceed with the following ∗ ∗
𝑎𝛽 𝑢 𝑣
theorem. Let 𝛥1 = 𝑑 𝑒𝑡(J0 ) = (𝑎+𝑢 ∗ )3 . Hence, the first Lyapunov number 𝛤 is given
by,
Theorem 3.3. Let 𝐸 ∈ [0, 𝐸 𝑚𝑎𝑥 ). 3𝜋
𝛤 =− {[𝑚10 𝑛10 (𝑚211 + 𝑚11 𝑛02 + 𝑚02 𝑛11 )
𝑎𝛽(𝛽−𝛿) ∗ ∗ ∗ 3∕2
(i) If 𝐸 𝑚𝑎𝑥 < (𝛽−𝛿+𝑘 2 , then 𝑃 (𝑢 , 𝑣 ) is locally asymptotically stable 2𝑚01 𝛥1
1 𝑎𝛿)
whenever it exists. + 𝑚10 𝑚01 (𝑛211 + 𝑚20 𝑛11 + 𝑚11 𝑛02 )
𝑎𝛽(𝛽−𝛿)
(ii) If 𝐸 𝑚𝑎𝑥 > (𝛽−𝛿+𝑘 , then ∃ 𝐸 𝐻 𝑜𝑝𝑓 such that 0 < 𝐸 𝐻 𝑜𝑝𝑓 < 𝐸 𝑚𝑎𝑥 ,
𝑎𝛿)2 1 + 𝑛210 (𝑚11 𝑚02 + 2𝑚02 𝑛02 )
and the system undergoes Hopf bifurcation at 𝐸 = 𝐸 𝐻 𝑜𝑝𝑓 , where 𝐸 𝐻 𝑜𝑝𝑓 =
[ ] − 2𝑚10 𝑛10 (𝑛202 − 𝑚20 𝑚02 ) − 2𝑚10 𝑚01 (𝑚220 − 𝑛20 𝑛02 )
𝑎(𝛽−𝛿) 𝛿 𝛽
1 − 𝛽−𝛿+𝑘 + 𝛽−𝛿+𝑘 .
𝑎𝛿 𝛽−𝛿
1 𝑎𝛿 1 − 𝑚201 (2𝑚20 𝑛20 + 𝑛11 𝑛20 ) + (𝑚01 𝑛10 − 2𝑚210 )
× (𝑛11 𝑛02 − 𝑚11 𝑚20 )]
𝑎𝛿 ∗
Proof. (𝑖) Here 𝐸 𝑚𝑎𝑥 = 1 − 𝛽−𝛿+𝑘1 𝑎𝛿
= 1 − 1+𝑘𝑢 𝑢∗ . Then we can write, − (𝑚210 + 𝑚01 𝑛10 )[3(𝑛10 𝑛03 − 𝑚01 𝑚30 ) + 2𝑚10 (𝑚21 + 𝑛12 )
∗ 1
𝑢∗
𝑡𝑟(J0 ) = − (1+𝑘𝑢 𝑢∗ )2 + 𝑎+𝑢∗
(𝐸 𝑚𝑎𝑥 − 𝐸). Thus 𝑃 ∗ is locally stable if, + (𝑛10 𝑚12 − 𝑚01 𝑛21 )]}.
1
𝑡𝑟(J0 ) < 0,
Due to the complexity of the above expression, we cannot determine
𝑎𝛽(𝛽 − 𝛿)
⟹ 𝐸 𝑚𝑎𝑥 − 𝐸 < , the sign of 𝛤 directly. Hence the nature of 𝛤 will be determined by
(𝛽 − 𝛿 + 𝑘1 𝑎𝛿)2 numerical example. If 𝛤 < 0, the Hopf bifurcation is supercritical, while
𝑎𝛽(𝛽 − 𝛿) it is subcritical if 𝛤 > 0.
⟹ 𝐸 𝑚𝑎𝑥 − < 𝐸.
(𝛽 − 𝛿 + 𝑘1 𝑎𝛿)2 As an example, consider 𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, and 𝑘1 = 0.7.
So, from Eq. (5) and from the given condition 𝐸 𝑚𝑎𝑥 < 𝑎𝛽(𝛽−𝛿)
we For this parameter set, we get 𝐸 𝑚𝑎𝑥 = 0.7608 and 𝐸 𝐻 𝑜𝑝𝑓 = 0.2982.
(𝛽−𝛿+𝑘1 𝑎𝛿)2 Fig. 1(a) shows the bifurcation diagram of system (4) with 𝐸 as the
have,
bifurcating parameter. The dashed red and solid blue curves represent
𝑎𝛽(𝛽 − 𝛿)
𝐸 𝑚𝑎𝑥 − < 0 < 𝐸 < 𝐸 𝑚𝑎𝑥 . the unstable and stable part of the predator-free equilibrium. The solid
(𝛽 − 𝛿 + 𝑘1 𝑎𝛿)2
green line is the stable part of the interior equilibrium, and the green
Hence, 𝑃 ∗ is locally asymptotically stable whenever it exists. curve represents the maximum and minimum value of the limit cycle
𝑎𝛽(𝛽−𝛿) 𝑎𝛽(𝛽−𝛿)
(𝑖𝑖) If 𝐸 𝑚𝑎𝑥 > (𝛽−𝛿+𝑘 𝑎𝛿)2
then ∃ 𝐸 𝐻 𝑜𝑝𝑓 = 𝐸 𝑚𝑎𝑥 − (𝛽−𝛿+𝑘 𝑎𝛿)2
(> 0). So from at the corresponding value of 𝐸 of the prey species. If 𝐸 < 𝐸 𝐻 𝑜𝑝𝑓 , then
1 1
the above analysis, we have 0 < 𝐸 𝐻 𝑜𝑝𝑓 < 𝐸 𝑚𝑎𝑥 . Hence if 𝐸 ∈ [0, 𝐸 𝐻 𝑜𝑝𝑓 ) the system is unstable, and if 𝐸 > 𝐸 𝐻 𝑜𝑝𝑓 then the system is stable.

