Chapter 6

A mathematical model on depletion of

grassland under the effect of invasive species
as a catastrophe: An application to Kaziranga
National Park

6.1 Introduction

One of the important issues that the world needs to address today is the
loss of habitat for wildlife due to various reasons such as pollution, industri-
alization, mining, over exploitation, invasion by alien species, other human
induced activities, urbanization, over growth of population, depletion of re-
sources such as forestry, fishery, fertile topsoil, crude oil, minerals, etc. Due to
loss of natural habitat, species lose the physical area and vital resources which
can lead to catastrophic effect. Tilman et al. (1994); Casagrandi & Gatto
(2002); Brook et al. (2003); Petchey et al. (2008) and Laurance (2010) have
studied the catastrophic effects on species due to habitat loss.
In recent years several investigations were carried out to study the effect
of toxicant/pollutant , degradation/ conservation on various ecosystems using
mathematical models (Shukla et al., 1988; Shukla & Dubey, 1997; Dubey et
al., 2009; Misra & Lata, 2013; Misra et al., 2014a & 2014b and Patra et al.,
(2017). Rai (2008) presented a model for the wetland part of KNP which pro-
vides clear perspectives on future management strategies and policy decisions.
Misra & Lata (2013) have proposed a mathematical models with the effect of
time delay on conservation of forestry biomass. They found in their analysis
that the density of forestry biomass may be conserved if the technological ef-

149
 6.2. Mathematical Model

fort is applied within an appropriate time. Another mathematical model was

presented by Ramdhani et al. (2015), where it was found that if the industrial
crowding increases, then density of biomass of forestry resources decreases and
commented that control of industrialization is necessary to protect the forestry
resources stability. Anderson et al. (2018) in their study found that, invasive
plant species have various negative impacts on the ecosystems they invade. It
has been found through studies that the invasive species lead to decrease in
species diversity causing economic loss as well as reduced forest health and
productivity. These invasive plants will not stop spreading unless controlled.
However, longer its management is postponed, the more the invasive plants
will spread and the more costly they will be to eradicate.
Given this brief literature survey, this chapter aims to study an autonomous
mathematical model with the effect of bad biomass such as Bombax ceiba, Mi-
mosa invisa, Eichornia, Mikenia etc which tend to invade the grassland and
directly reduce the suitable flora for rhinos and other herbivores of KNP. An
uncontrolled growth of such bad biomass may lead to catastrophic effect on
KNP. So, to control the bad biomass and to maintain the grassland at an
appropriate level control effort has been considered. Positivity and bounded-
ness of all solutions are discussed and also the local and global stability of the
interior equilibrium points of the system examined.

6.2 Mathematical Model

We consider the Kaziranga National Park, where patches of grassland area

has been found to be invaded by various species of plants such as Mimosa,
Mikenia, Eichhornia etc. These species have been found to cause serious eco-
logical degradation of the habitat if unchecked and is a major problem con-

150
 6.2. Mathematical Model

fronting the park authorities. Some assumptions and factors of importance for
the proposed model are listed below:

1. Bad biomass is the cumulative densities of Mikenia, Mimosa and Eich-

hornia which depletes the grassland biomass.

2. Densities of grassland biomass and bad biomass are governed by logistic

type equations.

3. Invasion by such species has catastrophic effect on the system since it

does not stop spreading unless human intervene. Thus it is a major cause
of decline of grassland and the whole grazing wildlife population.

4. Existence of interspecific competition between bad biomass and grassland

biomass for abiotic factors such as soil, water, light etc.

5. Effort required to control the growth of bad biomass is directly propor-

tional to existing density of bad biomass.

