You are on page 1of 57

Computational Ecology

and Software
Vol. 4, No. 2, 1 June 2014

International Academy of Ecology and Environmental Sciences


Computational Ecology and Software
ISSN 2220-721X
Volume 4, Number 2, 1 June 2014

Editor-in-Chief
WenJun Zhang
Sun Yat-sen University, China
International Academy of Ecology and Environmental Sciences, Hong Kong
E-mail: zhwj@mail.sysu.edu.cn, wjzhang@iaees.org

Editorial Board
Ronaldo Angelini (The Federal University of Rio Grande do Norte, Brazil)
Andre Bianconi (Sao Paulo State University (Unesp), Brazil)
Bin Chen (Beijing Normal University, China)
Daniela Cianelli (University of Naples Parthenope, Italy)
Alessandro Ferrarini (University of Parma, Italy)
Yanbo Huang (USDA-ARS Crop Production Systems Research Unit, USA)
Istvan Karsai (East Tennessee State University, USA)
Vladimir Krivtsov (Heriot-Watt University, UK)
Lev V. Nedorezov (University of Nova Gorica, Slovenia)
Fivos Papadimitriou (Environmental and Land Use Consultancies, Greece)
George P. Petropoulos (Institute of Applied and Computational Mathematics, Greece)
Vikas Rai (Jazan University, Saudi Arabia)
Santanu Ray (Visva Bharati University, India)
Kalle Remm (University of Tartu, Estonia)
Rick Stafford (University of Bedfordshire, UK)
Luciano Telesca (Institute of Methodologies for Environmental Analysis, Italy)
Bulent Tutmez (Inonu University, Turkey)
Ranjit Kumar Upadhyay (Indian School of Mines, India)
Ezio Venturino (Universita’ di Torino, Italy)
Michael John Watts (The University of Adelaide, Australia)
Peter A. Whigham (University of Otago, New Zealand)
ZhiGuo Zhang (Sun Yat-sen University, China)

Editorial Office: ces@iaees.org

Publisher: International Academy of Ecology and Environmental Sciences


Address: Flat C, 23/F, Lucky Plaza, 315-321 Lockhart Road, Wanchai, Hong Kong
Tel: 00852-6555 7188
Fax: 00852-3177 9906
Website: http://www.iaees.org/
E-mail: office@iaees.org
Computational Ecology and Software, 2014, 4(2): 82-88

Article

Spatial risk assessment of alien plants in China using biodiversity


resistance theory

YouHua Chen
Department of Renewable Resources, University of Alberta, Edmonton, T6G 2H1, Canada
E-mail: haydi@126.com

Received 24 February 2014; Accepted 27 March 2014; Published online 1 June 2014

Abstract
In the present study, the potential occurrence risk of invasive plants across different provinces of China is
studied using disease risk mapping techniques (empirical Bayesian smoothing and Poisson-Gamma model).
The biodiversity resistance theory which predicts that high-biodiversity areas will have reduced risk of species
invasion serves as the base for performing spatial risk assessment of plant invasion across provinces. The
results show that, both risk mapping methods identified that north-eastern part of China have the highest
relative risk of plant invasion. In contrast, south-western and south-eastern parts of China, which have high
woody plant richness, are predicted to possess low relative risks of plant invasion. Through spatial regression
analysis (simultaneous autoregression model), nine environmental variables representing energy availability,
water availability, seasonality, and habitat heterogeneity are used to explain the relative risk of plant invasion
across provinces of China. The fitting results suggest that, PRECrange and TEMrange are the most two
important covariates correlated with the occurrence risks of alien plants at provincial level in China. As
indicated by Moran’s I index, spatial regression analysis can effectively eliminate the potential biases caused
by spatial autocorrelation.

Keywords spatial ecology; risk assessment; invasion biology; plant distribution.

Computational Ecology and Software     
ISSN 2220­721X   
URL: http://www.iaees.org/publications/journals/ces/online­version.asp 
RSS: http://www.iaees.org/publications/journals/ces/rss.xml 
E­mail: ces@iaees.org 
Editor­in­Chief: WenJun Zhang 
Publisher: International Academy of Ecology and Environmental Sciences 

1 Introduction
Alien plants may cause severe ecological problems onto local ecosystems because they can better compete for
the resources than native species (Zhang and Chen, 2011). Biodiversity hotspots are believed to be more
resistant to species invasion: more diverse communities would have less species invasions (Law and Morton,
1996; Stachowicz et al., 2002; Zhang, 2011, 2012a). This biodiversity resistance theory can be an option for
performing risk assessment of species invasions.
Modeling plant invasion risk in China has been done in some previous studies recently (Bai et al., 2013;
Liu et al., 2005; Wu et al., 2006; Feng and Zhu, 2010). However, all these studies only rely on the richness

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 82-88 83

information of alien plants, the dependence of alien plant diversity and native plant diversity has never been
considered yet. As such, one may utilize the biodiversity resistance theory to quantify relative risk of alien
plants by explicitly incorporating the information of native plant diversity.
In the present study, we perform the relative risk assessment of alien plants in China by using spatial
disease mapping methods (including empirical Bayesian smoothing and Poisson-Gamma model) so as to
incorporate richness information of native woody plant species.

2 Materials and Methods


2.1 Invasive plant diversity data
The province-level 127 alien plant richness data are collected from previous studies (Chen, 2013; Wang et al.,
2009; Wu et al., 2006; Yan et al., 2012). The resultant data include provincial distributional information and
physiological trait patterns of each invasive plant found in China.
2.2 Woody plant diversity data
The population size at risk when performing risk mapping is required as an entry. In the present study, we
regard the population size at risk as the number of native woody plant species. As such, similar to the disease
transmission model, areas where there are higher population sizes would have less relative risk of disease
outbreak. In our modeling of relative risk estimation of plant invasion, we hold the recognition that areas with
high biological diversity should have less plant invasion risk (Law and Morton, 1996; Stachowicz et al., 2002).
2.3 Risk mapping using empirical Bayesian smoothing and Poisson-Gamma model
We use both empirical Bayesian smoothing and Poisson-Gamma model for predicting the outbreak risk of
invasive plants over the provinces of China. For testing whether the relative risk of plant invasion is
significantly higher than 1, we utilize Poisson exact test.
2.4 Identification of important covariates associated with the occurrence risk of alien plants
We use generalized linear models to identify important covariates which are closely related to the occurrence
risk of alien plants at provincial level in China. The fitting model is given by,

log( )  X   

where  is the relative risk of alien plants across various provinces of China calculated as above;  is the
associated coefficients for the covariates presented in terms of columns in the covariate matrix X.
We standardize all the covariates in X so that the estimated coefficients  are standardized partial
coefficients. We then choose those covariates with high values of  as the important ones for explaining the
occurrence risk of alien plants at provincial level in China.
However, the above model is spatially implicit, for which it can not be able to partial out the influence of
spatial autocorrelation and may inflate Type I errors. As such, we implemented the alternative model which
explicitly incorporate the spatial influence in the model as,

log( )  X   W log( )  

where W is the matrix with each element wij =1 if provinces i and j are adjacent (i.e., they share some
boundaries), otherwise wij =0.  measures the strength of spatial autocorrelation for the response variable
(log-transformed relative risk). This is the formula of spatial simultaneous autoregression lag model (SAR)
(Dormann et al., 2007; Dormann, 2007). Implication of SAR model has been done using R (R Development
Core Team, 2011) package “spdep” (Bivand et al., 2013).
The following environmental variables at provincial level are used for fitting the above model (Qian, 2013):
mean annual temperature (TEM, °C), mean temperature of the coldest month (TEMmin, °C), mean

IAEES www.iaees.org
84 Computational Ecology and Software, 2014, 4(2): 82-88

temperature of the warmest month (TEMmax, °C), annual precipitation (PREC, mm), annual actual
evapotranspiration (AET, mm), annual potential evapotranspiration (PET, mm), the range of mean annual
temperature within a province (TEMrange, °C), the range of annual precipitation within a province
(PRECrange, mm), and the range of elevation within a province (ELEVrange, m). These nine environmental
variables represent energy availability, water availability, seasonality, and habitat heterogeneity (Qian, 2013).
All these covariates are standardized before performing SAR modeling.

3 Results
As showed in Fig. 1, the woody plant diversity is highest in southern part of China, which was identified as
one of the global biodiversity hotspots of the world (Myers et al., 2000; Chen and Bi, 2007). Yunnan has the
highest woody plant richness, followed by Guangxi Province.
50
40
30
20
10
0

80 90 100 110 120 130

Fig. 1 Observed woody plant diversity at provincial level of China. Colors from light to dark grey indicate diversity from low to
high.
50
40
30
20
10
0

80 90 100 110 120 130

Fig. 2 Observed alien plant diversity at provincial level of China. Colors from light to dark grey indicate diversity from low to
high.

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 82-88 85

As shown in Fig. 2, the alien plant diversity is highest in southern and eastern part of China. Interestingly,
the biodiversity hotspot, Yunnan Province, also has very high alien plant diversity following Jiangsu Province
(Wang et al., 2009).

50
40
30
20
10

80 90 100 110 120 130

Fig. 3 Relative risk of plant invasiveness at provincial level of China using empirical Bayesian smoothing method. Provinces
which have red boundaries indicate that their relative risks of plant invasion are significantly higher than 1. Colors from light to
dark grey indicate invasive risks from low to high.
50
40
30
20
10

80 90 100 110 120 130

Fig. 4 Relative risk of plant invasiveness at provincial level of China using Poisson-Gamma model. Provinces which have red
boundaries indicate that their relative risks of plant invasion are significantly higher than 1. Colors from light to dark grey
indicate invasive risks from low to high.

IAEES www.iaees.org
86 Computational Ecology and Software, 2014, 4(2): 82-88

The relative risk of plant invasion based on empirical Bayesian smoothing and Poisson-Gamma models are
highly similar (Figs. 3-4). As seen, northern and western parts of China have high risk of plant invasion (dark
grey colors). In contrast, southern part of China is less sensitive to plant invasion (light grey colors). Using
Poisson exact test, it is identified that 13 provinces from northern part of China have significant relative risk
values higher 1 (those provinces with red boundaries in Figs. 3-4).
As indicated by Table 1, the SAR model predicted that PRECrange was the most significant covariate
(P<0.05) predicting the relative risk of plant invasion across different provinces of China. The next two
important variables are TEMrange and PET respectively, but their coefficients are marginally significant
(P<0.1). The coefficient  for quantifying the strength of spatial autocorrelation is equal to 0.7903, which is a
quite large value given that the interval for  is [0, 1].
Before applying SAR model, there is significant spatial autocorrelation for the relative risk of plant
invasion with Moran’s I=0.639 (P<0.001) (Table 1 and Figs. 3-4). However, after SAR model has been done,
the spatial autocorrelation has been removed to a great extent as indicated by the non-significant Moran’s I
value for the residuals of the model (Moran’s I=0.131, P>0.05).

