Heterogeneous carrying capacities and global extinction in metapopulations
Jakub Hesoun, Petr Stehlík · 2024
Read Heterogeneous carrying capacities and global extinction in metapopulations by Jakub Hesoun, Petr Stehlík (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-17461.txt.
Introduction
Jakub Hesoun and Petr Stehlík study when a species spread over two patches of land dies out everywhere. Their paper appeared in 2024. Their model mixes movement between patches with growth that has a strong Allee effect, meaning small populations shrink. The two patches hold different amounts of the species at best. With equal patches, the count of steady states ranges from three to nine depending on movement strength.
The authors give conditions on movement and growth under which extinction is the only steady state, and globally stable. They test the result against computer simulations and against several growth laws, for robustness. The paper adds proof where the extinction literature had mostly numbers. They close by linking the result to habitat breakup and the paradox of perfect mixing.
Catalog record
| CREATOR | Jakub Hesoun, Petr Stehlík |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 17 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-17461
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Heterogeneous carrying capacities and global extinction in metapopulations Jakub Hesoun, Petr Stehlík Abstract In this paper we consider a simple two patch reaction diffusion model with strong Allee effect, sufficiently distinct carrying capacities, similar reaction strengths, and strong diffusion. In the homogeneous case, i.e., in in the case of equal or similar capacities and reaction strengths, it is well known that the number of stationary solutions ranges from three (strong diffusion) to nine (weak diffusion). We provide sufficient conditions which includes the diffusion strength and reaction parameters that ensure that the extinction point is the unique and globally asymptotically stable equilibrium in the case of heterogeneous capacities. For the sake of robustness we consider several bistable reaction functions, compare our analytical result with numerical simulations, and conclude the paper with a short discussion on global extinction literature (which has provided mostly numerical results so far), as well as other related phenomena, e.g., fragmentation, the perfect mixing paradox, and the natural form the reaction diffusion patch models. MSC 37N25 , 39A12 , 92C42. 1 Introduction In this paper we consider a simple two-patch bistable reaction-diffusion model with heterogeneous capacities k_{1}\neq k_{2} \begin{cases}\displaystyle x_{1}^{\prime}=D(x_{2}-x_{1})+\lambda_{1}x_{1}\left(1-\frac{x_{1}}{k_{1}}\right)\left(\frac{x_{1}}{k_{1}}-\frac{1}{2}\right),\\ \displaystyle x_{2}^{\prime}=D(x_{1}-x_{2})+\lambda_{2}x_{2}\left(1-\frac{x_{2}}{k_{2}}\right)\left(\frac{x_{2}}{k_{2}}-\frac{1}{2}\right),\end{cases} (1) where D>0 represents the diffusion parameter and \lambda_{1},\lambda_{2}>0 heterogeneous reaction strengths. Without loss of generality we assume that k_{2}\leq k_{1} . The system ( 1 ) has been extensively studied in the symmetric case k_{1}=k_{2} and \lambda_{1}=\lambda_{2} . It was shown that there exist 9 stationary solutions (6 heterogeneous