4
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

Fig. 1. (a) Bifurcation diagram of system (4) with 𝐸 as the bifurcation parameter. The red curve (dashed) and the blue curve (solid) represent the unstable and stable part
of the predator-free equilibrium, respectively. 𝐸 𝐻 𝑜𝑝𝑓 and 𝐸 𝑇 𝐶 are the critical value of the Hopf bifurcation and transcritical bifurcation, respectively. (b), (c), (d) The phase
diagrams of system (4) with 𝐸 = 0.28 < 0.2982 = 𝐸 𝐻 𝑜𝑝𝑓 , 𝐸 = 0.31 > 0.2982 = 𝐸 𝐻 𝑜𝑝𝑓 , and 𝐸 = 0.77 > 0.7608 = 𝐸 𝑇 𝐶 , respectively. The remaining parameter values are
𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, and 𝑘1 = 0.7. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Moreover, the first Lyapunov number, 𝛤 = −4.98997𝜋 < 0, implies is ecologically appropriate, the equilibrium point 𝑃 ∗ (𝑢∗ , 𝑣∗ ) is feasible
that the Hopf bifurcation is supercritical. Fig. 1(b) shows the unstable provided 𝑘1 > 𝑘𝑚𝑖𝑛1
. From Eq. (8) we have,
equilibrium point 𝑃 ∗ surrounded by a stable limit cycle. As the bifur- [ ]
𝑢∗ 𝑢∗ 𝑢∗
cating parameter is increased, the equilibrium point becomes stable, 𝑡𝑟(J0 ) = − + 1 − 𝐸 − ≡ 𝜙(𝑘1 ).
(1 + 𝑘1 𝑢∗ )2 𝑎 + 𝑢∗ 1 + 𝑘1 𝑢∗
which is displayed in Fig. 1(c). Here the dimensionless parameter 𝐸
is the ratio of prey harvesting effort to its growth rate. In comparison The nature of 𝜙(𝑘1 ) will determine the stability of the equilibrium point.
to the harvesting effort, growth rate is a slow intrinsic parameter. So, Clearly,
the change is negligible compared to the changes in the harvesting
𝑢∗
effort. Hence, an increase in 𝐸 indicates a corresponding increase in lim 𝜙(𝑘1 ) = − (< 0). (10)
𝑘1 →𝑘𝑚𝑖𝑛 (1 + 𝑘1 𝑢∗ )2
the prey harvesting effort. Furthermore, Theorem 3.3 shows that either 1

the equilibrium point is locally asymptotically stable whenever it exists

Moreover,
or the unharvested system (𝐸 = 0) is unstable. However, increasing
𝑢∗ (1 − 𝐸)
harvesting effort causes the amplitude of the limit cycle to decrease, lim 𝜙(𝑘1 ) = (> 0). (11)
eventually leading to stabilization up to a certain range. A further
𝑘1 →+∞ 𝑎 + 𝑢∗
increase in harvesting efforts causes the extinction of predator species
Hence Eqs. (10) and (11) implies that as 𝑘1 increases from its lower
and, ultimately, of the prey species. Hence, by controlling harvesting
threshold, the sign of 𝜙(𝑘1 ) changes from negative to positive. Also, the
efforts, limit cycles can be prevented. However, if harvesting is not
controlled, it would eventually lead to the collapse of the entire system. value of 𝜙(𝑘1 ) gets saturated and attains a positive value as 𝑘1 increases.

Lemma 3.4. There always exists a unique value 𝑘(1) such that for 𝑘1 ∈
3.3. The destabilizing effect of 𝑘1 (1)
1
(𝑘𝑚𝑖𝑛
1
, 𝑘 1
) the equilibrium point 𝑃 ∗ (𝑢∗ , 𝑣∗ ) is locally asymptotically stable

and unstable for 𝑘1 > 𝑘(1) .

In this subsection, we will see the impact of the parameter 𝑘1 on the 1
1
dynamics of the temporal system (4). Let 𝑘𝑚𝑖𝑛
1
= 1−𝐸 − 𝑢1∗ . Since 𝐸 < 1

5
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

Fig. 2. (a) Variation of 𝑡𝑟(J0 ) with 𝑘1 is shown. The dark-shaded region indicates an unstable region, while the unshaded region indicates a stable region. (b) Bifurcation diagram
with 𝑘1 as the bifurcating parameter. Here the parameters values are 𝑎 = 0.38, 𝐸 = 0.3, 𝛽 = 0.72, and 𝛿 = 0.31.

Proof. From Eq. (8), 𝜙(𝑘1 ) = 0 is quadratic equation in 𝑘1 . Let 𝑘(1)