Table 1 Coefficient estimation for the covariates in the SAR model used for predicting relative risk of plant invasion across
different provinces of China. Here the relative risk values are from empirical Bayesian smoothing method, the result for Poisson-
Gamma method is almost identical and thus not presented here.
Covariates Estimate Std. Error z value Pr(>|z|)
(Intercept) 0.083 0.049 1.695 0.09
Area 0.119 0.102 1.171 0.242
Elev 0.527 0.883 0.597 0.55
TEM -0.496 1.219 -0.407 0.684
TEMmin -0.721 0.755 -0.955 0.339
TEMmax 0.505 0.669 0.755 0.45
PET 0.81 0.486 1.668 0.095
PREC -0.384 0.375 -1.026 0.305
AET 0.341 0.376 0.907 0.364
ElevRange -0.21 0.789 -0.267 0.79
TEMrange -0.267 0.151 -1.768 0.077
PRECrange -0.214 0.076 -2.827 0.005

Moran’s I P value
Before SAR 0.639 0.001
After SAR 0.131 0.717

4 Discussion
Different from previous studies, which showed that areas with high relative outbreak risk of alien plants are
typically those with high richness of alien plants, the present study found that those areas with low native plant
diversity would have higher risk of alien plant occurrence. The remarkable difference on the risk assessment is
due to the inclusion of biodiversity information from native woody plant species. As mentioned above, all the
previous studies (Bai et al., 2013; Liu et al., 2005; Wu et al., 2006; Feng and Zhu, 2010) did not explicitly
consider the interaction between native and alien plant diversity. The biodiversity resistance hypothesis (Law

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 82-88 87

and Morton, 1996; Stachowicz et al., 2002) provides a way to effectively incorporate diversity information
from native plants for quantifying invasive risk of alien plants over provinces.
Based on the comparative results (Figs 3-4), both Poisson-Gamma and empirical Bayesian smoothing
methods consistently identified that provinces from northern part of China are remarkably higher than those
provinces in southern part of China. This is expected, since we applied biodiversity resistance hypothesis when
modeling the invasive risk of alien plants. In southern and south-western part of China, the provinces there
have very high species richness in comparison to those provinces in northern part of China. In particular,
Yunnan province constitutes a well recognized biodiversity hotspot of the world (Chen and Bi, 2007; Myers et
al., 2000; Zhang et al., 2010; Hua, 2008; Zhang, 2012b). Thus, even though this province is found to have a
very high number of alien plants (Fig. 2), its relative risk of alien plant invasion is still comparatively low
(Figs. 3-4). The very reason is that the calculation of relative risk of alien plant invasion requires the
population at risk in the denominator, for which we use the native woody plant richness as the surrogate.
Yunnan Province has the highest woody plant diversity (Fig. 1), thus its relative risk is predicted to be low
accordingly.

Acknowledgements
This work is supported by China Scholarship Council.

References
Bai F, Chisholm R, Sang W, Dong M. 2013. Spatial risk assessment of alien invasive plants in China.
Environmental Science & Technology, 47: 7624-7632
Bivand R, Altman M, Anselin L . 2013. spdep: spatial dependence: weighting schemes, statistics and models.
http://CRAN.R-project.org/package=spdep
Chen Y. 2013. Distributional patterns of alien plants in China: the relative importance of phylogenetic history
and functional attributes. ISRN Ecology, 527052
Chen Y, Bi J. 2007. Biogeography and hotspots of amphibian species of China: Implications to reserve
selection and conservation. Current Science, 92: 480-489
Dormann C. 2007. Effects of incorporating spatial autocoorrelation into the analysis of species distribution
data. Global Ecology and Biogeography, 16: 129-138
Dormann C, McPherson J, Araujo M, et al. 2007. Methods to account for spatial autocorrelation in the analysis
of species distirbutional data: a review. Ecography (Cop), 30: 609-628
Feng J, Zhu Y. 2010. Alien invasive plants in China: risk assessment and spatial patterns. Biodiversity and
Conservation, 19: 3489-3497
Hua Z. 2008. Advances in biogeography of the tropical rain forest in southern Yunnan, southwestern China.
Tropical Conservation Science, 1: 34-42
Law R, Morton R. 1996. Permanence and the assembly of ecological communities. Ecology, 77: 762-775
Liu J, Liang S, Liu F, et al. 2005. Invasive alien plant species in China: regional distribution patterns. Diversity
and Distributions, 11: 341-347
Myers N, Mittermeier R, Mittermeier C, et al. 2000. Biodiversity hotspots for conservation priorities. Nature,
403: 853-858
Qian H. 2013. Environmental determinants of woody plant diversity at a regional scale in China. PLoS One, 8:
e75832

IAEES www.iaees.org
88 Computational Ecology and Software, 2014, 4(2): 82-88

R Development Core Team. 2011. R: A Language and Environment for Statistical Computing. Vienna, Austria
Stachowicz J, Fried H, Osman R, Whitlatch R. 2002. Biodiversity, invasion resistance, and marine ecosystem
function: reconciling pattern and process. Ecology, 83: 2575-2590
Wang Z, Chen Y, Chen Y. 2009. Functional grouping and establishment of distribution patterns of invasive
plants in China using self-organizing maps and indicator species analysis. Archives of Biological Science,
61: 71-78
Wu X, Luo J, Chen J, Li B. 2006. Spatial patterns of invasive alien plants in China and its relationship with
environmental and anthropological factors. Journal of Plant Ecology, 30: 576-584
Yan X, Shou H, Ma J. 2012. The problem and status of the alien invasive plants in China. Plant Diversity and
Resources, 34: 287-313
Zhang D, Chen M, Murphy R, et al. 2010. Genealogy and palaeodrainage basins in Yunnan Province:
phylogeography of the Yunnan spiny frog, Nanorana yunnanensis (Dicroglossidae). Molecular Ecology,
19: 3406-3420
Zhang WJ. 2011. Constructing ecological interaction networks by correlation analysis: hints from community
sampling. Network Biology, 1(2): 81-98
Zhang WJ. 2012a. Computational Ecology: Graphs, Networks and Agent-based Modeling. World Scientific,
Singapore
Zhang WJ. 2012b. Did Eucalyptus contribute to environment degradation? Implications from a dispute on
causes of severe drought in Yunnan and Guizhou, China. Environmental Skeptics and Critics, 1(2): 34-38
Zhang WJ, Chen B. 2011. Environment patterns and influential factors of biological invasions: a worldwide
survey. Proceedings of the International Academy of Ecology and Environmental Sciences, 1(1): 1-14

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 89-103

Article

Stability analysis of a discrete ecological model

Q. Din1, E. M. Elsayed2,3
1
Department of Mathematics, Faculty of Basic & Applied Sciences, University of PoonchRawalakot, Pakistan
2
Department of Mathematics, Faculty of Science,King Abdulaziz University, P.O. Box 80203, Jeddah 21589, Saudi Arabia
3
Department of Mathematics, Faculty of Science, Mansoura University, Mansoura 35516, Egypt
E-mail: qamar.sms@gmail.com,emmelsayed@yahoo.com

Received 14 January 2014; Accepted 20 February 2014; Published online 1 June 2014

Abstract
In this paper, we study the qualitative behavior of following discrete-time population model:
, ,
where parameters , , , , , and initial conditions , , , are positive real numbers. More
precisely, we investigate the existence and uniqueness of positive equilibrium point,boundedness character,
persistence, local asymptotic stability, global behavior and rate of convergence of unique positive equilibrium
point of this model. Some numerical examples are given to verify our theoretical results.

Keywords population models; difference equations; steady-states; boundedness; local and global character.

Computational Ecology and Software 
ISSN 2220­721X 
URL: http://www.iaees.org/publications/journals/ces/online­version.asp 
RSS: http://www.iaees.org/publications/journals/ces/rss.xml 
E­mail: ces@iaees.org 
Editor­in­Chief: WenJun Zhang 
Publisher: International Academy of Ecology and Environmental Sciences 

1 Introduction
In population dynamics, difference and differential equations are used in many models (Nedorezov, 2012;
Nedorezov and Sadykov, 2012; Elena et al., 2013). Exponential difference equations have many applications
in population dynamics. One can see the references (El-Metwally et al., 2001; Papaschinopoulos et al., 2011;
Papaschinopoulos et al., 2012) for some interesting results related to qualitative behavior of population models.
From Zhou and Zou (2003) and Liu (2010) it is clear that difference equations are much better as compared to
differential equations, when the populations are of non-overlapping generations. In literature, there are many
papers in which discrete dynamical systems are used to study the qualitative behavior of population models
(Ahmad, 1993; Zhou and Zou, 2003; Tang andZou, 2006; Din, 2013; Din and. Donchev, 2013; Din et al., 2013;
Din, 2014). In case of discrete–time models one can compute mathematical results more efficiently. There are
many research papers which are related to mathematical models in population dynamics. During the last few
decades, many researchers are attracted by such computational models. For detail of biological models see
(Edelstein-Keshet, 1988; Brauer and Castillo-Chavez, 2000; Allen, 2007). Nonlinear difference equations are

IAEES www.iaees.org
90 Computational Ecology and Software, 2014, 4(2): 89-103

very important in application. Naturally, such equations appear as discrete approximation of differential and
delay differential equations and these models are related to applied sciences such as ecology,biology, physics,
chemistry, physiology, engineering, bio-chemistry and economics. We cannot find the solutions of nonlinear
difference equations in all cases. So, one can study the behavior of solutions by local asymptotic stability of
equilibrium points. El-Metwally et al. (2001) studied the qualitative behavior of following population
model:   . Papachinopoulous et al. (2011) investigated the dynamics of following
population model: , . Recently, Papachinopoulous et al. (2012)
study following two exponential nonlinear difference equations:    ,
, and  , .
Our aim in this paper is to investigate existence and uniqueness of positive equilibrium point, boundedness
character, persistence, local asymptotic stability, global behavior of unique positive equilibrium point of
following four-dimensional discrete-time population model:
, ,                                            1
where parameters , , , , , and initial conditions , , , are positive real numbers. System (1) is
an extension of (Papachinopoulouset al., 2011).

2 Boundedness and Persistence


The following Theorem shows that every positive solution of (1) is bounded and persists.
⁄ ⁄
Theorem 2.1 Assume that 4 2 , 4 2 , then every
positive solution {( , } of system (1) is bounded and persists.

Proof. Consider the following second-order difference equations:


, . (2.1)
The solutions of (2.1) are given by


2 4
1


2 4 ,

⁄ ⁄
2 4 2 4 ,

where  , , , depend on initial values , , , .


⁄ ⁄
Assume that 4 2 , 4 2 . Then it follows that:

and  for all 1,2, . Evidently, the sequences

and are bounded for all 1,2, . Then, by comparison we obtain

 and  for all 1,2, . Hence we have

 and  for all 1,2, . This completes the proof.

3 Existence of Invariant Set for Solutions


Theorem 3.1 Let , be a positive solution of system of (1). Then ,

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 89-103 91

, , is invariant set for system (1).

Proof. Let , be a solution of system (1) with initial conditions ,   , and

, , . Then, it follows from system (1) that

  ,
1 1 1
and

  .
1 1 1
Hence, and . Suppose that result is true for 1, i.e., and . Then, for
1 from system (1) one has:

,
1 1 1

     
1 1 1
Hence, the proof follows by induction.

4 Existence and Uniqueness of Positive Equilibrium and Local Stability


⁄ ⁄
Theorem 4.1 Assume that 4 2 , 4 2 ,0 , 1 and

exp 1 exp .

Then the system (1) has a unique positive equilibrium point , such that , and

, .

Proof. Consider the following system of equations:


,

Then, one has and . Taking , where and

. Then, it follows that

. Now, 0 if and only if 1 0, . ., 1 exp .

Hence, it follows that 0 if and only if 1 exp . 

Furthermore, we have

F ,

IAEES www.iaees.org
92 Computational Ecology and Software, 2014, 4(2): 89-103

where . It is easy to see that 0 if and only if

, i.e., exp .

It follows that 0, if and only if  exp .

Hence, has positive solution in .


Moreover, we have that

1
1

1 0.

Hence, has a unique positive solution in . The proof is therefore completed.


