Extinction in a generalized Lotka-Volterra predator-prey model (Q1841235)

Following the idea of \textit{A. S. Ackleh} [Nolinear Anal., Theory Methods Appl. 33, No. 7, 729-745 (1998; Zbl 0931.34026)], the authors divide the prey and predator populations into \(M\) and \(N\) subpopulations, respectively. Assume that the growth (mortality) for the prey (predator) is subpopulation specific, while the mortality (growth) is driven by interaction with the entire predator (prey) population. Let \(x_i(t),i=1,2,\dots,M\) and \(y_j(t)\), \(j=1,2,\dots,N\) be the sizes of the \(i\)th prey subpopulation and the \(j\)th predator subpopulation, respectively. Let \(x(t)=(x_1(t),x_2(t),\dots,x_M(t))\), \(y(t)=(y_1(t),x_2(t),\dots,x_N(t))\), \(X(t)=\sum_{i=1}^Mx_i(t)\), \(Y(t)=\sum_{j=1}^Ny_j(t)\). Then the generalized predator-prey model is \[ \begin{aligned} \frac{dx_i(t)}{dt}=x_i(t)[a_i-b_iY(t)], &\quad i=1,2,\cdots,M,\\ \frac{dy_j(t)}{dt}=y_j(t)[-c_j+d_jX(t)], &\quad j=1,2,\cdots,N,\\ (x(0),y(0))\in \operatorname {int}\mathbb{R}^{M+N}_+. \end{aligned}\tag{1} \] Suppose the subpopulations are ordered such that \(\frac{a_1}{b_1}>\frac{a_k}{b_k}\), \(k=2,3,\cdots,M\) and \(\frac{d_1}{c_1}>\frac{d_k}{c_k}\), \(k=2,3,\dots,N.\) The subpopulation \(x_1\) and \(y_1\) are called dominant and the others are called nondominant subpopulations. Using the auxiliary function method, the authors show that for system (1) \[ \lim_{t\to\infty}x_k(t)=0,\quad k=2,\dots,M,\qquad \lim_{t\to\infty}y_k(t)=0, k=2,\dots,N, \] but \(x_1\) and \(y_1\) remained bounded and strictly positive as \(t\to\infty\). The numerical simulation shows that this dominant pair of subpopulations traces a trajectory in \(\operatorname {int} \mathbb{R}^2_+\) such that it becomes increasingly close to a classical Lotka-Volterra orbit.