1
and section. We consider a bounded region B ⊂ R2 with boundary 𝜕B and
𝑘(2) be its roots where, propose the following model,
1 √ ( )
1 1 4𝑎(1 − 𝐸) + 4𝑢∗ (1 − 𝐸) + (𝑢∗ )2 𝜕 𝑢(𝑥, 𝑦, 𝑡) 𝑢(𝑥, 𝑦, 𝑡) 𝑢(𝑥, 𝑦, 𝑡)𝑣(𝑥, 𝑦, 𝑡)
𝑘(1)
1
= − ∗
+ , 𝜕𝑡
=𝑢(𝑥, 𝑦, 𝑡) 1 −
1 + 𝑘1 𝑢(𝑥, 𝑦, 𝑡)
−
𝑎 + 𝑢(𝑥, 𝑦, 𝑡)
2(1 − 𝐸) 𝑢 2𝑢∗ (1 − 𝐸)
√ (12)
(13)
1 1 4𝑎(1 − 𝐸) + 4𝑢∗ (1 − 𝐸) + (𝑢∗ )2 − 𝐸 𝑢(𝑥, 𝑦, 𝑡) + 𝑑11 𝛥𝑢 + 𝑑12 𝛥𝑣,
𝑘(2)
1
= − − . 𝜕 𝑣(𝑥, 𝑦, 𝑡) 𝛽 𝑢(𝑥, 𝑦, 𝑡)𝑣(𝑥, 𝑦, 𝑡)
2(1 − 𝐸) 𝑢∗ 2𝑢∗ (1 − 𝐸) = − 𝛿 𝑣(𝑥, 𝑦, 𝑡) + 𝑑21 𝛥𝑢 + 𝑑22 𝛥𝑣.
𝜕𝑡 𝑎 + 𝑢(𝑥, 𝑦, 𝑡)
Clearly 𝑘𝑚𝑖𝑛
1
> 𝑘(2)
1
. By the continuity of 𝜙(𝑘1 ) in the interval [𝑘𝑚𝑖𝑛
1
, ∞)
and by Eq. (10) and (11), we get 𝑘𝑚𝑖𝑛 < 𝑘(1) . Therefore, 𝑘𝑚𝑖𝑛 ∈ (𝑘(2) , 𝑘(1) ). Here 𝑢(𝑥, 𝑦, 𝑡) and 𝑣(𝑥, 𝑦, 𝑡) are the non-dimensional prey and preda-
1 1 1 1 1
tor population at position (𝑥, 𝑦) ∈ B and time 𝑡, respectively. 𝑑11 (> 0),
From Eq. (10), for 𝑘1 ∈ (𝑘𝑚𝑖𝑛 1
, 𝑘(1)
1
), 𝑡𝑟(J0 ) < 0. Hence equilibrium
𝑑22 (> 0) and 𝑑12 , 𝑑21 are the self and cross-diffusion coefficients,
point 𝑃 ∗ (𝑢∗ , 𝑣∗ ) is stable and unstable for 𝑘1 > 𝑘(1) . This completes the
1 respectively. Moreover, 𝑑12 and 𝑑21 describe the population fluxes of
proof. □ prey and predator due to the presence of other species, respectively.
The paradox of enrichment, proposed by Rosenzweig [50], is a phe- 𝑑12 > 0 indicates that the prey species tends to diffuse in the lower
nomenon where the prey species’ enrichment leads to the system’s concentration of predator species, and 𝑑21 > 0 indicates the preda-
destabilization, thereby collapsing the entire trophic interaction. Sev- tor species tends to catch prey from a lower concentration group to
avoid group defense by a large number of prey species [19,54]. 𝛥 =
eral studies have been conducted regarding this enigma behavior [51– 𝜕2 2
+ 𝜕𝜕𝑦2 denotes the two-dimensional Laplacian operator. The above
53]. The Lemma 3.4, implies that increasing 𝑘1 tends to destabilize the 𝜕 𝑥2
system (13) is subjected to non-negative initial and no-flux boundary
equilibrium point 𝑃 ∗ (𝑢∗ , 𝑣∗ ). To verify this numerically, we consider
conditions,
the same parameter set as in Fig. 1 with 𝐸 = 0.3. The corresponding
value of trace is plotted in Fig. 2(a). It is observed that the increment 𝑢(𝑥, 𝑦, 𝑡) ≥ 0, 𝑣(𝑥, 𝑦, 𝑡) ≥ 0, (𝑥, 𝑦) ∈ B, (14)
in the influence parameter 𝑘1 makes 𝑡𝑟(J0 ) positive, hence unstable
𝜕𝑢 𝜕𝑣
via limit cycles. The bifurcation plot is also shown in Fig. 2(b) with = = 0, (𝑥, 𝑦) ∈ 𝜕B, 𝑡 > 0. (15)
𝜕𝑛 𝜕𝑛
𝑘1 as the bifurcating parameter. The red dot represents the predator
Where 𝑛 is the unit outward normal vector along 𝜕B and 𝜕 𝑛 represents
species at the stable interior equilibrium, while the red lines represent
the operator of the directional derivative along the direction 𝑛. By the
the predator species at the limit cycles. As 𝑘1 increases through 𝑘(1)
1
, the
law of thermodynamics, the eigenvalue of the diffusion matrix 𝐷 =
system experiences a Hopf bifurcation. This indicates the destabilizing
[𝑑𝑖𝑗 ]2×2 are positive, so we assume 𝑑11 𝑑22 − 𝑑12 𝑑21 > 0. This implies that
nature of the coefficient 𝑘1 . So, the species’ impact on its carrying
the flow species is strongly governed by its own density rather than the
capacity plays a major role in the overall dynamics of the system. Only
other species [24,55,56].
in the marginal case can the system be stable (i.e., 𝑘1 ∈ (𝑘𝑚𝑖𝑛 1
, 𝑘(1)
1
).
The system becomes unstable if the species highly influences its own
4.1. Turing instability
carrying capacity constructively. A positive value of 𝑘1 indicates that
the prey species work together to ameliorate their carrying capacity. So
In this subsection, sufficient conditions required for cross-diffusion-
increase in the parameter 𝑘1 correlates to the enrichment of the system.
driven instability of the system (13) will be derived. Turing instability
However, from the above discussions, an increment in the influence
arises when the stable temporal system becomes unstable as a result of a
parameter 𝑘1 always destabilizes the equilibrium point via limit cycles. small amplitude heterogeneous perturbation around the homogeneous
steady state. Linearizing (13) around the equilibrium point 𝑃 ∗ (𝑢∗ , 𝑣∗ ),
4. Spatio-temporal model 𝑢 =𝑢∗ + 𝜖1 𝑒𝜆𝑡+𝑖(𝜔𝑥 𝑥+𝜔𝑦 𝑦) ,
𝑣 =𝑣∗ + 𝜖2 𝑒𝜆𝑡+𝑖(𝜔𝑥 𝑥+𝜔𝑦 𝑦) ,

The spatio-temporal model with both self and cross-diffusion terms where√𝜆 is the growth rate of perturbation and 0 < 𝜖1 , 𝜖2 ≪ 1. Also,
corresponding to the temporal model (4) will be discussed in this 𝜔 = 𝜔2𝑥 + 𝜔2𝑦 is the wave number. Now, substituting the above in

6
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

Fig. 3. (a) Bifurcation diagram in the 𝐸 − 𝑘1 plane. The parameters values are 𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, 𝑑11 = 3.5, 𝑑12 = 1.2, 𝑑21 = 5.5, and 𝑑22 = 6. (b) Bifurcation diagram in
the 𝐸 − 𝑑21 plane. The parameters values are 𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, 𝑘1 = 0.7, 𝑑11 = 0.5, 𝑑12 = 0.2, and 𝑑22 = 0.25. In the non-feasible region, the equilibrium point does not
coexist. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Eq. (13) and neglecting higher-order terms, we get the characteristic

equation, Remark 3. If the initial two conditions of Eq. (17) holds, then solving
|J𝜔 − 𝜆𝐼| = 0, the last inequality, we can obtain the critical value for Turing bifurca-
( ) tion 𝐸 𝑇 𝑢𝑟 . Since 𝑎22 = 0, and in the absence of cross-diffusion (i.e 𝑑12 =
𝑑11 𝑑12
where J𝜔 = J0 −𝜔2 𝐷 and 𝐷 = . We obtain the characteristic 𝑑21 = 0), we get 𝑝2 < 0. Hence ℎ(𝜔2 ) > 0 ∀ 𝜔 i.e. 𝑑 𝑒𝑡(J𝜔 ) > 0 ∀ 𝜔.
𝑑21 𝑑22
equation, Hence, system (13) cannot induce Turing patterns without the presence
of cross-diffusion terms.
𝜆2 − 𝑡𝑟(J𝜔 )𝜆 + 𝑑 𝑒𝑡(J𝜔 ) = 0,
where, 5. Numerical simulations
𝑡𝑟(J𝜔 ) =𝑡𝑟(J0 ) − 𝜔2 (𝑑11 + 𝑑22 ),
In this section, we perform numerical simulations to verify the
𝑑 𝑒𝑡(J𝜔 ) =(𝑑11 𝑑22 − 𝑑12 𝑑21 )𝜔4 − (𝑑11 𝑎22 + 𝑑22 𝑎11 − 𝑑12 𝑎21 − 𝑑21 𝑎12 )𝜔2
above-discussed theoretical results and visualize various spatio-
+ 𝑑 𝑒𝑡(J0 ). temporal patterns. We consider a bounded square domain B ⊂ R2
The 𝑡𝑟(J0 ) and 𝑑 𝑒𝑡(J0 ) are given in Eq. (8). Hence, system (13) will be and zero-flux boundary conditions. This boundary condition indicates
stable at the equilibrium 𝑃 ∗ (𝑢∗ , 𝑣∗ ) if 𝑡𝑟(J𝜔 ) < 0 and 𝑑 𝑒𝑡(J𝜔 ) > 0. So, species cannot enter or leave the bounded square domain. Hence
any of these conditions must be violated for Turing instability. When allowing us to observe the self-organization of spatio-temporal patterns
the temporal system is stable around 𝑃 ∗ , then we notice that, without the effect of external input. Non-negative initial conditions are
𝑡𝑟(J𝜔 ) = 𝑡𝑟(J0 ) − 𝜔2 (𝑑11 + 𝑑22 ) < 𝑡𝑟(J0 ) < 0. employed. More precisely,
𝑢(𝑥, 𝑦, 0) =𝑢∗ − 𝜖 × 𝜂1 ,
So the only possible way for the system (13) to be unstable at the
equilibrium 𝑃 ∗ (𝑢∗ , 𝑣∗ ) is 𝑑 𝑒𝑡(J𝜔 ) < 0. Let, 𝑣(𝑥, 𝑦, 0) =𝑣∗ − 𝜖 × 𝜂2 ,