Consider a fourth-dimensional discrete dynamical system of the form:
, , ,
(4.1)
, , , , 0,1, ,
where  : and : are continuously differentiable functions, and , are real intervals.
Furthermore, a solution , of system (4.1) is uniquely determined by initial conditions ,
 for 1,0 . Along with system (4.1) we consider a vector map of the form , , , . A constant
solution or equilibrium point (fixed-point) of (4.1) is point that satisfies:
, , , ,
, , , .
The point , is also point called a fixed point of the vector map .
Definition 4.1 Assume that  , be aconstant solution (steady-state) of (4.1).
1. A constant solution , is called stable, if for an arbitrary  0 there exists a positive such that
for any initial condition , , 1, 0 if ∑ , , , then one has   ,
, for all 0, where · denotes usual Euclidean norm in .
2. A constant solution , is called unstable if it did not satisfy stability condition.
3. A constant solution , is locally asymptotically stable, if there exists a positive  such that
 
∑   , , and , , as ∞.
4. A constant solution( , ) is said to be global attracter if , , as ∞.
5. A constant solution ( , ) is said to be asymptotically global attractor if it is global attractor and stable.
Assume that , be a fixed-point of map defined by = , , , , where and are continuously
differentiable functions about ( , . The linearized system of (4.1)about the fixed-point ( , ) is given by:
 ,

where  and is a Jacobain matrix of (4.1) about the fixed-point , .

Lemma 4.1 (Sedaghat, 2003) Consider the discrete dynamical system of the form , 0,1, ,
such that be an equilibrium point of (4.1). Then following statements are true:
(i) If all eigenvalues of Jacobian  at lie inside an open unit disc | | 1, then the fixed-point is

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 89-103 93

locally asymptotically stable.


(ii) If one of the eigenvalue of Jacobian matrix has an absolute value greater than one, then is unstable.
To construct corresponding linearized form of system (1) we consider the following transformation:
, , ,, , , , , (4.2)
where , , and . The Jacobian matrix about the fixed point , of (1)
under the transformation (4.2) is given by
 0
,   0        .
1 0 0      0
0 1 0      0
Theorem 4.2The unique positive equilibrium point , of system (1) is locally asymptotically stable if the
following condition is satisfied:

1 1 1 1. (4.3)

Proof. The characteristic polynomial of Jacobain matrix , about , is given by


, (4.4)

where and .
Let  and .
Assume that condition (4.3) is satisfied and | | 1, then one has
| | | |
| |
 

 
1 1

1 1.

Then, by Rouche’s theorem and have same number of zeroes in an open unit disc
| | 1. Hence, all the roots of (4.3) satisfies | | 1, and it follows from Lemma 4.1 that the unique positive
equilibrium point , of system (1) is locally asymptotically stable.

5 Global Stability
Theorem 5.1 The unique positive equilibrium point , of system (1) is globally asymptotically stable.
Proof. Let , be an arbitrary positive solution of system (1) and let 0 Ι lim ∞ ,0 Ι
lim ∞ ,S lim ∞ ∞,S lim ∞ ∞. Then, from (1) one has:
Ι Ι Ι
Ι Ι Ι
Ι
S S S
Ι
S S S
Furthermore,  

Ι ,
1

Ι ,
1

IAEES www.iaees.org
94 Computational Ecology and Software, 2014, 4(2): 89-103

S ,
1 Ι

S .
1 Ι

Then, it follows that:

Ι ,
1
S
Ι S ,
1
Ι
Ι S , 
1 Ι

Ι
Ι S .
1 Ι

Ι S Ι
Hence one has Ι and Ι . Now consider the following functions:

, Ι, y ϵ J.
1 1
Moreover, for every , Ι :

0, 
1 1

0.
1 1
Hence, Ι and Ι .

6 Existence of Unbounded Solutions


Theorem 6.1 Let , 0,1 and {( , be a positive solution of system (3.1) Then following are true:

If  ,   ∞,   ∞.

(ii) If  ,   , ∞    ∞.

Proof. Let . Choosing the initial conditions , , , for the system (1) such that

, and , with , . Then from system (1) one has:

,
and
.
Suppose that result is true for 1, i.e., and . Then for 1 from system (1)
one has:
,
and
.
Hence, , for all 1,2, . Furthermore,

. Let with initial conditions , . Then,

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 89-103 95

2 4 2 4 . Hence by comparison . But ∞ ∞, so


lim ∞ ∞. Moreover, . Hence, lim ∞
.
The proof of is similar, therefore it is omitted.

7 Rate of Convergence
To discuss the rate of convergence of discrete dynamical system, we consider the following system:
, (7.1)
where  is a -dimensional vector, is a constant matrix, and : is a matrix function
such that:
0, (7.2)
as ∞, where . denotes a matrix norm which associated with the vector norm of the form:
, √ .
Lemma 7.1 (Perron’s theorem) (Pituk, 2002) Assume that the condition (7.2) holds true. If be a solution
of (7.1) then either 0 for all ∞, or

lim

is defined and it is equal to the absolute value of one of the eigenvalues of matrix .
Lemma 7.2 (Pituk, 2002) Assume that condition (7.2) holds. If is a solution of (7.1), then either 0
for all ∞, or

lim ,

is defined and is equal to the absolute value of one the eigenvalues of matrix .
Let , be any solution of the system (1) such that lim and lim , where
∞ ∞

,  and , . To find the error terms, one has from the system (1)

and

   

Let , and , then one has

and

where

IAEES www.iaees.org
96 Computational Ecology and Software, 2014, 4(2): 89-103

, , , 0, , ,

0,     . 

Moreover,

, , , 0, , 
∞ ∞ ∞ ∞ ∞

∞ , ∞ 0, ∞ . 

Now the limiting system of error terms can be written as

 0
  0        , 
1 0 0      0
0 1 0      0

which is similar to linearized system of (1) about the equilibrium point  , .

Using Lemma 7.1 one has the following result.

Theorem 7.1 Assume that , be a solution of (1) such that lim ∞ , lim ∞ , where

, be unique equilibrium point of (1), then the error vector of every solution of system (1)

satisfies both of the following asymptotic relations:

lim ∞ , , , , ,

lim , , , , ,

where  , , , , are the characteristic roots of the Jacobian matrix , .

8 Examples

In this section, we consider the following numerical examples.

Example 8.1 Let  1.1, 0.01, 2.9, 1.2, 0.02, 2.304.Then, system (1) can be written

as

1.1 0.01 2.9 , 1.2 0.02 2.9 (8.1)

with initial conditions 1.3, 1.2, 1.3, 1.5.

In this case the unique positive equilibrium point of the system (8.1)is given by

, 7.93981,1.22552 . Moreover, in Fig. 8.1 the plot of is shown in Fig. 8.1a, the plot of is

shown in Fig. 8.1b, and an attractor of the system is shown in Fig. 8.1c.

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 89-103 97

(8.1a) Plot of for the system (8.1)

(8.1b) Plot of for the system (8.1)

IAEES www.iaees.org
98 Computational Ecology and Software, 2014, 4(2): 89-103

(8.1c) An attractor of the system (8.1)


Fig. 8.1 Plots for the system (8.1).

Example 8.2 Let  1.5, 0.026, 3.14, 1.67, 0.0096, 2.6. Then, system (1) can be
written as
1.5 0.026 3.14 ,
1.67 0.0096 2.6     (8.2)
with initial conditions 2.2, 2.1, 2.35, 2.25.
In this case the unique positive equilibrium point of the system (8.2) is given by
, 2.16435,2.41379 .  Moreover, in Fig. 8.2 the plot of is shown in Fig. 8.2a, the plot of is
shown in Fig. 8.2b, and an attractor of the system is shown in Fig. 8.2c.

(8.2a) Plot of for the system (8.2)

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 89-103 99

(8.2b) Plot of for the system (8.2)

(8.2c) An attractor of the system (8.2)


Fig.8.2 Plots for the system (8.2)

Example 8.3 Let  1.5, 0.0001, 3.5, 1.67, 0.00002, 2.77. Then, system (1) can be
written as
2.3 0.0001 3.5
                                                                 1.67 0.00002 2.77     (8.3)
with initial conditions 2.25, 2.15, 2.36, 2.27.  In this case the unique positive
equilibrium point of the system (8.3) is given by , 1.71494,3.33035 .

IAEES www.iaees.org
100 Computational Ecology and Software, 2014, 4(2): 89-103

Moreover, in Fig. 8.3 the plot of is shown in Fig. 8.3a, the plot of is shown in Fig. 8.3b, and an
attractor of the system is shown in Fig. 8.3c.

(8.3a) Plot of for the system (8.3)

(8.3b) Plot of for the system (8.3)

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 89-103 101

(8.3c) An attractor of the system (8.3)


Fig. 8.3 Plots for the system (8.3).

Example 8.4 Let 2.3, 0.0009, 4.5, 2.09, 0.0008, 6.35. Then system (1) can be
written as:
2.3 0.0009 4.5 ,
                                                                 2.09 0.0008 6.35     (8.4)
with initial conditions 2.1, 2.2, 1.1, 1.5. In this case the unique positive equilibrium
point of the system (8.4) is given by , 2.3581,5.24482 . Moreover, in Fig. 8.4 the plot of is
shown in Fig. 8.4a,the plot of is shown in Fig. 8.4b,and an attractor of the system (8.4) is shown in Fig.
8.4c.

(8.4a) Plot of for the system (8.4)

IAEES www.iaees.org
102 Computational Ecology and Software, 2014, 4(2): 89-103

(8.4b) Plot of for the system (8.4)

(8.4c) An attractor of the system (8.4)


Fig. 8.4 Plots for the system (8.4).

9 Conclusion
This work is related to qualitative behavior of a four-dimensional exponential discrete population model. We
prove that system (1) has a unique positive equilibrium point which is globally asymptotically stable. Usually,
biologists believe that a globally asymptotically stable equilibrium point is very important in ecological point
of view. Method of linearization is used for local asymptotic stability of steady-state of (1). Furthermore, we
prove the boundedness and persistence of positive solutions of (1), and an invariant set for its solutions is

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 89-103 103

investigated. We prove the rate of convergence of positive solutions of system (1) which converge to its
unique positive equilibrium point.