Based on these assumptions and in conformity with the reality in the KNP, the
dynamics of the problem is governed by the system of non-linear differential
equations given below:

dG  G
= r1 G 1 − − c1 GB − c2 G2 B
dt K
dB  B
= r2 B 1 − − eBE − aGB (6.1)
dt L
dE
= δ1 B − δ2 E
dt

with initial conditions G(0) ≥ 0, B(0) ≥ 0, E(0) ≥ 0, 0 ≤ δ1 , δ2 ≤ 1.

where, G is the density of grassland biomass, B is the density of bad biomass
and E is the measure of effort applied for conservation of grassland biomass.
The constants r1 and K are intrinsic growth rate and carrying capacity of

151
 6.2. Mathematical Model

grassland biomass. r2 and L are intrinsic growth rate and carrying capacity
of bad biomass respectively. The constants c1 and c2 are the depletion rate
coefficients of intrinsic growth rate and carrying capacity of grassland biomass
respectively, due to bad biomass (Misra & Lata, 2013). The constants e and
a are effort coefficient needed to control the bad biomass and interspecific
interference coefficient of bad biomass and grassland biomass respectively. δ1
and δ2 respectively correspond to the growth rate and depreciation rate of
effort.
Boundedness:
To analyze the model, the boundedness of the dependent variable involved
is needed. For this the region of attraction is obtained from the following
lemma.

Lemma 6.2.1 The set

ν
Ω = (G, B, E) ∈ R3+ :

0<W =G+B+E ≤
η

is a region of attraction for all solutions initiating in the interior of the positive
orthant, where η is a constant such that

K(r1 + η)2 L(r2 + δ1 + η)2

0 < η < δ2 , ν = + .
4r1 4r2

Proof: Let us consider W (t) = G(t) + B(t) + E(t) and η > 0 be a constant.
Then,

dW r1 G2 r2 B 2
+ ηW =(r1 + η)G − + (r2 + δ1 + η)B −
dt K L (6.2)
2
− (c1 + a)GB − c2 G B − eBE − (δ2 − η)E

Now, we choose η such that 0 < η < δ2 , then the equation (6.2) can be written

152
 6.3. Equilibrium Analysis

as

dW r1 G2 r2 B 2
+ ηW ≤ (r1 + η)G − + (r2 + δ1 + η)B −
dt K L
K(r1 + η)2 L(r2 + δ1 + η)2
= + = ν (say)
4r1 4r2

By using differential inequality (Birkhoff & Rota, 1982), we obtain

ν
0 < W (G(t), B(t), E(t)) ≤ (1 − e−ηt ) + (G(0), B(0), E(0))e−ηt
η

ν
which gives 0 < W (t) ≤ η
as t → ∞.

6.3 Equilibrium Analysis

The model system (6.1) developed to study the interaction between grass-
land and bad biomass has four equilibrium points, viz. P0 (0, 0, 0), P1 (K, 0, 0),
P2 (Ḡ, B̄, 0) and P3 (G∗ , B ∗ , E ∗ ).
Here, P2 is the equilibrium point in the absence of effort to control bad
biomass and P3 is the interior equilibrium point. The points P0 (0, 0, 0) and
P1 (K, 0, 0) always exist. The existence of the other points are discussed below.

Existence of P2 (Ḡ, B̄, 0):

In the absence of control effort E, Ḡ and B̄ can be obtained by solving the
following equations.

G

r1 1 − − c1 B − c2 GB = 0 (6.3)
K
B

r2 1 − − aG = 0 (6.4)
L

153
 6.3. Equilibrium Analysis

Equation (6.4) gives,

L
B= (r2 − aG) (6.5)
r2

which upon substitution in equation (6.3) yields a quadratic equation in G

given as,
B1 G2 + B2 G + B3 = 0 (6.6)

where,

aLc2
B1 =
r2
ac1 L r1
B2 = − − c2 L
r2 K
B3 =r1 − Lc1

√
−B2 + B22 −4B1 B3
Now, solving for the roots of the equation (6.6) gives Ḡ = 2B1
,
which is a positive real root if the following condition holds.

r1 < Lc1 (6.7)

Knowing the value of Ḡ , the value of B̄ is evaluated from equation (6.5). It

may be noted that for B̄ to be positive, the following must hold :

r2 > aḠ (6.8)

Existence of P3 (G∗ , B∗ , E∗ ):
To see the existence of P3 (G∗ , B ∗ , E ∗ ) , we note that G∗ , B ∗ and E ∗ are
the positive solutions of the following equations.