ℎ(𝜔2 ) = 𝑝1 𝜔4 − 𝑝2 𝜔2 + 𝑑 𝑒𝑡(J0 ). (16) where 𝜖 = 1 × 10−5 and 𝜂1 , 𝜂2 are random numbers between [−0.5, 0.5].
FTCS (Forward Time Centered Space) scheme is used to discretize the
where,
system (13) in time and space (cf. [24]). We choose B = [0, 400] ×
𝑝1 = 𝑑11 𝑑22 − 𝑑12 𝑑21 (> 0), [0, 400], time step 𝛥𝑡 = 0.01 and space step 𝛥𝑥 = 𝛥𝑦 = 1. Numerical
simulations are done using the MATLAB software. The spatial allotment
of prey and predator species are just transposed of each other. For
𝑝2 = 𝑑11 𝑎22 + 𝑑22 𝑎11 − 𝑑12 𝑎21 − 𝑑21 𝑎12 . this reason, we only present the spatial distribution of prey species.
The minimum value of ℎ occurs at 𝜔2𝑚𝑖𝑛 =
𝑝2
and ℎ(𝜔2𝑚𝑖𝑛 ) = 𝑑 𝑒𝑡(J0 ) − Moreover, simulations are carried out for a long time until no further
2𝑝1 pattern changes are observed.
𝑝22
. So Turing instability occurs when ℎ(𝜔2𝑚𝑖𝑛 ) < 0 and correspondingly
4𝑝1 √
𝑑 𝑒𝑡(J0 ) 5.1. Impact of parameter 𝐸
the critical wave number is 𝜔𝑐 𝑟 = 𝑝
. So, combining the above
1
analysis, we have, The non-dimensional parameter 𝐸 in the system (13) is the ratio
of harvesting effort to the intrinsic growth rate of prey species. An
Theorem 4.1. The system (13) undergoes Turing instability around the increase in 𝐸 indicates an increase in the harvesting effort or a decrease
equilibrium point 𝑃 ∗ (𝑢∗ , 𝑣∗ ) if the following conditions holds, in the prey growth rate. We observe that variation of this parameter
𝑡𝑟(J0 ) =𝑎11 + 𝑎22 < 0, produces various types of Turing patterns. Using the expression of
𝑝2 =𝑑11 𝑎22 + 𝑑22 𝑎11 − 𝑑12 𝑎21 − 𝑑21 𝑎12 > 0, 𝑎𝑛𝑑 trace from Eq. (8) and equating it to zero, we obtain the Hopf curve.
√ Similarly, we obtain the Turing curve by equating the expression of 𝑝3
𝑝3 =𝑝2 − 4𝑝1 𝑑 𝑒𝑡(J𝜔 ), (17)
from Eq. (17) to zero. Both these curves are plotted in the 𝐸 − 𝑘1 plane
=𝑑11 𝑎22 + 𝑑22 𝑎11 − 𝑑12 𝑎21 − 𝑑21 𝑎12 in Fig. 3(a) and are denoted by blue and red curves, respectively. The
√
− 4(𝑑11 𝑑22 − 𝑑12 𝑑21 )𝑑 𝑒𝑡(J0 ) > 0. curves divide the 𝐸 − 𝑘1 plane into four regions: stable, Hopf, Turing,

7
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

Fig. 4. (a) Dispersion relation is plotted with 𝐸 as the varying parameter. (b) Plot of 𝑑 𝑒𝑡(J𝜔 ) against 𝜔2 . The dotted line represents the plot at Turing bifurcation (𝐸 𝑇 𝑢𝑟 ). Here
the parameters values are 𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, 𝑘1 = 0.7, 𝑑11 = 3.5, 𝑑12 = 1.2, 𝑑21 = 5.5, and 𝑑22 = 6.

Fig. 5. Snapshots of the evolution of prey species after time 𝑡 = 50,000 for different values of (a) 𝐸 = 0.5, (b) 𝐸 = 0.6, (c) 𝐸 = 0.65, (d) 𝐸 = 0.675, (e) 𝐸 = 0.718, and (f) 𝐸 = 0.75. The
rest of the parameters are 𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, 𝑘1 = 0.7, 𝑑11 = 3.5, 𝑑12 = 1.2, 𝑑21 = 5.5, and 𝑑22 = 6.

and the Hopf-Turing region. We take the parameter set from the Turing 𝐸 can drive the system to Turing instability. Hence, we observe that
region and vary the parameter 𝐸 to observe the formation of Turing the more constructive the species is towards their carrying capacity,
patterns. the higher the chance the parameter 𝐸 (or the harvesting effort) has
We consider 𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, 𝑑11 = 3.5, 𝑑12 = to drive the system to Turing instability. Ecologically, this may suggest
1.2, 𝑑21 = 5.5, and 𝑑22 = 6. The black dotted line is the threshold curve that as the prey species begin to work together to better their living
beyond which the equilibrium point ceases to coexist. If 𝑘1 = 0.7, then conditions, a small variation in harvesting effort (which can also be
from Eqs. (8) and (17), we get 𝐸 𝐻 𝑜𝑝𝑓 ∼ 0.2982 and 𝐸 𝑇 𝑢𝑟 ∼ 0.3763, regarded as an external disturbance) can make the system spatially
respectively. The dispersion relation is shown in Fig. 4 for varying 𝐸 unstable, and there are likely chances for the system to generate Turing
and fixed 𝑘1 . Also the variation of 𝑑 𝑒𝑡(J𝜔 ) with respect to 𝜔2 (wave patterns.
number) is displayed. In both figures, the dotted line represents the From Fig. 5, we observe the emergence of hot spot when the value of
curve at Turing bifurcation (𝐸 𝑇 𝑢𝑟 ). From Fig. 3(a), we see that for 𝐸 > 𝐸 is low, and as the parameter value increases, a mixture of hot spot and
𝐸 𝑇 𝑢𝑟 and less than the threshold value (𝐸 𝑚𝑎𝑥 ), Turing instability occurs, stripes is seen. With further increase in its values, labyrinth types pattern
and patterns start to arise. Moreover, from Fig. 3(a), the Turing region are seen, followed by a mixture of cold spots and stripes. As 𝐸 approaches
decreases for decreasing value of 𝑘1 . This indicates that for a lower the threshold value, cold spot dominates the entire region. Hence, we
value of 𝑘1 , 𝐸 induced spatial instability is not seen or observed only observe a variety of patterns as parameter 𝐸 is varied. From the model
for a very small range. However, for higher values of 𝑘1 , the parameter (13), it is clear that the predator depends solely on prey for food, so