References
Ahmad S. 1993. On the nonautonomous Lotka-Volterra competition equation. Proceedings of the American
Mathematical Society, 117: 199-204
Allen LJS. 2007. An Introduction to Mathematical Biology. Prentice Hall, USA
Brauer F, Castillo-Chavez C. 2000.Mathematical Models in Population Biology and Epidemiology. Springer
Din Q. 2013.Dynamics of a discrete Lotka-Volterra model. Advances in DifferenceEquations, 1: 1-13
Din Q. 2014.Global stability of a population model. Chaos, Soliton and Fractals, 59: 119-128.
Din Q, Donchev T. 2013.Global character of a host-parasite model. Chaos, Soliton and Fractals, 54: 1-7
Din Q, Khan AQ, Qureshi MN. 2013. Qualitative behavior of a host-pathogen model. Advances in Difference
Equations, 1: 263
Edelstein-Keshet L. 1988. Mathematical Models in Biology. McGraw-Hill, USA
Elaydi S. 2005. An Introduction to Difference Equations (3rd ed). Springer-Verlag, New York, USA
Elena E, Grammauro M, Venturino E. 2013. Predator’s alternative food sources do not support ecoepidemics
with two strains-diseased prey. Network Biology, 3(1): 29-44
El-Metwally E, Grove EA, Ladas G, et al. 2001.On the difference equation   .
Nonlinear Analysis, 47: 4623-4634
Grove EA, Ladas G. 2004. Periodicities in Nonlinear Difference Equations. Chapman and Hall/CRC Press,
Boca Raton, USA
Liu X. 2010. A note on the existence of periodic solution in discrete predator-prey models. Applied
Mathematical Modelling, 34: 2477-2483
Nedorezov LV. 2012. Continuous-discrete model of population dynamics with time lag in a reaction of intra-
population self-regulative mechanisms. Network Biology, 2(4):139-147
Nedorezov LV, Sadykov AM. 2012. About a modification of Rogers model of parasite-host system dynamics.
Proceedings of the International Academy of Ecology and Environmental Sciences, 2(1): 41-49
Papaschinopoulos G, Radin MA, Schinas CJ. 2011. On the system of two difference equations of exponential
form: , . Mathematical and Computer Modelling, 54:
2969-2977
Papaschinopoulos G, Schinas CJ. 2012. On the dynamics of two exponential type systems of difference
equations. Computers & Mathematics with Applications, 64(7): 2326-2334
Pituk M. 2002.More on Poincare’s and Perron’s theorems for difference equations. Journal of Difference
Equations and Applications, 8: 201-216
Sedaghat H. 2003. Nonlinear difference equations: Theory with applications to social science models. Kluwer
Academic Publishers, Dordrecht, Netherlands
Tang X, Zou X. 2006.On positive periodic solutions of Lotka-Volterra competition systems withdeviating
arguments. Proceedings of the American Mathematical Society, 134: 2967-2974
Zhou Z, Zou X. 2003. Stable periodic solutions in a discrete periodic logistic equation. Applied Mathematics
Letters, 16(2): 165-171

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 104-115

Article

Qualitative behavior of an anti-competitive system of third-order


rational difference equations

Q. Din1, M. N. Qureshi2, A. Q. Khan2


1
Department of Mathematics, Faculty of Basic & Applied Sciences, University of PoonchRawalakot, Pakistan
2
Department of Mathematics, University of Azad Jammu & Kashmir, Muzaffarabad, Pakistan
E-mail: qamar.sms@gmail.com,nqureshi@ajku.edu.pk,abdulqadeerkhan1@gmail.com

Received 6 December 2013; Accepted 10 January 2014; Published online 1 June 2014

Abstract
In this paper, our aim is to study the equilibrium points, local asymptotic stability, global behavior of an
equilibrium points and rate of convergence of an anti-competitive system of third-order rational difference
equations of the form:

, , 0,1, ,

where the parameters , , , , , and initial conditions , , , , , are positive real


numbers. Some numerical examples are given to verify our theoretical results.

Keywords rational difference equations; stability; global character; rate of convergence.

Computational Ecology and Software 
ISSN 2220­721X 
URL: http://www.iaees.org/publications/journals/ces/online­version.asp 
RSS: http://www.iaees.org/publications/journals/ces/rss.xml 
E­mail: ces@iaees.org 
Editor­in­Chief: WenJun Zhang 
Publisher: International Academy of Ecology and Environmental Sciences 

1 Introduction
Systems of rational difference equations have been studied by several authors. Especially there has been a
great interest in the study of the attractivity of the solutions of such systems.Cinar (2004) investigated the
periodicity of the positive solutions of the system of rational difference equations:

 , .

Stevic (2012) studied the system of two nonlinear difference equations:

 , ,

where  , , , are some sequences or . Stevic(2012) studied the system of three nonlinear difference
equations:

 , , ,

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 104-115 105

where the parameters , , , 1,2,3 are real numbers.Bajo and Liz (2011) investigated the global
behavior of difference equation:

for all values of real parameters , .Touafek and Essayed(2012) studied the periodic nature and got the form of
the solutions of the following systems of rational difference equations:

 , .

Recently, Zhang et al. (2012) studied the dynamics of a system of rational third-order difference equation:

 , , 0,1, .

Our aim in this paper is to investigate the qualitative behavior of positive solution for an anti-competitive
system of third-order rational difference equations:

                , , 0,1, (1)

where  , , , , , and initial conditions , , , , , are positive real numbers.


Let us consider six-dimensional discrete dynamical system of the form:
, , , , ,
(2)
, , , , ,
0,1, , where : and : are continuously differentiable functions and , are

some intervals of real numbers. Furthermore, a solution , of system (2) is uniquely determined
by initial conditions , for 2, 1,0 . Along with the system (2), we consider the
corresponding vector map , , , , , , , . An equilibrium point of system (2) is a
point , that satisfies
, , , , ,
, , , , , .
The point , is also called a fixed point of the vector map .
Definition 1. Let  , be an equilibrium point of the system (2).
(i) An equilibrium point  , is said to be stable if for every 0 there exists  0 such thatfor
every initial conditions , , 2, 1,0 if ∑ , , implies ,
, for all 0, where · is usual Euclidian norm in .
(ii) An equilibrium point  , is said to be unstable if it is not stable.
(iii) An equilibrium point  , is said to be asymptotically stable if there exists  0 such that
∑ , , and ,  ,  as ∞.
(iv) An equilibrium point  , is called global attractor if ,  , as ∞.
(v) An equilibrium point  , is called asymptotic global attractor if it is a global attractor andstable.
Definition 2. Let  , be an equilibrium point of a map
, , , , , , ,
where and are continuously differentiable functions at  , . The linearized system of (2) about the
equilibrium point  , is given by
,

IAEES www.iaees.org
106 Computational Ecology and Software, 2014, 4(2): 104-115

where  and is Jacobean matrix of the system (2) about the equilibrium point  , .

To construct corresponding linearized form of the system (1) we consider the following transformation:
, , , , , , , , , , (3)

where  , , , , , . The Jacobian matrix

about the fixed point  , under the transformation (3) is given by


0 0
1 0 0 0 0 0
 , 0 1 0 0 0 0 .
0 0
0 0 0 1 0 0
0 0 0 0 1 0

where  , , and .

Theorem 1. (Sedaghat, 2003). For the system , 0,1, , of difference equations such
that   be a fixed point of . If all eigenvalues of the Jacobian matrix  about  lie inside the open unitdisk
| | 1, then is locally asymptotically stable. If one of them has a modulus greater than one, then is
unstable.

2 Main Results
Let  , be an equilibrium point of the system (1), then system (1) has only 0,0 equilibrium point.
Theorem 2. Let , be positive solution of system (1), then for every 0 the following results hold.

, 6 6 , 0,1,2,
 0
, 6 6 , 3,4,5.

, 6 6 , 0,1,2,
  0
, 6 6 , 3,4,5.

Proof.  The result is obviously true for 0. Suppose that results are true for 1, . .,

, 6 6 , 0,1,2,
0
, 6 6 , 3,4,5.

and

, 6 6 , 0,1,2,
0
, 6 6 , 3,4,5.

Now, for 1 one has

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 104-115 107

0 , 3,4,5,

and

0 , 0,1,2.

So, the induction is complete. Similarly, inequalities hold by induction.


Theorem 3. For the equilibrium point 0,0 of the system (1) the following results hold:
If  and , then equilibrium point 0,0 of the system (1) is locally asymptotically stable.
If or , then equilibrium point 0,0 is unstable.
Proof. .The linearized system of (1) about the equilibrium point 0,0 is given by
0,0 ,

0 0 0 0 0
1 0 0 0 0 0
0 1 0 0 0 0
where  and 0,0 .
0 0 0 0 0
0 0 0 1 0 0
0 0 0 0 1 0
The characteristic polynomial of is given by

. (4)

The roots of are

exp  ,

for 0,1, ,5. Now, it is easy to see that | | 1 for all 0,1, ,5. Since all eigenvalues of Jacobian
matrix  0, 0 about (0, 0) lie in open unit disk | | 1. Hence, the equilibrium point (0, 0) is locally
asymptotically stable.
(ii). It is easy to see that if or , then there exist at least one root of equation (4) such that
| | 1. Hence by Theorem 1 if  or , then (0, 0) is unstable.
Theorem 4. The system (1) has no prime period-two solutions.
Proof. Assume that , , , , , , , be prime period two solution of system (1) such that
, 0 for 1,2 and , . Then, from system (1) one has

, , (5)

and

, . (6)

After some tedious calculations from (5) and (6), one has
4 0,
and
4 0,
which is contradiction. Hence, system (1) has no prime period-two solutions.
Theorem 5. Let   and , then equilibrium point (0, 0) of the system (1) is globally
asymptotically stable.
Proof. For  and , from Theorem 3 (0, 0) is locally asymptotically stable. From Theorem 2, it is

IAEES www.iaees.org
108 Computational Ecology and Software, 2014, 4(2): 104-115

easy to see that every positive solution , of the system (1) is bounded, . ., 0  and 0
for all 0,1,2, , where max  , , and max  , , .Now, it is sufficient to
prove that , is decreasing. From system (1) one has

This implies that  and . Also

This implies that and . So and


. Hence, the subsequences , , , , , and
, , , , , are decreasing. Therefore the sequences and
are decreasing. Hence lim ∞ lim ∞ 0.

3 Rate of Convergence
In this section we will determine the rate of convergence of a solution that converges to the unique equilibrium
point of the system (1). The following results give the rate of convergence of solutions of a system of
difference equations
, (7)
where  is an dimensional vector, is a constant matrix, and : is a matrix
function satisfying
0 (8)
as ∞, where · denotes any matrix norm which is associated with the vector norm
.
Proposition 1 (Perron’s theorem) (Pituk, 2002) Suppose that condition (8) holds. If is a solution of (7),
then either 0 for all large or

lim ∞ (9)

or

lim ∞ (10)

exist and is equal to the modulus of one the eigenvalues of matrix .

Assume that lim , lim .Then error terms for the system (1) are given by
∞ ∞

∑ ∑ ,
∑ ∑ .
Set and , one has
∑ ∑  , ∑ ∑ .
where


, ∏
, ∏
,

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 104-115 109

0 for 0,1 , ∏
, 0for 0,1 ,  ∏
,



, ∏
, ∏
.

Taking the limit, we obtain lim ∞ for 0,1,2 , lim ∞ 0 for 0,1 ,

lim ∞ , lim ∞ 0 for  0,1 , lim ∞ and lim ∞ for

0,1,2 . So, the limiting system of error terms can be written as ,where  .

Using the Proposition (1), one has following result.

Theorem 6. Assume that , be a positive solution of the system (1) such that lim  and

lim  where  , 0,0 . Then, the error term of every solution of (1) satisfies both of the

following asymptotic relations

lim  , , lim  , ,
∞ ∞

where  , are the characteristic roots of Jacobian matrix  , about 0,0 .

4 Examples
In order to verify our theoretical results we consider several interesting numerical examples in this section.
These examples represent different types of qualitative behavior of solutions to the system of nonlinear
difference equations (1). All plots in this section are drawn with Mathematica.
Example 1. Consider the system (1) with initial conditions  1.9 , 0.9, 0.8, 1.9 ,
1.5, 1.9 . Moreover, choosing the parameters 31, 32, 1.9, 521, 522,
1.8.Then , the system (1) can be written as

, , 0,1, , (11)
. .

and with initial condition 1.9 , 0.9, 0.8, 1.9 , 1.5, 1.9. . Moreover, in
Fig. 1 the plot of is shown in Fig. 1a, the plot of is shown in Fig. 1b and an attractor of the system (9)
is shown in Fig. 1c.
Example 2. Consider the system (1) with initial conditions  3.5 , 2.6, 2.1, 1.2 ,
2.8, 2.9. Moreover, choosing the parameters  1.1, 1.12, 0.004, 0.8, 0.81,
0.007. Then , the system (1) can be written as
. .
, , 0,1, , (12)
. . . .

and with initial condition 3.5 , 2.6, 2.1, 1.2 , 2.8, 2.9. Moreover, in Fig.
2 the plot of is shown in Fig. 2a, the plot of is shown in Fig. 2b and an attractor of the system (12) is
shown in Fig. 2c.
Example 3. Consider the system (1) with initial conditions  3.4 , 2.6, 2.1, 1.2 ,

IAEES www.iaees.org
110 Computational Ecology and Software, 2014, 4(2): 104-115

2.8, 2.7. Moreover, choosing the parameters  1500, 1535, 1.9, 1400, 1450,
350. Then, the system (1) can be written as

, , 0,1, , (13)
.

and with initial condition 3.4 , 2.6, 2.1, 1.2 , 2.8, 2.7. Moreover, in Fig.
3 the plot of is shown in Fig. 3a, the plot of is shown in Fig. 3b and an attractor of the system (13) is
shown in Fig. 3c.

