MULTISTABILITY ON A LESLIE-GOWER TYPE PREDATOR-PREY MODEL WITH NONMONOTONIC FUNCTIONAL RESPONSE

BETSABE GONZALEZ-YANEZ, EDUARDO GONZALEZ-OLIVARES , JAIME MENA-LORCA Grupo de Ecolog Matemtica, Instituto de Matemticas a a a Ponticia Universidad Catlica de Valpara o so E-mails: betsabe.gonzalez@ucv.cl, ejgonzal@ucv.cl, jmena@ucv.cl

In this work, a bidimensional continuous-time dierential equations system is analyzed which is derived from a Leslie type predator-prey model by considering a nonmonotonic functional response (or Holling type IV or Monod Haldane). This functional response is employed to explain a class of prey antipredator strategies and we study how it inuences in bifurcation and stability behavior of model. System obtained can have one, two or three equilibrium point at interior of the rst quadrant, but here we describe the dynamics of the particular cases when system has one or two equilibrium points. Making a time rescaling we obtain a polynomial dierential equations system topologically equivalent to original one and we prove that for certain subset of parameters, the model exhibits biestability phenomenon, since there exists an stable limit cycle surrounding two singularities of vector eld one of these stable. We prove that there are conditions on the parameter values for which the unique equilibrium point at the rst quadrant is stable and surrounded by two limit cycles, the innermost unstable and the outhermost stable. Also we show the existence of separatrix curves on the phase plane that divide the behavior of the trajectories, which have dierent limit sets, and we have that solutions are highly sensitives to initial conditions. However, the populations always coexist since the singularity (0, 0) is a nonhyperbolic saddle point.

1. Introduction In this work we present the rst part of analysis of a deterministic continuous predator-prey model considering two important aspects for describe the interaction: a logistic type growth function for predator and a functional response of predators of non-monotonic type. The rst aspect characterize
Partially Partially

supported by Fondecyt project 1040833 and DI. PUCV project 124794/2004 supported by DI. PUCV project 124704/2006 359

360

Leslie type predator-prey models23 or logistic predator prey model6,36 or Leslie-Gower model4,21, in which the conventional environmental carrying capacity Ky is a function of the prey quantity x, that is, dependent of the available resources27 . Here we consider that Ky = nx, proportional to prey abundance as in the May-Holling-Tanner model3,32. Although Leslie models can leads anomalies in their predictions 36, because it predicts that even at very low prey density, when the consumption rate by individual predator is essentially zero, predator population can nevertheless increase, if predator prey ratio is very small36, this models are recently employed18. This form of modeling diers from a more common Gause type model12,16,44 in which the predator equation is based in mass action principle, because the numerical response is dependent of functional response. It is well known that functional response of predators or consumption rate function refers to the change in the density prey attached per unit time per predator as the prey density changes12,31. In most predator-prey models considered in the ecological literature, the predator response to prey density is assumed to be monotonic increasing, the inherent assumption being that the more prey in the environment, the better o the predator 13. However, there is evidence that indicates that this need not always be the case, for instance when a type of antipredator behavior (APB) exists. Group defence is one of this, and the term is used to describe the phenomenon whereby predators decrease, or even prevented altogether, due to the increased ability of the prey to better defend or disguise themselves when their number are large enough13,31,40,41 and in this case a non-monotonic functional response is better. For example, lone musk ox can be successfully attached by wolves, however large herds of them can be attached but with rare success. An another manifestation of APB in which a non-monotonic functional response can be used, is the phenomenon of aggregation, a social behavior of prey, in which prey congregate on a ne scale relative to the predator, so that the predators hunting is not spatially homogeneous35, such as succeed with miles long schools of certain class of shes. In this case, a primary advantage of schooling seems to be confusion of predator when it attacks. The more important benets of aggregation than group defence is an increased in wariness. Moreover, aggregation can be both decrease in vulnerability to attack and increase the time group member can devote to activities other than surveillance35. Also related examples of non-monotone consumption occur at the mi-

361

crobial level where evidence indicates that when faced with overabundance of nutrient the eectiveness of the consumer can begin to decline. This is often seen when micro-organism are used for waste discomposing or for water purication, phenomenon that is called inhibition 13,31,41. The functional response curves, in particular a non-monotonic curve that it shows in Figure 1, has an upper bound on the rate of predation per individual predator at some prey density, in contrast to the old LotkaVolterra model which had assumed a linear relationship between prey density and the rate of predation over the entire range of prey densities35 .

h( x)

x
Figure 1. A non-monotonic functional response.

Here we use the function h(x) = qx/(x2 + a) also employed and correspond to Holling type IV functional response35 in which is generalized as h(x) = qx/(x2 + bx + a) in7,43,44 for a Gause model. This generalized expression is derived by Collings10, which arms that this type of functional response seems a reasonable possibility if it is assumed that prey and webbing densities are directly related. In19 it is afrmed that the corresponding Gause model has an unique stable limit cycle, but in44, it is proved the existence of two limit cycle. Moreover the system exhibit bifurcation of cusp-type with codimension two3,17 or BogdanovicTakens bifurcation22,43,44. The same model is modied in 25 considering delay and in28 for a reaction diusion system. The model here studied is partially analyzed by Collings10 and Li and Xiao24. Collings employs a computer program to determine the behavior of system. In24, the same model is analyzed in R+ = {(x, y) : x > 0, y 0}, 2 since it is not well-dene at x = 0. Using a topologically equivalent polynomial system2,8,33 to original one, we show that the singularity (0, 0), is a
14,15,31,40,41,42

362

nonhyperbolic saddle point. We determine conditions for existence of one, two and three equilibrium points or singularities at interior of the rst quadrant and we demonstrate that when a unique positive singularity there exist, then a subset of parameters there is for which two (innitesimal) limit cycles can surround these singularity (multiple Hopf bifurcation), the innermost unstable and the outhermost stable which it answers partially the question formulated in9. In24, by simulation is shown the possibility of existence of two limit cycles, one of this surrounding both singularities and the other around only one of them. We prove that this last limit cycle is non-innitesimal5, and it can not obtained by Hopf bifurcation. Then, it can coexist an repellor node with an unstable limit cycle that appear because an homoclinic of the other singularity is broken and a stable limit cycle surrounding both singularities. 2. The Model In this work we denote by x = x(t) and y = y(t) the prey and predator population size of respectively (measured in biomass, density or number), assuming that they vary continuously with time, uniform distribution over space, neither age or sex structure and as variables as parameters are of the deterministic nature. We consider that the functional response is nonmonotonic and then the Leslie type model (or Leslie-Gower model)4,6,21,23 is expressed by the following autonomous bidimensional dierential equations system: X :
dx dt dy dt

= r (1 = s 1

x K) y n x

q y a + x2

(1)

System (1) is of Kolmogorov type12,26 and all the parameters are positives, that is, = (r, K, a, q, s, n) R6 with a < K and have the following + meanings: r and s are the intrinsic growth rates or biotic potential of the prey and predators respectively. K is the prey environment carrying capacity. q is the predator maximum consumption rate per capita, i.e., the maximum number of prey that necessary can be eaten by a predator in each time unit.