8
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

Fig. 6. Snapshots of the evolution of prey species after time 𝑡 = 20,000 for different values of (a) 𝑘1 = −0.1, (b) 𝑘1 = 0.1, and (c) 𝑘1 = 1. The rest of the parameters are
𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, 𝐸 = 0.67, 𝑑11 = 3.5, 𝑑12 = 1.2, 𝑑21 = 5.5, and 𝑑22 = 6.

Fig. 7. Time series plot of average prey and average predator density for different values of influence parameter 𝑘1 . The parameter sets are the same as in Fig. 6.

with an increase in the value of 𝐸, the predator species declines more 5.3. Impact of cross-diffusion coefficient 𝑑21
rapidly as compared to the prey species. Hence, with a lesser number
of predators in the region, the prey species begin to diffuse and venture As discussed earlier, a positive value of 𝑑21 indicates the predator
out more, thereby leading to the formation of several types of patterns. species prefers to take prey from a lower concentration group to evade
group defense by a large number of prey species. In this subsection,
we will see the effect of the cross-diffusion term 𝑑21 in the formation
5.2. Impact of parameter 𝑘1 of patterns of system (13). We take the same parameters as before,
𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, 𝑑11 = 3.5, 𝑑12 = 1.2, 𝑑22 = 6 and fix 𝐸 = 0.5
and 𝑘1 = 0.7. Setting, 𝑑21 = 5.5, 7, 10 respectively, we obtain spatial
The parameter 𝑘1 indicates the prey species’ influence on their own patterns as shown in Fig. 8. In Fig. 8(a) hot spot patterns are seen in the
carrying capacity. In Section 3.3, we have already seen the destabilizing entire domain when 𝑑21 = 5.5. Now, as the cross-diffusion parameter
effect of 𝑘1 in the temporal model. Here, we will observe its impact 𝑑21 is increased to 7, slowly hot stripe pattern begins to emerge. As a
on the spatio-temporal model (13). We fix 𝐸 = 0.67 and consider the result, mixed type of hot spot and hot stripe patterns. When the parameter
value is further increased to 10, the hot stripe gets longer, and such
same parameter set as in Fig. 5. The negative value of 𝑘1 indicates the
patterns start prevailing over the entire domain. In Fig. 9, the time
destructive role of prey species towards their own carrying capacity,
series evolution of average prey and average predator species for the
while its positive value indicates a constructive role. For 𝑘1 = −0.1, hot
above values of the cross-diffusion parameter is plotted. We find that an
and cold stripe patterns are seen in Fig. 6(a). With a gradual increase increase in 𝑑21 leads to a decrease in the average prey density. On the
in its value, cold stripe gets dense, and we can see the formation of other hand, the average predator density increases with an increase in
labyrinth pattern for 𝑘1 = 1. To further analyze the effect of the 𝑑21 . Now, places with a lower prey species density correspond to places
influence parameter on the spatial dynamics of the system (13), we with higher predator density due to the positive value of 𝑑21 . As this
plot the time series evolution of average prey and predator density value increases, the intensity of predators attacking regions with lower
for different values of 𝑘1 in Fig. 7. The initial value for both prey prey density increases. As a result, prey species start forming long hot
and predator species is taken as a small random perturbation from the stripe patterns. However, in doing so, their average density decreases,
equilibrium point. Since only the predator equilibrium point depends and they become more accessible to predators. As a result, the average
predator density increases.
on the value of 𝑘1 , so in Fig. 7 the initial value of prey species for
different values of 𝑘1 are the same while for the predator species, the
5.4. Spatio-temporal chaos
initial value changes as 𝑘1 varies. Moreover, we can also observe that
as 𝑘1 increases, the steady state pattern takes a much shorter time to In this section, we investigate the emergence of non-Turing patterns,
settle down. So, ecologically, we can interpret that as the prey species especially the spatio-temporal chaotic pattern. Many ecologically mean-
start to better their living conditions, both the species start to localize ingful initial conditions have been employed by researchers [57–59].
faster, forming Turing patterns. The average prey and predator density Sun et al. [58] introduced the following initial condition.
also rise as the value of 𝑘1 increases, indicating that increment in 𝑘1 is 𝑢(𝑥, 𝑦, 0) = 𝑢∗ − 10−5 (𝑥 − 200),
beneficial to the system. 𝑣(𝑥, 𝑦, 0) = 𝑣∗ − 10−5 (𝑦 − 200).

9
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

Fig. 8. Snapshots of evolution of prey species after time 𝑡 = 50,000 for different value of (𝑎) 𝑑21 = 5.5, (𝑏) 𝑑21 = 7, (𝑐) 𝑑21 = 10. The rest of the parameters are
𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, 𝑘1 = 0.7, 𝐸 = 0.5, 𝑑11 = 3.5, 𝑑12 = 1.2, and 𝑑22 = 6.