 G
r1 1 − − c1 B − c2 B = 0 (6.9)
K

154
 6.3. Equilibrium Analysis

 B
r2 1 − − eE − aG = 0 (6.10)
L
δ1 B − δ2 E = 0 (6.11)

Equation (6.10) implies

L 
B= r2 − eE − aG (6.12)
r2

using this in equations (6.9) & (6.11 ), we get

 G  c1 L   c GL 
2

f (G, E) = r1 1− − r2 −eE −aG − r2 −eE −aG = 0 (6.13)
K r2 r2

and
δ1 L  
g(G, E) = r2 − eE − aG − δ2 E = 0 (6.14)
r2

From equation (6.14) we note the following:

δ1 r2
i. If G = 0, then E = r δ = E1 where E1 is a solution of g(0, E) = 0
eδ1 + 2L 2

r2
ii. If E = 0, then G = a
= G1 where G1 is a solution of g(G, 0) = 0.

∂g
dG
 ∂g
iii. dE
= − ∂E ∂G
< 0, as

∂G eδ1 L
=− − δ2 < 0 and
∂E r2
∂g aδ1 L
=− <0
∂G r2

Equation (6.13) implies that,

r2 (Lc1 −r1 )
i. If G = 0, then E = ec1 L
= E2 where E2 is a solution of f (0, E) = 0.
Here, if Lc1 > r1 , then E > 0.

ii. If E = 0, we get G = G2 (say), where G2 is the positive root of f (G, 0) =

155
 6.4. Stability Analysis

0 which yeilds B1 G2 + B2 G + B3 = 0, where

Lc2 a
B1 =
r2
ac1 L r1
B2 = − − Lc2
r2 K
B3 =r1 − c1 L

Thus, G2 is positive if r1 < c1 L.

iii. From equation (6.13),

∂f Le(c1 + c2 G)
= > 0,
∂E r2
∂f r1 ac1 L c2 L   ac GL
2
=− + − r2 − eE − aG + and
∂G K r2 r2 r2
dG ∂f . ∂f ∂f ∂f
=− > 0 if and have opposite signs .
dE ∂E ∂G ∂E ∂G

If we also assume that , G2 > G1 , then the two isoclines (6.13) & (6.14)
intersect at a unique point (G∗ , E ∗ ), and B ∗ can be calculated from equation
(6.12). It may be noted that B ∗ is positive if the following inequality holds:

r2 > eE ∗ + aG∗ (6.15)

6.4 Stability Analysis

This section is devoted to discuss the local stability behavior of the system
(6.1).
Evaluating the Jacobian at each equilibrium point and using the eigenvalue
method & Routh Hurwitz criteria the following results are obtained.

i. P0 is unstable.

156
 6.4. Stability Analysis

ii. P1 is stable if aK < r2 otherwise unstable.

iii. P2 is locally asymptotically stable iff

r1 r2 ḠB̄ r2 c2
+ ḠB̄ 2 + ac2 Ḡ2 B̄ > aB̄c1 Ḡ
KL L

otherwise unstable.

iv. To study the stability of P3 (G∗ , B ∗ , E ∗ ) it is noted that the characteristic

equation of the Jacobian matrix evaluated at P3 is

λ3 + C1 λ2 + C2 λ + C3 = 0 (6.16)

where,

r2 B ∗ G∗ r1
C1 =δ2 + + + c2 G∗ B ∗
L K
δ2 r 2 B ∗ ∗
 r2 B ∗  G∗ r1 ∗ ∗

C2 = + eδ1 B + δ2 + + c2 G B (6.17)
L L K
+ aB ∗ (c1 G∗ − c2 G∗2 )
 G∗ r  δ r B ∗ 
1 ∗ ∗ 2 2
C3 = + c2 G B + eδ1 B + δ2 aB ∗ (c1 G∗ − c2 G∗2 )
∗
K L

By Routh Hurwitz criteria it implies that, all the solutions of (6.16) have
negative real parts iff,

Ci > 0, i = 1, 3 and C1 C2 > C3 (6.18)

c1
Clearly, C1 > 0, C3 > 0 iff G∗ < c2
. It is also easy to verify that
c1
C1 C2 > C3 holds true for G∗ < c2
.