and 3 homogeneous) for sufficiently small D and three homogeneous solutions (two stable ones and one unstable) once D is large enough, [ 11 , 12 , 20 ] . This result has deep consequences both for the theory of metapopulations [ 13 ] but also for patterns of discrete-space dynamical systems [ 15 ] . In contrast to this homogeneous case, our main result provides sufficient conditions on the heterogeneous capacities k_{i} , and reaction strengths \lambda_{i} which ensure that for sufficiently strong diffusion, there is a unique equilibrium – the origin which corresponds to the extinction of both populations, see also Fig. 4 . Theorem 1 . Assume that \max\{\lambda_{1},\lambda_{2}\}<4D , 2k_{2}<k_{1} , and \max\left\{\frac{2\sqrt{3}k_{1}^{2}}{9\left(k_{1}-k_{2}\right)\left(2k_{1}-k_{2}\right)},\,\frac{\sqrt{3}k_{1}^{2}}{18\left(k_{1}-2k_{2}\right)\left(k_{1}-k_{2}\right)}\right\}<\frac{\lambda_{1}}{\lambda_{2}}<\frac{9\left(k_{1}-2k_{2}\right)\left(k_{1}-k_{2}\right)}{2\sqrt{3}k_{2}^{2}}. (2) Then the system ( 1 ) has the unique stationary solution (x_{1}^{*},y_{1}^{*})=(0,0) which is globally asymptotically stable. In other words, the sufficient condition from Thm. 1 requires that the first patch must be at least twice as large, diffusion strong enough, and the reaction strengths sufficiently close (inequalities ( 2 )). Comparable conditions have been obtained for similar models numerically [ 1 , 2 , 3 ] . In Sec. 2 we reduce the number of parameters in the model ( 1 ) and provide some straightforward results on the number of stationary solutions for similar capacities. In Sec. 3 we prove Thm. 1 and in Sec. 4 we generalize Thm. 1 to general bistabilities with arbitrary viability constants and provide explicit necessary and sufficient condition for sawtooth caricature. We conclude with a brief discussion on the optimality of our results and literature in Sec. 5 . 2 Preliminaries We first observe, that for small D>0 we have 9 stationary solutions. Theorem 2 . If D>0 is sufficiently small then the problem ( 1 ) has 9 nonnegative stationary solutions. Proof. The results follows by a standard observation that for D=0 there are 9 stationary solutions and the application of the implicit function theorem which is enabled by nonzero derivatives of the cubic f in 0,1/2,1 . See, e.g., [ 20 ] for more details. ∎ Using x=x_{1}/k_{1},y=x_{2}/k_{2},\tau=Dt , we reduce the number of parameters in ( 1 ) and consider the system \begin{cases}\displaystyle x^{\prime}=\gamma y-x+\alpha f(x),\\ \displaystyle y^{\prime}=\frac{1}{\gamma}x-y+\beta f(y).\end{cases} (3) where f(s):=s(1-s)(s-1/2) and \gamma=\frac{k_{2}}{k_{1}}\in(0,1],\quad\alpha=\frac{\lambda_{1}}{D}>0,\quad\beta=\frac{\lambda_{2}}{D}>0. (4) Figure 1 : Illustration of Lemma 3 . Nullclines \mathcal{N}_{x} and \mathcal{N}_{y} from ( 5 ) for \alpha=6 , \beta=8 Figure 2 : A priori estimates \widehat{\Omega}_{i} from Lemma 4 (light gray regions) and Lemma 5 (the hatched subsets \Omega_{i}\subset\widehat{\Omega}_{i} ). Next, we show that there is always a nontrivial equilibrium once \gamma\in[1/2,1) , i.e., in the case in which the capacities do not differ too much. Lemma 3 . Let \alpha,\beta>0 , then the problem ( 3 ) has • at least three nonnegative stationary solutions if \gamma\in(1/2,1) , • at least two