Fig. 9. Time series plot of average prey and average predator density for different values of cross-diffusion coefficient 𝑑21 . The parameter sets are the same as in Fig. 8, and the
trajectory remains the same with a further increase in time.

the activities of species can have a substantial negative or positive

To investigate chaotic patterns, we take a parameter set from the impact on their carrying capacity. So a linear function of species density
Hopf region (see Fig. 3(b)). For the system to be in this region, the is used instead of the constant carrying capacity to account for this
parameter 𝐸 must be small, or the influence parameter 𝑘1 must be high. change. The effect is constructive if the function’s slope is positive.
We consider the following values of the parameters, 𝑎 = 0.38, 𝛽 = Together, the species improve their living conditions and surroundings.
0.72, 𝛿 = 0.31, 𝑘1 = 0.7, 𝐸 = 0.1, 𝑑11 = 0.5, 𝑑12 = 0.2, 𝑑21 = The impact is destructive if the slope is negative. Moreover, we consider
0.25, and 𝑑22 = 0.25. Fig. 10 shows the emergence of chaotic patterns. harvesting of the prey species, which follows a proportional law.
A spiral wave pattern begins to appear because of the above initial Three steady states exist for the system: the trivial, the predator-
condition. Slowly, this spiral pattern begins to break, and we observe free, and the interior equilibrium. The conditions for their existence
the formation of chaotic patterns. As enrichment increases in the tem- and stability are also derived. Considering the dimensionless harvesting
poral case, the size of the limit cycle expands and moves closer to the effort (𝐸) as the bifurcating parameter, it is proven analytically that
axis. Due to stochastic environmental changes, a decreased population either the system is locally asymptotically stable or exhibits a Hopf bi-
density in a realistic ecosystem increases the likelihood that a species furcation. The stability of limit cycles is also investigated. Additionally,
would go extinct [60]. Thus, there is a risk of species extinction when the amplitude of the stable limit cycle diminishes as the parameter 𝐸 is
enrichment levels rise. But when spatio-temporal dynamics are taken increased. As the rise continues, a stable steady state forms, suggesting
into account, a transition to spatio-temporal chaos is seen. To illustrate that the harvesting effort stabilizes the system’s dynamics (see Fig. 1).
this change, Fig. 11 shows the time series of prey density and the spatial So temporal analysis reveals that prey harvesting has a stabilizing
average of prey density for the temporal and spatio-temporal cases, effect on the equilibrium point when the unharvested system is at a
respectively. The initial value is considered a slight perturbation from non-equilibrium state. This observation is consistent with most of the
the equilibrium state in both figures, and the rest of the parameter previous studies [14,44,45]. Srinivasu et al. [61] also reported this type
values are the same. In the temporal scenario (without diffusion), of behavior in a prey-predator system with simultaneous harvesting
limit cycles with large amplitudes are observed (Fig. 11(a)). When of both species. They discussed methods to avoid this cyclic behavior
spatial heterogeneity is taken into account, the species density, how- and eventually stabilize the system by increasing the harvesting efforts.
ever, becomes extremely erratic and uncorrelated in space Fig. 11(b). When spatial aspects are considered, variations of harvesting effort
As a result, the population’s oscillatory behavior is desynchronized, result in the formation of various Turing patterns (see Fig. 5). From
lowering the threat of extinction. this, we can infer that there is some sensitivity to slight variations in
the value of the harvesting parameter, which causes abrupt changes
6. Conclusion to the dynamical system. Furthermore, the simulation also shows the
formation of spatio-temporal chaotic patterns for low harvesting effort
In this study, we investigated a predator–prey system with variable (see Fig. 10).
carrying capacity. A modified Rosenzweig-MacArthur model is used to Yukalov et al. [40,41] and Pati et al. [42] have incorporated sim-
depict the predator–prey relationship. The modification is done because ilar variable carrying capacity in their system. However, their study

10
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

Fig. 10. Evolution of chaotic patterns for prey species at different time (𝑎) 𝑡 = 180, (𝑏) 𝑡 = 500, (𝑐) 𝑡 = 850, (𝑑) 𝑡 = 1200, (𝑒) 𝑡 = 5000, (𝑓 ) 𝑡 = 10 000. The parameter values are
𝑎 = 0.38, 𝛽 = 0.72, 𝛿 = 0.31, 𝑘1 = 0.7, 𝐸 = 0.1, 𝑑11 = 0.5, 𝑑12 = 0.2, 𝑑21 = 0.25, and 𝑑22 = 0.25.

is unlikely as enrichment rises, it does cause the limit cycles’ amplitude

to grow (see Section 3.3). As a result, the population density is brought
to a very low level, where random environmental variations increase
the likelihood of extinction.
Species exist in both time and space in ecology. To understand this
spatial inhomogeneity, we utilize a reaction–diffusion equation with
linear cross-diffusion. Neumann boundary conditions and non-negative
initial conditions are employed. The Turing instability requirements are
derived. It is further proved that the system (13) cannot exhibit Turing
patterns in the absence of cross-diffusion factors. To visualize various
Turing patterns, extensive numerical simulations are performed. Most
spatio-temporal analyses are done using two parameters, 𝐸 and 𝑘1 .
The bifurcation diagram, plotted in Fig. 3(a), assists us in selecting
appropriate parameter values from the Turing region. The subsequent
formation of various Turing patterns as 𝐸 is varied is shown in Fig. 5.
As 𝐸 is increased, the transition from hot spots → hot and stripes →
labyrinths → cold and stripes → cold spots, is shown. Such a wide range of
patterns was not found when 𝑘1 was varied (Fig. 6). This could result
from the predator species becoming saturated at low densities as the
𝑘1 increases. The temporal system becomes unstable when 𝑘1 increases
Fig. 11. Effect of enrichment to the system (a) without diffusion, and (b) with further, and we leave the Turing region. Furthermore, the patterns form
diffusion. The initial conditions are taken as a small perturbation from the equilibrium
point in both cases.
more quickly as the species grows more constructive (see Fig. 7). A
further investigation is done into the influence of the cross-diffusion
term 𝑑21 in the development of spatial patterns. With an increase in
the diffusion coefficient 𝑑21 , the hot spots get replaced by hot stripes
focussed more on the dynamical concept and the occurrence of delay- (see Fig. 8). The average prey density lowers, and they become more
induced supercritical and subcritical Hopf bifurcations. The novelty accessible to the predator, which in turn leads to an increase in the
of our research lies in the fact that we have explored their concept average predator density (see Fig. 9).
of variable carrying capacity from a more ecological perspective. The Chaos is not rare in an ecosystem. A recent study by Rogers
species’ constructive and destructive characteristics are investigated. It et al. [62] used various chaos-detecting methods in a global database
has been noted that the greater the positive influence of species on and found that chaos was prevalent in more than 30% of them. Chaotic
their carrying capacity, the more likely it is to destabilize the system. dynamics facilitate species’ biodiversity and make the system more ro-
This positive influence is directly tied to an increase in species’ carrying bust. Using numerical simulations, Petrovskii et al. [60] demonstrated
capacity. Thus, enrichment does lead to destabilization in our system that a shift from oscillatory regime to spatio-temporal chaos could
via the formation of stable limit cycles (see Fig. 2). Although extinction avert species extinction. They suggested that if the temporal system