Thus, we state the following theorem.

157
 6.5. Existence Hopf-bifurcation

Theorem 6.4.1 The positive equilibrium point is locally asymptotically

stable iff
c1
G∗ < (6.19)
c2

6.5 Existence Hopf-bifurcation

The aim of this section is to investigate the Hopf-bifurcation analysis of

(6.1) around the interior equilibrium point P3 . Here we have considered the
depreciation rate of effort δ2 as bifurcation parameter.
Choosing δ2 as the bifurcation parameter, then for δ2 = δ2∗ (critical value),
the necessary and sufficient condition for the Hopf-bifurcation to occur are:

1. C1 (δ2∗ ) > 0 and C3 (δ2∗ ) > 0

2. f (δ2∗ ) ≡ C1 (δ2∗ )C2 (δ2∗ ) − C3 (δ2∗ ) = 0 (6.20)

h dλ i
j
3. Re 6= 0, j = 1, 2, 3
dδ2 δ2 =δ2∗

where λj is the eigenvalue of the Jacobian variational matrix associated with

P3 (G∗ , B ∗ , E ∗ ).
At the critical value δ2 = δ2∗ , the characteristic equation (6.16) can be
written as
(λ2 + C2 )(λ + C1 ) = 0

This has three roots i.e. λ1,2 = ±iω, λ3 = µ, where,

µ = −C1
p (6.21)
ω = C2

where C1 and C2 are expressed in (6.17).

Thus at δ2 = δ2∗ , the characteristic equation (6.16) has purely imaginary

158
 6.5. Existence Hopf-bifurcation

roots while the third root is negative. To show the transversality condition,
let at any point δ2 in −neighbourhood of δ2∗ , λ1,2 = b1 (δ2 ) ± ib2 (δ2 )
Substituing this in charcteristic equation (6.16) and operating real amd
imaginary parts, and taking its derivative, we have,

R(δ2 )b01 (δ2 ) − S(δ2 )b02 (δ2 ) + L(δ2 ) = 0

S(δ2 )b01 (δ2 ) + R(δ2 )b02 (δ2 ) + N (δ2 ) = 0

where

R(δ2 ) =3{b21 (δ2 ) − b22 (δ2 )} + 2C1 b1 (δ2 ) + C2

S(δ2 ) =6b1 (δ2 )b2 (δ2 ) + 2C1 b2 (δ2 )

L(δ2 ) =C10 (b21 (δ2 ) − b22 (δ2 ) + C20 b1 (δ2 ) + C30

N (δ2 ) =2C10 b1 (δ2 )b2 (δ2 ) + C20 b2 (δ2 )

h i h i
dλj LR+SN
Hence, Re dδ2
=− R2 +S 2
6= 0.
δ2 =δ2∗ δ2 =δ2∗
Therefore the transversality condition hold and Hopf-bifurcation occurs at
δ2 = δ2∗ around the interior equilibrium P3 (G∗ , B ∗ , E ∗ ). Thus we can write
the following theorem:

Theorem 6.5.1 There is a simple Hopf-bifurcation at equilibrium point P3

under conditions (6.20), at some critical value of the parameter δ2 ), given by
tthe equation f (δ2∗ ) = 0.

