Derivation and analysis of a discrete predator-prey model (Q2141313)

For population dynamics, sometimes discrete mathematical models are more appropriate than continuous ones, especially for species with nonoverlapping generations. A usual way to derive discrete models is to discretize continuous ones. Unfortunately, in some cases, the resulting discrete models can have negative solutions or have dynamics quite different from the continuous ones. In this paper, to avoid these drawbacks as possible, the authors formulated a discrete predator-prey model by not discretizing a continuous one. The model is based on the same assumptions as the following continuous predator-prey model, \[ x'=rx\left(1-\frac{x}{K}\right)-\alpha xy, \quad y'=-dy+\gamma xy. \tag{*} \] First, for a single species, it is assumed that its population at time \(t+1\) is a factor of the population at time \(t\), that is, \[ X_{t+1}=\frac {1+p(t)}{1+q(t)}X_t, \tag{\(\dagger\)} \] where \(p(t)\) captures the processes contributing to the increase of the population and \(q(t)\) captures the processes contributing to the decrease in the population between time steps \(t\) and \(t+1\). A possible justification of the construction of (\(\dagger\)) is as follows. Model (\(\dagger\)) is related to the continuous population model, \[ x'(t)=(p(t)-q(t))x(t), \] which has the solution \[ x(t+1)=\frac {e^{\int_t^{t+1}p(s)ds}}{e^{\int_t^{t+1}q(s)ds}}x(t). \] Using \[ \lim_{n\to\infty}\left(1+\frac{F}{n}\right)^n=e^F \] and arguing that the processes were to take place discretely (that is, \(n=1\)) yield that \(e^F\) would be replaced by \(1+F\), which produces (\(\dagger\)). Next, the derivation of (\(\dagger\)) is extended to consider the interaction of several species \(X_i\) (\(i=1\), \(2\), \(\ldots\), \(k\)), which results in \[ X_i(t+1)=\frac {1+p_i(t,X_1,X_2,\ldots,X_k)}{1+q_i(t,X_1,X_2,\ldots,X_k)}X_i(t). \] In particular, for the predator-prey model, \(X_1\) represents prey and \(X_2\) represents predator. Then, assuming that the prey population grows logistically to a carrying capacity in the absence of the predator population and the predator cannot survive in the absence of the prey population, the authors obtained the following discrete predator-prey model, \[ X_{t+1}=\frac {1+r}{1+\frac{r}{K}X_t+\alpha Y_t}X_t, \qquad Y_t=\frac {1+\gamma X_t}{1+d}Y_t.\tag{\(\ddagger\)} \] Model (*) and model (\(\ddagger\)) have the same structure of equilibria. There are at most three equilibria, \(E_0(0,0)\), \(E_K(K,0)\), and \(E^*(\frac{d}{\gamma},\frac{r(\gamma K-d)}{\alpha \gamma K})\) (only when \(\gamma K>d\)). The authors extended standard phase plane analysis by introducing the next iterate root-curve associated with the nontrivial prey nullcline. Using this curve in combination with the nullclines and direction field, they showed that the prey-only equilibrium \(E_K\) is globally asymptotic stable if \(d\ge \gamma K\), the same condition on the global stability of \(E_K\) for (*). But when \(\gamma K>d\), the coexistence equilibrium \(E^*\) of (*) is globally asymptotically stable while, for (\(\ddagger\)), \(E^*\) is locally asymptotically stable if \(d<\gamma K<1+2d\) and it is unstable if \(\gamma K>1+2d\). Moreover, model (\(\ddagger\)) undergoes a transcritical bifurcation if \(d=\gamma K\) and undergoes a supercritical Neimark-Sacker bifurcation at \(E^*\) when \(\gamma=\gamma_{\mathrm{crit}}=\frac{1+2d}{K}\).