(a) Plot of for the system (11)

(b) Plot of for the system (11)

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 104-115 111

(c) An attractor of the system (11)


Fig. 1 Plots for the system (11).

(a) Plot of for the system (12)

IAEES www.iaees.org
112 Computational Ecology and Software, 2014, 4(2): 104-115

(b) Plot of for the system (12)

(c) An attractor of the system (12)


Fig. 2 Plots for the system (12).

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 104-115 113

(a) Plot of for the system (13)

(b) Plot of for the system (13)

IAEES www.iaees.org
114 Computational Ecology and Software, 2014, 4(2): 104-115

(c) An attractor of the system (13)


Fig.3 Plots for the system (13).

5 Conclusions
In the paper, we investigate some dynamics of a six-dimensional discrete system. The system has only one
equilibrium point  0,0 . The linearization method is used to show that equilibrium point 0,0 is locally
asymptotically stable. Also system has no prime period-two solutions. The main objective of dynamical
systems theory is to predict the global behavior and rate of convergence of the system. In the paper, we use
simple technique to prove the global asymptotic stability of equilibrium point  0,0 . Some numerical examples
are provided to support our theoretical results.

Acknowledgements
This work was supported by the Higher Education Commission of Pakistan.

References
Agarwal RP. 2000. Difference Equations and Inequalities (Second Edition). Marcel Dekker, New York, USA
Bajo I, Liz E.2011.Global behaviour of a second-order nonlinear difference equation.Journal of Difference
Equations and Applications, 17(10): 1471-1486

Cinar C.2004. On the positive solutions of the difference equation system ,   .

Applied Mathematics and Computation, 158: 303-305


Din Q. 2013.Dynamics of a discrete Lotka-Volterra model. Advances in Difference Equations, 1: 1-13

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 104-115 115

Din Q. Global character of a rational difference equation, Thai Journal of Mathematics (in press)
Din Q. 2014.Global stability of a population model. Chaos, Soliton and Fractals, 59: 119-128.
Din Q, Donchev T. 2013.Global character of a host-parasite model. Chaos, Soliton and Fractals, 54: 1-7
Din Q. On a system of rational difference equation. Demonstratio Mathematica (in press)
Din Q, Khan AQ, Qureshi MN. 2013. Qualitative behavior of a host-pathogen model. Advances in Difference
Equations, 1: 263
Din Q, Qureshi MN, Khan AQ, Dynamics of a fourth-order system of rational difference equations, Advances
in Difference Equations, doi:10.1186/1687-1847-2012-216
Grove EA, Ladas G. 2004.Periodicities in Nonlinear Difference Equations. Chapman and Hall/CRC Press,
Boca Raton, USA
Kalabusic S, Kulenovic MRS, Pilav E. 2009. Global dynamics of a competitive system ofrational difference
equations in the plane. Advances in Difference Equations, Article ID 132802
Kalabusic S, Kulenovic MRS, Pilav E. 2011. Dynamics of a two-dimensional system of rational difference
equations of Leslie--Gower type. Advances in Difference Equations. doi: 10.1186/1687-1847-2011-29
Khan AQ, Qureshi MN, Din Q. 2013.Global dynamics of some systems of higher-order rational difference
equations.Advances in Differential Equations. doi: 10.1186/1687-1847-2013-354
Kocic VL, Ladas G: 1993. Global Behavior of Nonlinear Difference Equations of Higher Orderwith
Applications. Kluwer Academic, Dordrecht, Netherlands
Kurbanli AS. 2011. On the behavaior of positive solutions of the system of rational difference

       , , . Advances in Difference Equations, 40

Kurbanli AS, Cinar C, Yalcinkaya I. 2011. On the behavior of positive solutions of the system ofrational

difference equations , . Mathematical and Computer Modelling, 53:

1261-1267
Pituk M. 2002. More on Poincare's and Perron's theorems for di_erence equations.Journal of Difference
Equations and Applications, 8: 201-216
Sedaghat H. 2003.Nonlinear difference equations: Theory with applications to social science models. Kluwer
Academic Publishers, Dordrecht, Netherlands
Shojaei M, Saadati R, Adibi H. 2009. Stability and periodic character of a rational third orderdifference
equation. Chaos, Solitons and Fractals, 39: 1203-1209
Stevic S. 2012.On a third-order system of difference equations. Applied Mathematics and Computation, 218:
7649-7654
Touafek N, Elsayed EM. 2012. On the solutions of systems of rational difference equations. Mathematical and
Computer Modelling, 55: 1987-1997
Zhang Q, Yang L, Liu J.2012. Dynamics of a system of rational third order difference equation. Advances in
Difference Equations. doi:10.1186/1687-1847-2012-136

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 116-128

Article

Complex dynamics of a stochastic discrete modified Leslie-Gower


predator-prey model with Michaelis-Menten type prey harvesting

A. Elhassanein
Dept. of Math., Faculty of Science, Damanhour University, Damanhour, Egypt
E-mail: el_hassanein@yahoo.com

Received 30 January 2014; Accepted 5 March 2014; Published online 1 June 2014

Abstract
This paper introduced a stochastic discretized version of the modified Leslie-Gower predator-prey model with
Michaelis-Menten type prey harvesting. The dynamical behavior of the proposed model was investigated. The
existence and stability of the equilibria of the skeleton were studied. Numerical simulations were employed to
show the model's complex dynamics by means of the largest Lyapunov exponents, bifurcations, time series
diagrams and phase portraits. The effects of noise intensity on its dynamics and the intermittency phenomenon
were also discussed via simulation.

Keywords ecosystems; Leslie-Gower predator-prey model; skeleton; bifurcations; chaos.

Computational Ecology and Software     
ISSN 2220­721X   
URL: http://www.iaees.org/publications/journals/ces/online­version.asp 
RSS: http://www.iaees.org/publications/journals/ces/rss.xml 
E­mail: ces@iaees.org 
Editor­in­Chief: WenJun Zhang 
Publisher: International Academy of Ecology and Environmental Sciences 

1 Introduction
Deterministic nonlinear predator-prey models (ODE models) are widely used to understand the dynamics of
the ecosystems (Rosenzweig and MacArthur, 1963; Clark, 1976; Steven, 1994; Srinivasu, 2001; Kondoh, 2003;
Murdoch et al., 2003; Nedorezov and Sadykov, 2012). In recent years, there has been an increasing interest in
Leslie-Gower type predator-prey model (Hsu and Huang, 1995; Li and Xiao, 2007; Liang and Pan, 2007; Song
et al., 2009; Tian and Weng, 2011; Tian and Zhu, 2012). The local and global stability for a predator-prey
model of modified Leslie-Gower and Holling-type II with time-delay has been considered by Lin and Ho
(2006). Bifurcation analysis of a Leslie-Gower prey-predator model with Holling-type III functional response
has been studied by Li and Xiao (2007). The global stability of a Leslie-Gower prey-predator model with
proportional harvesting in both prey and predator has been studied by Zhang et al. (2011). By defining a
suitable Lyapunov function, the global stability of the unique interior equilibrium of the system was shown,
which means that suitable harvesting has no influence on the persistent property of the harvesting system.
Mena-Lorca et al. (2007) studied the dynamics of the Leslie-Gower model subjected to the Allee effect with
proportionate harvesting. The dynamical behavior of the periodic prey-predator model with a modified Leslie-
Gower Holling-type II scheme and impulsive effect has been studied by Song and Li (2008). Phase portraits

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 116-128 117

near the interior equilibria of a Leslie-Gower model with constant harvesting in prey has been considered by
Zhu and Lan (2010). Gupta and Chandra (2013) introduced a modified version of the Leslie-Gower prey-
predator model with Holling-type II functional response in the presence of nonlinear harvesting in prey. Many
authors (Agarwal, 2000; Agarwal, 1997; Freedman, 1980; Murray, 1989) have argued that the discrete time
models governed by difference equations are more appropriate than the continuous ones when the populations
have nonoverlapping generations. Discrete time models can also provide efficient computational models of
continuous models for numerical simulations. Discrete Leslie-Gower predator-prey model has also been
studied by many authors. Huo and Li (2004) considered a discrete Leslie-Gower predator-prey model. They
obtained sufficient conditions which guarantee the permanence of the model. Under the assumption that the
coefficients in the model are periodic, the existence of periodic solution is also obtained. Agiza et al (2009)
investigated discrete-time prey-predator model with Holling type II. The existence and stability of three fixed
points have been analyzed. The bifurcation diagrams, phase portraits and Lyapunov exponents have been
obtained for different parameters of the model. The fractal dimension of a strange attractor of the model has
been also calculated. They showed that the discrete model exhibits rich dynamics compared with the
continuous model. It has been shown that for the discrete-time prey-predator models the dynamics can produce
a much richer set of patterns than those observed in continuous-time models, see Danca (1997), Jing and Yang
(2006), Liu and Xiao (2007). The main objective of this paper is to propose a stochastic discrete version of the
modified Leslie-Gower predator-prey model with Michaelis-Menten type prey harvesting, and analyze its
chaotic behavior.
The organization of this paper is as follows. In section 2, a stochastic discrete version of the modified
Leslie-Gower predator-prey model with Michaelis-Menten type prey harvesting is formulated. In section 3, the
stability condition of the system are derived. The simulation is used in section 4, to discuss the analytical
results and to show the effects of noise intensity on the dynamics of the system.

2 The Stochastic Discrete Model


In the case of continuous time the modified Leslie-Gower predator-prey model with Michaelis-Menten type
prey harvesting has the following form
1 ,
(2.1) 
1                                    

here, and are the prey and predator population densities respectively and are intrinsic
growth rate and environmental carrying capacity for the prey species respectively. is the maximum value
of the per capita reduction rate of prey, measures the extent to which the environment provides protection
to prey and predator (under the assumption that the extent to which the environment provides protection to
both the predator and prey is the same (Ji et al., 2009, 2011), measures the growth rate of the predator
species,  is the maximum value of the per capita reduction rate of predator,  is the catchability coefficient,
is the effort applied to harvest the prey species, and , and are suitable constants. All the parameters
are assumed to be positive due to biological considerations.
Using a suitable non-dimensional scheme, the system (2.1) can be transformed into the following system,
Gupta and Chandra (2013).
1 ,
(2.2)
1                        

IAEES www.iaees.org
118 Computational Ecology and Software, 2014, 4(2): 116-128

with the initial conditions, 0 0, 0 0 where ,  , , , , and  are all positive.