11
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

is on the verge of extinction, individuals from other lattice sites can re- References
colonize this empty site. This movement is possible due to the diffusion
terms. The Hopf region in Fig. 3 indicates that the system is unstable in [1] May RM. Simple mathematical models with very complicated dynamics. Nature
1976;261:459–67.
both temporal and spatial domains. Taking the parameter set from this
[2] Segel LA, Jackson JL. Dissipative structure: an explanation and an ecological
region, we observe the emergence of chaotic spatio-temporal patterns example. J Theoret Biol 1972;37(3):545–59.
(see Fig. 10) for various time intervals. An increase in the influence [3] Kot M. Elements of mathematical ecology. Cambridge University Press; 2001.
parameter 𝑘1 increases the carrying capacity. So, without spatial het- [4] Biswas S, Bhutia LT, Kar TK. Transient and asymptotic dynamics of
erogeneity, an increase in 𝑘1 always destabilizes the equilibrium point Bazykin’s prey-predator model on managing reactivity, resilience, and maximum
sustainable yield. Eur Phys J Plus 2023;138(3):256.
via limit cycles. Through this destabilization, the predator and prey
[5] Rihan FA. Delay differential equations and applications to biology. Springer;
species are synchronized, and when they reach low-density values, 2021.
stochastic perturbation increases the likelihood of species extinction. [6] Sajan, Anshu, Dubey B. Study of a cannibalistic prey–predator model with
However, in the presence of spatial heterogeneity, such catastrophic Allee effect in prey under the presence of diffusion. Chaos Solitons Fractals
events are unlikely to occur as the species become highly erratic and 2024;182:114797.
[7] Kumbhakar R, Hossain M, Pal N. Dynamics of a two-prey one-predator model
uncorrelated in space. As a result, the behavior of both predator and
with fear and group defense: A study in parameter planes. Chaos Solitons Fractals
prey species becomes desynchronized, lowering the risk of extinction. 2024;179:114449.
By keeping the parameter set identical, the effect of enrichment on [8] Chowdhury SN, Banerjee J, Perc M, Ghosh D. Eco-evolutionary cyclic dominance
the absence and presence of spatial heterogeneity is illustrated in among predators, prey, and parasites. J Theoret Biol 2023;564:111446.
Fig. 11. Limit cycles with large amplitudes are observed in the temporal [9] Berryman AA. The orgins and evolution of predator-prey theory. Ecology
1992;73(5):1530–5.
case, while chaotic trajectory is seen in the spatio-temporal case. This
[10] Rosenzweig ML, MacArthur RH. Graphical representation and stability conditions
indicates that spatial heterogeneity lowers the species’ fluctuations and of predator-prey interactions. Amer Nat 1963;97(895):209–23.
decreases their danger of eradication. Recently, Moustafa et al. [17] [11] Smith HL. The Rosenzweig-MacArthur predator-prey model. Phoenix: School of
also demonstrated a possible proposition to the paradox of enrichment Mathematical and Statistical Sciences, Arizona State University; 2008.
by considering a fractional-order system. [12] Kar TK. Stability analysis of a prey–predator model incorporating a prey refuge.
Commun Nonlinear Sci Numer Simul 2005;10(6):681–91.
We hope that our work could be beneficial in studying the effect of
[13] Peng G, Jiang Y, Li C. Bifurcations of a Holling-type II predator–prey system
spatial heterogeneity in prey-predator systems with variable carrying with constant rate harvesting. Int J Bifurcation Chaos 2009;19(08):2499–514.
capacities and harvesting. We can also investigate the corresponding [14] Barman B, Ghosh B. Explicit impacts of harvesting in delayed predator-prey
stochastic model by incorporating some noise into our deterministic models. Chaos Solitons Fractals 2019;122:213–28.
model. Furthermore, the study of predator–prey systems with the Allee [15] Rihan FA, Lakshmanan S, Hashish AH, Rakkiyappan R, Ahmed E. Fractional-
order delayed predator–prey systems with Holling type-II functional response.
effect and variable carrying capacity would also be an interesting
Nonlinear Dynam 2015;80:777–89.
study. For a more realistic ecological system, time delay can also be [16] Javidi M, Nyamoradi N. Dynamic analysis of a fractional order prey–predator
incorporated and studied. We leave this for our future work. interaction with harvesting. Appl Math Model 2013;37(20–21):8946–56.
[17] Moustafa M, Mohd MH, Ismail AI, Abdullah FA. Dynamical analysis of a
CRediT authorship contribution statement fractional-order Rosenzweig–MacArthur model incorporating a prey refuge.
Chaos Solitons Fractals 2018;109:1–13.
[18] Wyse SK, Martignoni MM, Mata MA, Foxall E, Tyson RC. Structural sensitivity in
Lakpa Thendup Bhutia: Writing – review & editing, Writing – the functional responses of predator–prey models. Ecol Complex 2022;51:101014.
original draft, Visualization, Validation, Software, Methodology, For- [19] Yang J, Zhang T, Yuan S. Turing pattern induced by cross-diffusion in a
mal analysis, Conceptualization. Samir Biswas: Writing – review & predator–prey model with pack predation-herd behavior. Int J Bifurcation Chaos
editing, Writing – original draft, Visualization, Validation, Software, 2020;30(07):2050103.
[20] Zhang X-C, Sun G-Q, Jin Z. Spatial dynamics in a predator-prey model with
Methodology, Formal analysis, Conceptualization. Esita Das: Writing Beddington-DeAngelis functional response. Phys Rev E 2012;85(2):021924.
– review & editing, Conceptualization. Tapan Kumar Kar: Writing – [21] Wang X, Efendiev M, Lutscher F. How spatial heterogeneity affects transient
review & editing, Validation, Supervision, Conceptualization. Bidhan behavior in reaction–diffusion systems for ecological interactions? Bull Math Biol
Bhunia: Writing – review & editing, Software, Conceptualization. 2019;81:3889–917.
[22] Ning L-Y, Luo X-F, Li B-L, Wu Y-P, Sun G-Q, Feng T-C. An effective Allee effect
may induce the survival of low-density predator. Results Phys 2023;53:106926.
Declaration of competing interest [23] Abid W, Yafia R, Aziz-Alaoui MA, Aghriche A. Turing instability and Hopf
bifurcation in a modified Leslie–Gower predator–prey model with cross-diffusion.
The authors declare that they have no known competing finan- Int J Bifurcation Chaos 2018;28(07):1850089.
cial interests or personal relationships that could have appeared to [24] Ghorai S, Poria S. Turing patterns induced by cross-diffusion in a predator-prey
system in presence of habitat complexity. Chaos Solitons Fractals 2016;91:421–9.
influence the work reported in this paper.
[25] Liang J, Sun G. Effect of nonlocal delay with strong kernel on vegetation pattern.
J Appl Anal Comput 2024;14(1):473–505.
Acknowledgments [26] Liang J, Sun G-Q. Effects of climate change on vegetation pattern in Baotou,
China. Nonlinear Dynam 2024;112(10):8675–93.
The research work of Lakpa Thendup Bhutia is financially supported [27] Hou L-F, Sun G-Q, Perc M. The impact of heterogeneous human activity on
vegetation patterns in arid environments. Commun Nonlinear Sci Numer Simul
by Council of Scientific and Industrial Research (CSIR), India (File
2023;126:107461.
No. 08/0003(13400)/2022-EMR-I, dated: 4th March 2022) and the re- [28] Guin LN, Pal S, Chakravarty S, Djilali S. Pattern dynamics of a reaction-
search work of Samir Biswas is financially supported by the University diffusion predator-prey system with both refuge and harvesting. Int J Biomath
Grants Commission (UGC), India (NTA Ref. No.: 201610175117, dated: 2021;14(01):2050084.
1st April 2021). The work by Esita Das is financially supported by [29] Zhang L, Wang W, Xue Y. Spatiotemporal complexity of a predator–prey system
with constant harvest rate. Chaos Solitons Fractals 2009;41(1):38–46.
the Indian Institute of Engineering Science and Technology (IIEST),
[30] Upadhyay RK, Roy P, Datta J. Complex dynamics of ecological systems under
Shibpur, India (No. 1999/Exam, dated: October 01, 2021), and the nonlinear harvesting: Hopf bifurcation and turing instability. Nonlinear Dyn
research work of Bidhan Bhunia is financially supported UGC-NFSC 2015;79:2251–70.
(Ref. No. 191620076432). [31] Fan M, Wang K. Optimal harvesting policy for single population with periodic
coefficients. Math Biosci 1998;152(2):165–78.
[32] Lutscher F, Wang X. Reactivity of communities at equilibrium and periodic orbits.
Data availability
J Theoret Biol 2020;493:110240.
[33] Gupta RP, Chandra P. Bifurcation analysis of modified Leslie–Gower predator–
No data was used for the research described in the article. prey model with Michaelis–Menten type prey harvesting. J Math Anal Appl
2013;398(1):278–95.