159
 6.6. Numerical Simulation

6.6 Numerical Simulation

To validate the feasibility of the analytical results obtained in previous

section the model is numerically simulated with the following parameter values:

r1 = 2, K = 100, c1 = 0.07, c2 = 0.02, r2 = 3,

(6.22)
L = 1000, e = 0.4, a = 0.00001, δ1 = 0.191, δ2 = 0.2

with initial conditions (1, 1, 1). For the above set of values of param-

(a) (b)
80
G
B
E
70
10

60 9

50 7

6
G, B, E

40 5

3
30
2

1
20
30
25
80
20
10 60
15
10 40

5 20
0
0 50 100 150 B 0 0 G
Time

Fig. 6.1. (a) Local asymptotic stability of interior equilibrium point and (b)
Limit cycle

eters, conditions for the existence of interior equilibrium P3 (G∗ , B ∗ , E ∗ )

is satisfied. Thus positive equilibrium point P3 (G∗ , B ∗ , E ∗ ) is given by
G∗ = 16.7692, B ∗ = 4.1065, E ∗ = 5.9750. Figure (6.1(a)) shows local asymp-
totic stability behavior of the interior equilibrium point P3 and figure (6.1(b))
shows the existence of a limit cycle for the system.

160
 6.6. Numerical Simulation

Next, for another set of parameter values

r1 = 2, K = 100, c1 = 0.07, c2 = 0.02, r2 = 5,

(6.23)
L = 1000, e = 0.16, a = 0.00001, δ1 = 0.26,

and different values of depletion rate δ2 are taken. The trajectories of G, B

and E are shown in igures (6.2 (a)-(d)). From these figures it is interpreted
that the density of grassland biomass settles at higher equilibrium level than
bad biomass if depreciation rate is lesser than the increase of given effort.
Now, to verify the theorem (6.5.1), the depletion rate has been considered as

(a) (b)
250 250
G G
200 B 200 B
E E
150
G, B, E

G, B, E

150
100
100
50

0 50

−50 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Time Time

(c) (d)
250 250
G G
200 B 200 B
E E
G, B, E

G, B, E

150 150

100 100

50 50

0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Time Time

Fig. 6.2. Graph trajectories of grassland biomass and bad biomass in presence
of effort for (a) δ2 = 0.02, (b) δ2 = 0.03, (c) δ2 = 0.04 and (d) δ2 = 0.06

a bifurcation parameter and other parameter values are taken to be

r1 = 2, K = 100, c1 = 0.07, c2 = 0.02, r2 = 3,

(6.24)
L = 1000, e = 0.4, a = 0.00001, δ1 = 0.0991,

The critical point of effort depletion rate δ2 = δ2∗ = .06. Thus, for these set

161
 6.6. Numerical Simulation

120
G
B
100 E

80
G, B, E

60

40

20

0
0 10 20 30 40 50 60 70 80
Time

Fig. 6.3. Existence of Hopf bifurcation for the model system for bifurcation
parameter δ2 = 0.0891

(a) (b)
120 120
G G
B B
E E
100 100

80 80
G, B, E

G, B, E

60 60

40 40

20 20

0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Time Time

Fig. 6.4. Variation of grassland biomass and bad biomass with time for
different values of δ2 > δ1 , (a) δ2 = 0.3 and (b) δ2 = 0.55

of parameters and for δ2 = .0891 > .06 = δ2∗ , Hopf oscillations is observed as
shown in Fig (6.3).
Also, for the same set of parameters when δ2 exceeds the bifurcation point,
the behaviour of the system is shown in figures (6.4 (a)-(b)). It is observed

162
 6.7. Conclusion

that as the depletion rate increases the density of bad biomass increases and
grassland biomass decreases.

6.7 Conclusion

In this chapter a mathematical model for the depletion of grassland biomass

under the effect of invasive species as a catastrophe has been presented and
analyzed. This model explains how the bad biomass growth affect the growth
of grassland biomass. The growth of grassland and bad biomass follows the
logistic growth model. Eigenvalue method and Routh Hurwitz criteria have
been applied to discuss the existence of equilibrium points and local stability.
It was observed that the conservative effort could play a vital role in controlling
the undesirable growth of bad biomass. Also, it was found that the bad biomass
increases and grassland decreases if depletion rate of effort applied goes below
a critical value. From analysis of the proposed model, it is found that the
grassland biomass may be conserved by applying suitable effort and the results
are verified through numerical simulation. This model suggest that the control
of bad biomass in an efficient manner is important to maintain the grassland
biomass of the park in balanced condition.

163