Applying the forward Euler scheme to model (2.2) we obtain the following stochastic discrete modified
Leslie-Gower predator-prey model with Michaelis-Menten type prey harvesting model
1 ,
(2.3)
1                              

where is the step size ,  , , , , and  are defined as model (2.2) and ( ,  ) are assumed to be an
i.i.d. white noise sequence conditional upon the history of the time series, which is denoted Ω
, that is, , \Ω 0 and  \Ω \Ω , and is a scalar
parameter of the noise intensity.

3 The Skeleton (Free Noise System)


In this section we study the chaotic behaviour of the free noise system (2.3) caused by the change of time step.
Where 0, the system (2.3) becomes
1 ,
(3.1)
1                              
Fixed points of the system (3.1) are derived in the following.
Lemma 3.1. The fixed points of the system (3.1), are
  0,0 ;
  , , where

1 1 , and

  , , where 1 1 ,

Proof. The fixed points of the system (3.1) are obtained as the solution of the algebraic system:

1 ,
1                              
which is obtained by setting and in (3.1), it is easy to complete the proof.
The local stability analysis of the system (3.1) can be studied by computing the variation matrix
corresponding to each fixed point. The variation matrix of the system at state variable is given by
1 1 2
, . (3.2)
1 1

Theorem 3.1. The fixed point is unstable fixed point for all parameters values.
Proof. In order to prove this result, we estimate the eigenvalues of Jacobian matrix at . The Jacobian
matrix for is
1 1
.
0 1

Hence the eigenvalues of  are 1 1 , 1 , and since all parameters are positive

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 116-128 119

the proof is completed.


Theorem 3.2. The fixed point , 1,2 is locally asymptotically stable if one of the following
conditions is satisfied:
1 max , max  , , max  , ;
2 max , max  , min , min  , ;
3 max , max  , min , min  , ;
4   min , min  , , min  ,

otherwise it is unstable fixed point, where , , , ,

, , , 1 , and 2 .

Proof. The Jacobian matrix (3.2) at , 1,2 has the form


 
1  
.
1

where 1 , and 2 .

The corresponding characteristic equation of matrix , 1,2 is


0,
where 2 and 1
If the eigenvalues of the Jacobian matrix of a fixed point are inside the unit circle of the complex plan,
interior fixed point is local stable. Using Jury's conditions (Chatterjee and Yilmaz, 1992; Zhu and Lan, 2010),
we have necessary and sufficient condition for local stability of interior fixed point which are the necessary
and sufficient condition for , 1, which completes the proof.

4 Numerical Simulation
4.1 Deterministic system

The main purpose of this subsection is to investigate the qualitative behavior of the solution of the nonlinear

system (3.1). To provide some numerical evidence for its chaotic behavior, we present various numerical

results here to show the chaoticity including its bifurcation diagrams, Lyapunov exponents, and fractal

dimension. In Fig. 1(a) bifurcation diagram of the system (3.1) is plotted on the interval 0.2 0.9 for

initial point , 0.01,0.02 with , , , , , 1,2,2,1,2.1,3 . In Fig. 2, a plot of the Lyapunov

exponent for attractors of the system (3.1) according to Fig. 1 is presented. A positiveness of this exponent for

0.6689 confirms the chaotic character of attractors in this parametrical zone (here, the value

0.6689 is a tangent bifurcation point). In Fig. 3 phase portraits are given for different values of to show the

chaotic behavior of the system. For the given parameters the only positive equilibrium point is

IAEES www.iaees.org
120 Computational Ecology and Software, 2014, 4(2): 116-128

0.030662,1.01533 which is attracting point for 0.2 0.66890 as we can see in Fig. 3. (a). Fig. 3.(b)

show the period-two orbit in the parameter zone 0.6689< l <0.8195. The period-four orbit in the parameter

zone 0.8195 0.8527 is clear in Fig. 3(c). The chaotic attractor for 0.8527 0.9 is clear in Fig.

3(d). Which means that the system (3.1) undergoes a discrete Hopf bifurcation. One of the commonly used

characteristics for classifying and quantifying the chaoticity of a dynamical system is fractal dimensions,

(Cartwright, 1999; Zhu and Lan, 2010). Via simulation we get two Lyapunov exponents 0.08026 0
.  
0.35263 for 0.871, which means that 1 1.2276. There for the system (3.1)
.

exhibits a fractal structure and its attractor has a fractal dimension which is chaotic behavior.

Fig. 1 Bifurcation diagram of (3.1), for 0.2 0.9 and initial point , 0.01,0.02 with
, , , , , 1,2,2,1,2.1,3 .

Fig. 2 Maximum Lyapunov exponents corresponding to Fig. 1.

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 116-128 121

Fig. 3 Phase portrait of the system (3.1) for: (a)  0.6 ; (b)  0.6679 (c)  0.671 (d)  0.87 with
, , , , , 1,2,2,1,2.1,3 .

4.2 Stochastic model


Along with the deterministic systems (3.1) we consider a stochastic system forced by additive noise (2.3),
where and are uncorrelated Gaussian random processes with parameters , 0 and  1
1, and is a scalar parameter of the noise intensity. We study a behavior of this stochastic model for
different sets of parameters and . In Figs. 4, 5 and 6, stochastic attractors of the system (2.3) are plotted on
the interval 0.2 0.9 for three values of the noise intensity 0.0002, 0.0004, and 0.0006 .
Lyapunov exponents corresponding to each value of noise intensity are given in Fig. (7) (a), (b) and (c)
respectively. As we can see, noise deforms the deterministic attractor (compare Figs. 2 and 7). As noise
intensity increases, a border between order and chaos moves to the left. The changes of the arrangement of
attractors are accompanied by the changes in dynamical characteristics (compare Lyapunov exponents in Figs.
2 and 8). The most essential difference between stochastic and deterministic attractors is observed near the
bifurcation point 0.6689. The underlying reason is that in the vicinity of this bifurcation point attractors
are highly sensitive to random disturbances. Consider in detail a behavior of stochastic system (2.3)
near  0.6689. Compare the stochastic response of this system for two fixed values 0.668 and
0.671. Consider 0.668. For low noise 0.0002, random states are concentrated near the stable
deterministic equilibrium (see Figs. 8 and 12(a)), they have distribution with mean equal approximately
0.0306,1.0153 and covariance matrix

IAEES www.iaees.org
122 Computational Ecology and Software, 2014, 4(2): 116-128

0.00000057 0.00000038
Σ .
0.00000038 0.00002028
For 0.0006 one can see stochastic oscillations of large amplitude. Indeed, as the noise intensity increases,
the dispersion of random states near growsc see Table 1. After these oscillations, iterations come to the
vicinity of the point again and so on (see Figs. 9 and 12(b)). In this case, the stochastic model (2.3)
exhibits a coexistence of two different dynamical regimes even if the system (3.1) has a stable equilibrium
only. This type of dynamics of the system (2.3) can be determined as a noise-induced intermittency
(Bashkirtseva and Ryashko 2013). In Figs. 10 and 11, time series of the stochastic system (2.3) with
0.671 for 0.0002 and 0.0002 are plotted. As can be seen, noise-induced intermittency for this
0.671 is observed for the lower noise intensity, see also Fig. 13. For stochastic attractors and their
dynamic characteristics, a dependence on noise level is illustrated in Figs. 14 and 15 for 0.668 and
0.671. In Fig. 14, one can see a sharp growth of the size of the attractor as noise intensity exceeds some
critical value. A change of the sign of Lyapunov exponent from minus to plus can be interpreted as a transition
from regular to noise-induced chaotic regime (see Fig. 16). Thus, the results presented here give us a
qualitative description of noise-induced transitions from the regular regime to intermittency.

Fig. 4 Bifurcation diagrams of the stochastic model (2.3) with 0.0002, 0.2 0.9, and , , , , ,
1,2,2,1,2.1,3 .

Fig. 5 Bifurcation diagrams of the stochastic model (2.3) with 0.0004, 0.2 0.9, and , , , , ,
1,2,2,1,2.1,3 .

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 116-128 123

Fig. 6 Bifurcation diagrams of the stochastic model (2.3) with 0.0006, 0.2 0.9, and , , , , ,
1,2,2,1,2.1,3 .

Fig. 7 Lyapunov exponents of the stochastic model (2.3) with 0.2 0.9: for (a) 0.0002; (b) 0.0004; (c)
0.0006.

Fig. 8 Time series of the stochastic model (2.3) with l =0.668, 0.0002 and , , , , , 1,2,2,1,2.1,3 .

IAEES www.iaees.org
124 Computational Ecology and Software, 2014, 4(2): 116-128

Fig. 9 Time series of the stochastic model (2.3) with  .668, 0.0006 , , , , , 1,2,2,1,2.1,3 .

Fig. 10 Time series of the stochastic model (2.3) with .671,  0.0002 and , , , , , 1,2,2,1,2.1,3 .

Fig. 11 Time series of the stochastic model (2.3) with .671, 0.0006 and , , , , , 1,2,2,1,2.1,3 .

Fig. 12 Phase portrait of the system (2.3) with .668, and , , , , , 1,2,2,1,2.1,3 for: (a) 0.0002;
(b) 0.0006.

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 116-128 125

Fig. 13 Phase portrait of the system (2.3) with .671 and , , , , , 1,2,2,1,2.1,3 for: (a) 0.0002;
(b) 0.0006.

Fig. 14 Attractors of the stochastic system (2.3) for 0 0.0006 ; 0.668 and , , , , ,
1,2,2,1,2.1,3 .

Fig. 15 Attractors of the stochastic system (2.3) for 0 0.0006 ; 0.671 , and , , , , ,
1,2,2,1,2.1,3 .

Fig. 16 Lyapunov exponents of the stochastic system (2.3) with 0 0.0006 and
, , , , , 1,2,2,1,2.1,3 for: (a)  0.668; (b) 0.671.

IAEES www.iaees.org
126 Computational Ecology and Software, 2014, 4(2): 116-128

Table 1 Mean of the stochastic states , of the system (2.3) at different values of , and , , , , ,
1,2,2,1,2.1,3 .
╲    0.668 0. 671
0.0002 (0.030572,1.0153) (0.032278,1.0142)
0.0006 (0.029719,1.0147) (0.033094,1.0145)

Table 2 Covariance matrices of the stochastic states , of the system (2.3) at different values of , and
, , , , , 1,2,2,1,2.1,3 .
╲    0.668 0. 671
0.0002 0.00000057 0.00000038 0.00000140   0.00001124
0.00000038 0.00002028 0.00001124  0.00195619
0.0006 0.00001082 0.00000606 0.00001211  0.00001732
0.00000606 0.00016785 0.00001732  0.00203988

5 Conclusion
From a mathematical as well as biological point of view the predator-prey models can be formulated as
systems of differential or difference equations (Nedorezov and Sadykov, 2012). The current paper have
proposed a stochastic discrete modified Leslie-Gower predator-prey model with Michaelis-Menten type prey
harvesting, where the protection provided by the environment for both the prey and predator is the same. The
model shows rich and varied dynamics. The local stability of fixed points have been discussed. The results
show that the origin is unstable equilibrium point of the system. There is a unique interior equilibrium point
which is locally stable for certain parametric restrictions. The effectiveness of the time step on the dynamics of
the system has been shown. The chaotic behaviour of the system has been proved. We focus on the study of
the noise-induced type-I intermittency phenomenon and chaotization observed near tangent bifurcation. The
remarkable feature of the dynamics of the model considered here is that small noises generate large-amplitude
chaotic oscillation.