12
L.T. Bhutia et al. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena 191 (2025) 115790

[34] Ang TK, Safuan HM. Harvesting in a toxicated intraguild predator–prey fishery [49] Kumar S, Kharbanda H. Chaotic behavior of predator-prey model with
model with variable carrying capacity. Chaos Solitons Fractals 2019;126:158–68. group defense and non-linear harvesting in prey. Chaos Solitons Fractals
[35] Ganguli C, Kar TK, Mondal PK. Optimal harvesting of a prey–predator model 2019;119:19–28.
with variable carrying capacity. Int J Biomath 2017;10(05):1750069. [50] Rosenzweig ML. Paradox of enrichment: destabilization of exploitation
[36] Meyer PS, Ausubel JH. Carrying capacity: a model with logistically varying ecosystems in ecological time. Science 1971;171(3969):385–7.
limits. Technol Forecast Soc Change 1999;61(3):209–14. [51] Roy S, Chattopadhyay J. The stability of ecosystems: a brief overview of the
[37] Thornley JHM, France J. An open-ended logistic-based growth function. Ecol paradox of enrichment. J Biosci 2007;32:421–8.
Model 2005;184(2–4):257–61. [52] Gounand I, Mouquet N, Canard E, Guichard F, Hauzy C, Gravel D. The paradox
[38] Safuan HM, Towers I, Jovanoski Z, Sidhu H. Coupled logistic carrying capacity of enrichment in metaecosystems. Amer Nat 2014;184(6):752–63.
model. Anziam J 2011;53:C172–84.
[53] Abrams PA, Walters CJ. Invulnerable prey and the paradox of enrichment.
[39] Al-Moqbali MKA, Al-Salti NS, Elmojtaba IM. Prey–predator models with variable
Ecology 1996;77(4):1125–33.
carrying capacity. Math 2018;6(6):102.
[54] Dubey B, Das B, Hussain J. A predator–prey interaction model with self and
[40] Yukalov VI, Yukalova EP, Sornette D. Punctuated evolution due to delayed
cross-diffusion. Ecol Model 2001;141(1–3):67–76.
carrying capacity. Physica D 2009;238(17):1752–67.
[41] Yukalov VI, Yukalova EP, Sornette D. Extreme events in population dynamics [55] Guin LN, Djilali S, Chakravarty S. Cross-diffusion-driven instability in
with functional carrying capacity. Eur Phys J Spec Top 2012;205(1):313–54. an interacting species model with prey refuge. Chaos Solitons Fractals
[42] Pati NC, Ghosh B. Delayed carrying capacity induced subcritical and super- 2021;153:111501.
critical Hopf bifurcations in a predator–prey system. Math Comput Simulation [56] Vitagliano V. Some phenomenological and thermodynamic aspects of diffusion
2022;195:171–96. in multicomponent systems. Pure Appl Chem 1991;63(10):1441–8.
[43] Legović T, Klanjšček J, Geček S. Maximum sustainable yield and species [57] Medvinsky AB, Petrovskii SV, Tikhonova IA, Malchow H, Li B-L. Spatiotemporal
extinction in ecosystems. Ecol Model 2010;221(12):1569–74. complexity of plankton and fish dynamics. SIAM Rev 2002;44(3):311–70.
[44] Ghosh B, Kar TK, Legović T. Relationship between exploitation, oscillation, MSY [58] Sun G-Q, Jin Z, Li L, Li B-L. Self-organized wave pattern in a predator-prey
and extinction. Math Biosci 2014;256:1–9. model. Nonlinear Dynam 2010;60:265–75.
[45] Kar TK, Matsuda H. Global dynamics and controllability of a harvested prey– [59] Wang W, Zhang L, Wang H, Li Z. Pattern formation of a predator–prey system
predator system with Holling type III functional response. Nonlinear Anal Hybrid with Ivlev-type functional response. Ecol Model 2010;221(2):131–40.
Syst 2007;1(1):59–67. [60] Petrovskii S, Li B-L, Malchow H. Transition to spatiotemporal chaos can resolve
[46] Perko L. Differential equations and dynamical systems, vol. 7, Springer Science the paradox of enrichment. Ecol Complex 2004;1(1):37–47.
& Business Media; 2013.
[61] Srinivasu PDN, Ismail S, Naidu CR. Global dynamics and controllability of a
[47] Liu P, Shi J, Wang Y. Bifurcation from a degenerate simple eigenvalue. J Funct
harvested prey-predator system. J Biol Systems 2001;9(01):67–79.
Anal 2013;264(10):2269–99.
[62] Rogers TL, Johnson BJ, Munch SB. Chaos is not rare in natural ecosystems. Nat
[48] Han R, Dey S, Banerjee M. Spatio-temporal pattern selection in a prey–predator
Ecol Evol 2022;6(8):1105–11.
model with hunting cooperation and Allee effect in prey. Chaos Solitons Fractals
2023;171:113441.

13