References
Agarwal RP. 2000. Difference Equations and Inequalities: Theory, Method and Applications. Monographs and
Textbooks in Pure and Applied Mathematics (Volume 2). Marcel Dekker, New York, USA
Agarwal RP, Wong PJY. 1997. Advance Topics in Difference Equations. Kluwer, Dordrech. Germany
Agiza HN, ELabbasy EM, EL-Metwally H, Elsadany AA 2009. Chaotic dynamics of a discrete prey-predator
model with Holling type II. Nonlinear Analysis, RWA 10: 116-129
Bashkirtseva I, Ryashko L. 2013: Stochastic sensitivity analysis of noise-induced intermittency and transition
to chaos in one-dimensional discrete-time systems. Physica A, 392: 295-306
Cartwright J HE 1999. Nonlinear stifiness, Lyapunov exponents, and attractor dimension. Physical Letters A,
264: 298-304
Chatterjee S, Yilmaz M. 1992. Chaos, fractals and statistics. Statistical Science, 7: 49-68
Clark CW 1976. Mathematical Bioeconomics the Optimal Management of Renewable Resources. Wiley, New
York
Danca M, Codreanu S, Bako B 1997. Detailed analysis of a nonlinear prey-predator model. Journal of

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 116-128 127

Biological Physics, 23: 11-20


Freedman HI 1980. Deterministic Mathematics Models in Population Ecology. Marcel Dekker, New York,
USA
Gupta RP, Chandra P. 2013. Bifurcation analysis of modified Leslie-Gower predator-prey model with
Michaelis--Menten type prey harvesting. Journal of Mathematical Analysis and Applications, 398: 278-295
Hsu SB, Huang TW. 1995. Global stability for a class of predator-prey systems. SIAM Journal of Applied
Mathematics, 55: 763-783
Huo H, and Li WT. 2004. Stable periodic solution of the discrete periodic Leslie-Newer predator-prey model.
Mathematical and Computer Modelling 40: 261-269
Ji C, Jiang D, Shi N. 2009. Analysis of a predator--prey model with modified Leslie-Gower and Holling-type
II schemes with stochastic perturbation. Journal of Mathematical Analysis and Applications, 359: 482-498
Ji C, Jiang D, Shi N. 2011. A note on a predator--prey model with modified Leslie-Gower and Holling-type II
schemes with stochastic perturbation. Journal of Mathematical Analysis and Applications, 377(1): 435-440
Jing ZJ, Yang J. 2006. Bifurcation and chaos discrete-time predator--prey system. Chaos, Solitons and Fractals,
27, 259-277
Kondoh M 2003. Foraging adaptation and the relationship between food-web complexity and stability. Science,
299: 1388-1391
Li Y, Xiao D. 2007. Bifurcations of a predator-prey system of Holling and Leslie types. Chaos, Solitons and
Fractals, 34, 606-620
Liang Z, Pan H. 2007. Qualitative analysis of a ratio-dependent Holling-Tanner model. J. Math. Anal. Appl.
334, 954-964
Lin CM, Ho CP. 2006. Local and global stability for a predator-prey model of modified Leslie-Gower and
Holling-type II with time-delay. Tunghai Science, 8: 33-61
Liu X, Xiao D 2007. Complex dynamic behaviors of a discrete-time predator-prey system. Chaos, Solitons and
Fractals, 32: 80-94
Mena-Lorca J, Gonzalez-Olivares E, Gonzalez-Yanz B. 2007. The Leslie-Gower predator-prey model with
Allee effect on prey: a simple model with a rich and interesting dynamics. In: Proceedings of the
International Symposium on Mathematical and Computational Biology BIOMAT 2006 (Mondaini R, ed).
105, 132, E-papers Servios Editoriais Ltda, R' io de Janeiro, Brazil
Murdoch W, Briggs C, Nisbet R. 2003. Consumer-Resource Dynamics. Princeton University Press, New York,
USA
Murray J D 1989. Mathematical Biology. Springer-Verlag, New York, USA
Nedorezov LV, Sadykov AM. 2012. About a modification of Rogers model of parasite-host system dynamics.
Proceedings of the International Academy of Ecology and Environmental Sciences, 2(1): 41-49
Song Y, Yuan S, Zhang J. 2009. Bifurcation analysis in the delayed Leslie-Gower predator-prey system.
Applied Mathematical Modelling, 33: 4049-4061
Rosenzweig ML, MacArthur RH. 1963. Graphical representation and stability conditions of predator--prey
interactions. American Naturalist, 97: 209
Song XY, Li YF. 2008. Dynamic behaviors of the periodic predator-prey model with modified Leslie-Gower
Holling-type II schemes and impulsive effect. Nonlinear Analysis, RWA 9: 64-79
Srinivasu PDN. 2001. Bioeconomics of a renewable resource in presence of a predator. Nonlinear Analysis,
RWA 2: 497-506
Steven HS. 1994. Nonlinear dynamics and chaos: with applications to physics, biology, chemistry and
engineering. Perseus Books Publishing, USA

IAEES www.iaees.org
128 Computational Ecology and Software, 2014, 4(2): 116-128

Tian C, Zhu P. 2012. Existence and asymptotic behavior of solutions for quasilinear parabolic systems. Acta
Applicandae Mathematicae. DOI 10.1007/s 10440-012-9701-7
Tian Y, Weng P. 2011. Stability analysis of diffusive predator-prey model with Modified Leslie-Gower and
Holling-Type II schemes. Acta Applicandae Mathematicae, 114(3): 173-192
Zhang N, Chen F, Su Q, Wu T. 2011. Dynamic behaviors of a harvesting Leslie-Gower predator-prey model.
Discrete Dynamics in Nature and Society. http://dx.doi.org/10.1155/2011/473-94.
Zhu CR, Lan KQ. 2010. Phase portraits, Hopf-bifurcations and limit cycles of Leslie-Gower predator--prey
systems with harve sting rates. Discrete and Continuous Dynamical Systems-Series B, 14: 289-306

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 129-134

Article

Synthetic Aperture Radar (SAR) image segmentation by fuzzy c-


means clustering technique with thresholding for iceberg images

Usman Seljuq1, Rashid Hussain2


1
Sir Syed University of Engineering & Technology Karachi-75300, Pakistan
2
Faculty of Engineering Science and Technology, Hamdard University, Karachi 74600, Pakistan
E-mail: rashid.hussain@hamdard.edu.pk

Received 27 December 2013; Accepted 5 February 2014; Published online 1 June 2014

Abstract
Fuzzy c-means (FCM) clustering algorithm is widely used for image segmentation. The purpose of clustering
is to identify natural groupings of data from a large data set, which results in concise representation of
system’s behavior. It can be used to detect icebergs regardless of ambient conditions like rain, darkness and
fog. As a result SAR images can be used for iceberg surveillance. In this paper we have investigate FCM with
thresholding for iceberg image segmentation for Synthetic Aperture Radar (SAR) images. The results showed
that the assessment parameters; mean and entropy have lower values for efficient segmentation.

Keywords Synthetic Aperture Radar (SAR); fuzzy c-means clustering; thresholding; image segmentation.

Computational Ecology and Software     
ISSN 2220­721X   
URL: http://www.iaees.org/publications/journals/ces/online­version.asp 
RSS: http://www.iaees.org/publications/journals/ces/rss.xml 
E­mail: ces@iaees.org 
Editor­in­Chief: WenJun Zhang 
Publisher: International Academy of Ecology and Environmental Sciences 

1 Introduction
Image segmentation is a preliminary step in pattern recognition and scene analysis understanding. It partitions
the image into its constituent regions and objects. Image processing is effected by illumination conditions,
environmental disturbances, noise and temperature fluctuation. Under some severe conditions of unexpected
and improper illumination, the blurring of images makes it very difficult for target recognition, so under such
conditions segmentation is necessary. The SAR system is used in investigating environmental and ecological
activities (Hussain, 2012). SAR images are not affected by clouds, sunlight, day and night effects. The
satellites and airplanes equipped with SAR are used to monitor icebergs. SAR is an active microwave that
captures images (Topouzelis, 2008). One of the most important applications of SAR system is to collect the
data from the ground surface through image reconstruction. The ground surface has different areas such as
roads, deserts, pounds, buildings, grassland and cultivated plants, so those areas have to be segmented
according to the applications. Thus segmentation problem arises and because of this various segmentation
methods have been proposed. Correct segmentation is an important issue in SAR image segmentation. Several
methods are used for SAR image segmentation such as clustering algorithms, threshold methods and

IAEES www.iaees.org
130 Computational Ecology and Software, 2014, 4(2): 129-134

morphological methods. This paper reexamines the approach of SAR clustering based image segmentation.
Fuzzy clustering has a potential application in the fields of image processing and pattern recognition. Some of
the recent research includes an Improved FCM Algorithm Based on the SVDD for Unsupervised
Hyperspectral Data Classification (Niazmardi et al., 2013), Fuzzy C-Means Algorithms for Very Large Data
(Havens et al., 2012), Weighted Fuzzy C-Means Clustering Algorithm for Remotely Sensed Image
Classification (Chen et al., 2011), Unsupervised Subpixel Mapping of Remotely Sensed Imagery Based on
Fuzzy C-Means Clustering Approach (Zhang et al., 2014), A Multiple-Kernel Fuzzy C-Means Algorithm for
Image Segmentation (Chen et al., 2011), and Change Detection in Synthetic Aperture Radar Images based on
Image Fusion and Fuzzy Clustering (Gong et al., 2012).
Clustering is commonly used in pattern recognition, machine learning, biomedical and image analysis. In
fuzzy logic FCM is a technique of clustering that allows one piece of data belongs to two or more clusters (Pal
and Bezdek, 1995).
This paper is organized as follows: section 2 represents the need for image segmentation. Section 3
discusses the proposed FCM algorithm for the segmentation of SAR iceberg images with thresholding. The
simulation results and evaluation of the reconstructed image are represented in section 4 and section 5. Finally,
in section 6 conclusions are given.

2 Iceberg Image Segmentation


Image segmentation is used to locate the boundaries and objects in images. One of the most practical
applications of image segmentation is to locate the objects in SAR images (roads, buildings, forests etc.) image
processing is effected by random noise, environmental disturbances due to atmospheric pressure or
temperature fluctuations and illumination conditions, so image segmentation is necessary that divides an image
into regions of different characteristics.
SAR image segmentation falls into two categories, a) Segmentation based on grey levels b) Segmentation
based on texture. Image segmentation thresholding techniques can be found in (Sahoo et al., 1988). This paper
deals with SAR image segmentation based on grey levels with global thresholding technique.

3 Investigating Fuzzy C-Mean Clustering With Thresholding


The investigation is based upon the combination of FCM clustering with thresholding algorithm. Clustering of
data is a technique by which large sets of data are grouped into smaller ones.
3.1 Fuzzy clustering sets of clusters of similar data
The partition should have two properties, (a) homogeneity inside clusters: the data, which belong to one cluster,
should be as similar as possible; (b) heterogeneity between clusters: The data, which belongs to different
clusters, should be as different as possible. Many techniques have been used for clustering data; in this paper
fuzzy c-means clustering is used (Sahoot et al., 1988). FCM was developed by Dunn in 1973 (Dunn, 1973),
improved by Bezdek in 1981 (Bezdek, 1981), and further developed by many researchers (Cannon et al., 1986;
Lyer et al., 2002; Wang et al., 2006; Hussain, 2012).
In 2010, Krinidis and Chatzis proposed research incorporating local spatial information into FCM
algorithm to get improved clustering performance (Krinidis, 2010).
3.2 FCM algorithm
FCM divides a set of S-dimensional vectors X = [X1,……..,Xn] into c clusters. Each cluster is a fuzzy set
defined by the sample space X = [X1,……..,Xn]. The ith cluster is presumed to have the center vector vi =
[vi1 , . . . . , vis], where Xj = [xj1 , . . . . , xjs] represents the jth sample for ( j=1, . . . , n ). FCM determines the
fuzzy partition matrix U and cluster centers vi ( 1,2,..,c) by minimizing the objective function J which is

IAEES www.iaees.org
Computational Ecology and Software, 2014, 4(2): 129-134 131

defined by (1).
c n

J(U, v1,v2,. . . . ,vc ; X) = ∑ ∑ umij d2ij (1)


i=1 j=1

m in equation (1) adjusts the weighting effect. Large values of m increases the fuzziness
where dij is the Euclidean distance and is defined

S
dij   (v
K 1
ik  xjk ) 2
(2)

vi and uij is given by

u ij
m
X
vi  j 1


n
j 1
uij m
(3)

uij  1 c
dij (4)
(d
k 1 kj
) 2 /( m 1)

where m≠1
3.3 Thresholding
In fuzzy c-means clustering the segmented part of SAR image is not visible clearly. Thusthresholding is
applied on the segmented image (Otsu, 1979; Gonzalez, 2002). There are two techniques of thresholding: local
and global thresholding. In local thresholding the image is divided into no of sub-images, the threshold for
each sub-images depends upon the local properties of the point. In global thresholding the entire image is
segmented with one or more values.
We have used global thresholding method. The thresholded image g(x,y) is defines as
g(x,y )= { 1 if f(x,y) ≥ T
{ 0 if f(x,y) < T (5)
where T is constant, this approach is called global thresholding (Dunn., 1973).

4 Evaluation of the Reconstructed Images


To evaluate the accuracy of the segmented image following steps are followed.
(1) Iceberg image is taken as input.
(2) Segmentation algorithm is applied on the image.
(3) Assessment parameters are applied for fuzzy c-means clustering algorithm.
The proposed algorithm is applied by using MATLAB .The images are resized to 256x256, performance of
the algorithm is evaluated by following assessment parameters. 1) Mean 2) Entropy. Mean and entropy must
be a less values for a better segmentation algorithm.
Entropy is the amount of information based on the probabilistic occurrence of the picture elements. These
probabilities can be found from the image histogram. Entropy describes the average information of every pixel.
The more gray levels are present in the image, the higher the entropy. Suppose that two images have

IAEES www.iaees.org
132 Com
mputational Ecollogy and Softw
ware, 2014, 4(2): 129-134

probabiliity functions p and q, the entropy


e of p with
w respect to
t q is given by
b (6).
k
p (i )
d   p(i ) log
g2
i 1 q(i ) (6)

5 Results and Discusssion


From thee figure 1 the difference beetween the fuuzzy c-means clustering annd fuzzy c-meeans thresholld for icebergg
images can
c be observve. The fuzzyy c-means thhreshold givess better repreesentation of the image ass compare too
fuzzy c-m
means clutterring.

(
(a) (b) (c)
Fig. 1 (a) original image (b) FCM (cc) FCM with thresholding.

Table 1 FC
CM with hresho
olding for Fig. 1.
1

S.No.. STAST
TICAL ORIGINAL
L FCM FCM W
WITH
PARAMEETERS IMAGE THRESHOLD
1 MEAN 98.4815 0.4413 0.2169
2 ENTROPY
Y 7.3351 4.0313 0.7545

(a) (b
b) (c)
Fig. 2 (a) original image (b) FCM (cc) FCM with thresholding.

IAEES www.iaees.org
w
Computational Ecology and Software, 2014, 4(2): 129-134 133

Table 2 FCM with thresholding for Fig. 2.

SNO STASTICAL ORIGINAL FCM FCM WITH


PARAMETERS IMAGE THRESHOLD
1 MEAN 97.4866 0.4707 0.1869
2 ENTROPY 7.4895 7.0259 0.695

From Table 1 and 2, the less values of assessment parameters like mean and entropy shows that the fuzzy
c-means threshold method is better for iceberg images when compare with fuzzy c-Means clustering. With the
overall performance evaluation we can say that fuzzy c-means threshold method gives desirable results when
compare to fuzzy c-means method.

6 Conclusion
Fuzzy c-means thresholding method has been tested on various SAR images. The scheme implemented here
makes use of thresholding and fuzzy c-means clustering algorithm for image segmentations. This approach
made it possible to segment patterns in SAR images, which could go undetected by conventional image
clustering methods. This analysis demonstrates how new computational techniques help in satellite imaging
for environmental and ecological monitoring.

Acknowledgments
We wish to thank Matthew R. Lopez from Sandia National Laboratories Albuquerque, NM, USA, Ellen O'
Leary from Radar Data Center, Jet Propulsion Laboratory Pasadena, CA and J.C Bezdek for facilitating
theoretical concepts, and to the Graduate School of Engineering Sciences and Information Technology,
Hamdard University for their logistic support and services.

References
Bezdek JC. 1981. Pattern Recognition with Fuzzy Objective Function Algorithms. Plenum Press, New York,
USA
Cannon RL, Dave JV, Bezdek, JC. 1986. Efficient implementation of the fuzzy c-means clustering algorithms.
IEEE Transaction on Pattern Analysis and Machine Intelligence, 248-255
Chen L, Chen CLP, Lu M. 2011. A multiple-kernel fuzzy c-means algorithm for image segmentation. IEEE
Transactions on Systems, Man, and Cybernetics-Part B, 41: 1263–1274
Dunn JC. 1973. A fuzzy relative of the ISODATA process and its use in detecting compact well- separated
clusters. Journal of Cybernetics, 3: 32-57
Gong MG, Zhou ZQ, Ma JJ. 2012. Change detection in synthetic aperture radar images based on image fusion
and fuzzy clustering. IEEE Transactions on Image Processing, 21(4): 2141-2151
Gonzalez R, Woods R. 2002. Digital Image Processing (2nd Edition). Prentice-Hall, USA
Havens TC, Bezdek JC, Leckie C, et al. 2012. Fuzzy c-means algorithms for very large data. IEEE
Transactions on Fuzzy Systems, 20(6): 1130–1146
Hung CC, Kulkarni S, Kuo BC. 2011. A new weighted fuzzy c-means clustering algorithm for remotely
sensed image classification. IEEE Journal on Signal Processing, 5(3): 543-553
Hussain R. 2012. Synthetic Aperture Radar (SAR) images features clustering using fuzzy c-means (FCM)

IAEES www.iaees.org
134 Computational Ecology and Software, 2014, 4(2): 129-134

clustering algorithm. Computational Ecology and Software, 2(4): 220-225


Krinidis S, Chatzis V. 2010. A robust fuzzy local information c-means clustering algorithm. IEEE
Transactions on Image Processing, 19(5): 1328-1337
Lyer NS, Kandel A, Schneider M. 2000. Feature-based fuzzy classification for interpretation of mammograms.
Fuzzy Sets and Systems, 114(2): 271-280
Niazmardi S, Homayouni S, Safari A. 2013. An improved FCM algorithm based on the SVDD for
unsupervised hyperspectral data classification. IEEE Journal of Remote Sensing, 6(2): 831–839
Otsu N. 1979. A threshold selection method from gray-level histograms. IEEE Transactions on Systems, Man
and Cybernetics, 9(1): 62-66
Pal NR, Bezdek JC. 1995. On cluster validity for the fuzzy c-means model. IEEE Trans Fuzzy Systems, 3(3):
370-379
Sahoo PK, Soltani S, Wong AKC, et al. 1988. A survey of thresholding technique. Computer Vision Graphics
Image Process, 41: 233-260
Topouzelis KN. 2008. Oil spill detection by SAR images: dark formation detection, feature, extraction and
classification algorithms. Sensors, 8: 8: 6642-6659
Wang WN, Zhang YJ, Li Y, et al. 2006. The global fuzzy c-means algorithm. IEEE Proceedings of the 6th
World Congress on Intelligence. Control and Automation Clustering, 1: 3604-3607
Zhang Y, Du Y, Li X, Fang SM. 2014. Unsupervised subpixel mapping of remotely sensed imagery based on
fuzzy c-means clustering approach. IEEE Geoscience and Remote Sensing Letters, 11(5): 1024-1028

IAEES www.iaees.org
Computational Ecology and Software

The Computational Ecology and Software (CES) is an open access, peer-reviewed online journal that considers
scientific articles in all different areas of computational ecology. It is the transactions of the International
Society of Computational Ecology. The journal is concerned with the ecological researches, constructions and
applications of theories and methods of computational sciences including computational mathematics,
computational statistics and computer science. It features the simulation, approximation, prediction, recognition,
and classification of ecological issues. Intensive computation is one of the major stresses of the journal. The
journal welcomes research articles, short communications, review articles, perspectives, and book reviews. The
journal also supports the activities of the International Society of Computational Ecology. The topics to be
covered by CES include, but are not limited to:

 Computation intensive methods, numerical and optimization methods, differential and difference
equation modeling and simulation, prediction, recognition, classification, statistical computation
(Bayesian computing, randomization, bootstrapping, Monte Carlo techniques, stochastic process, etc.),
agent-based modeling, individual-based modeling, artificial neural networks, knowledge based systems,
machine learning, genetic algorithms, data exploration, network analysis and computation, databases,
ecological modeling and computation using Geographical Information Systems, satellite imagery, and
other computation intensive theories and methods.
 Artificial ecosystems, artificial life, complexity of ecosystems and virtual reality.
 The development, evaluation and validation of software and algorithms for computational ecology. The
development and evaluation of apparatus, instruments and machines for ecological and environmental
analysis, investigation and monitoring based on the software of computational ecology.
 Methodological applications of computational ecology in the researches of ecology and environmental
sciences.

Authors can submit their works to the email box of this journal, ces@iaees.org and (or) wjzhang@iaees.org. All
manuscripts submitted to CES must be previously unpublished and may not be considered for publication
elsewhere at any time during review period of this journal.

In addition to free submissions from authors around the world, special issues are also accepted. The organizer of
a special issue can collect submissions (yielded from a research project, a research group, etc.) on a specific
topic, or submissions of a conference for publication of special issue.

Editorial Office: ces@iaees.org

Publisher: International Academy of Ecology and Environmental Sciences


Address: Flat C, 23/F, Lucky Plaza, 315-321 Lockhart Road, Wanchai, Hong Kong
Tel: 00852-6555 7188
Fax: 00852-3177 9906
E-mail: office@iaees.org
Computational Ecology and Software
ISSN 2220-721X
Volume 4, Number 2, 1 June 2014

Articles

Spatial risk assessment of alien plants in China using biodiversity resistance


theory
YouHua Chen 82-88

Stability analysis of a discrete ecological model


Q. Din, E. M. Elsayed 89-103

Qualitative behavior of an anti-competitive system of third-order rational


difference equations
Q. Din, M. N. Qureshi, A. Q. Khan 104-115

Complex dynamics of a stochastic discrete modified Leslie-Gower


predator-prey model with Michaelis-Menten type prey harvesting
A. Elhassanein 116-128

Synthetic Aperture Radar (SAR) image segmentation by fuzzy c-means


clustering technique with thresholding for iceberg images
Usman Seljuq, Rashid Hussain 129-134

The International Academy of Ecology and Environmental Sciences (IAEES) is a nonprofit and registered
international organization. It devotes to promote global ecology and environmental sciences and protect global
ecological environments, by publishing scientific publications, conducting research activities, launching
environmental programs, disseminating knowledge and technologies, sponsoring conferences, and providing
information and discussion spaces, etc. The responsibility for these publications rests with the International
Academy of Ecology and Environmental Sciences.

IAEES
http://www.iaees.org/

You might also like