nonnegative stationary solutions if \gamma=1/2 . Proof. For \gamma=1/2, we have a non-trivial solution x^{*}=1/2,y^{*}=1 . For \gamma<1/2 , stationary solutions of ( 3 ) lie on intersections of continuous nullclines \mathcal{N}_{x} , \mathcal{N}_{y} given by \mathcal{N}_{x}:y=\frac{x}{\gamma}-\alpha\frac{f(x)}{\gamma}=:\nu_{x}(x),\quad\mathcal{N}_{y}:y=\frac{x}{\gamma}+\beta f(y). (5) Comparing the positions of \mathcal{N}_{x} , \mathcal{N}_{y} with respect to the line \ell:y=\frac{x}{\gamma} , we observe that \mathcal{N}_{x} lies above \ell if x\in(0,1/2)\cup(1,\infty) and \mathcal{N}_{y} lies above the line \ell if y\in(1/2,1) . Consequently, there are at least three intersections of \mathcal{N}_{x} and \mathcal{N}_{y} . These three intersections (x_{i}^{*},y_{i}^{*}) , i=1,2,3 satisfy (due to continuity) x_{1}^{*}=y_{1}^{*}=0 , y_{2}^{*}\in(1/2,1) , and x_{3}^{*}\in(\gamma,1) . Since for x>\gamma,\ \mathcal{N}_{y} is located above the line y=1 , we have that y_{3}^{*}>y_{2}^{*} , see Fig. 2 . ∎ 3 Global extinction In this section we consider \gamma\in(0,1/2) and prove our main result, Thm. 1 . We begin with auxiliary a priori estimates on nontrivial solutions, see Fig. 2 . Lemma 4 . Let \alpha,\beta>0 , \gamma\in(0,\,1/2) and (x^{*},y^{*}) be a non-trivial stationary solution of ( 3 ). Then (x^{*},y^{*})\in\widehat{\Omega}_{1}\cup\widehat{\Omega}_{2}\cup\widehat{\Omega}_{3} , where \displaystyle\begin{split}\widehat{\Omega}_{1}&:=\left\{(x,\,y)\in\mathbb{R}^{2}:x\in(0,\,\gamma),\,y\in\left(\frac{1}{2},\,1\right),\,y>\frac{x}{\gamma}\right\},\\ \widehat{\Omega}_{2}&:=\left\{(x,\,y)\in\mathbb{R}^{2}:x\in\left(\frac{1}{2},\,1\right),\,y\in\left(1,\,\frac{x}{\gamma}\right)\right\},\\ \widehat{\Omega}_{3}&:=\left\{(x,\,y)\in\mathbb{R}^{2}:x\in\left(\frac{1}{2},\,1\right),\,y\in\left(0,\,\frac{1}{2}\right)\right\}.\end{split} (6) Proof. The stationary solution of ( 3 ) has to comply with ( 5 ) which implies that \gamma y-x=-\alpha f(x)=\beta\gamma f(y). (7) Since \alpha,\beta,\gamma>0 the values of f(x) and f(y) must have opposite signs. We distinguish between two cases. First, let us assume that f(x)<0 and f(y)>0 . This implies that y\in(1/2,1) and x<\gamma , since 0<\gamma y-x<\gamma-x This defines the set \widehat{\Omega}_{1} . Next, let f(x)>0 which leads to x\in(1/2,1) and f(y)<0 . We have two subcases. If y\in(0,1/2) we get the set \widehat{\Omega}_{3} . Similarly we arrive to \widehat{\Omega}_{2} by considering y>1 and observing that y<x/\gamma from ( 7 ). ∎ Once we consider small values of \alpha,\beta (i.e., large diffusion D in the original problem ( 1 )) we can further reduce the regions \widehat{\Omega}_{i} from ( 6 ), see Fig. 2 . Lemma 5 . Let \alpha,\beta\in(0,4) , \gamma\in(0,1/2) and (x^{*},y^{*}) be a non-trivial stationary solution of ( 3 ). Then (x^{*},y^{*})\in\Omega_{1}\cup\Omega_{2} , where \displaystyle\begin{split}\Omega_{1}&=\left\{(x,\,y)\in\mathbb{R}^{2}:x\in\left(\frac{\gamma}{2},\,\gamma\right),\,y\in\left(\frac{x}{\gamma},\,1\right)\right\},\\ \Omega_{2}&=\left\{(x,\,y)\in\mathbb{R}^{2}:x\in\left(\frac{1}{2},\,1\right),\,y\in\left(\frac{1}{2\gamma},\,\frac{x}{\gamma}\right)\right\}.\end{split} (8) Proof. The concavity of f(s) on (1/2,1) yields f(s)\leq\frac{1}{4}\left(s-\frac{1}{2}\right) and we can rewrite ( 5 ) into \displaystyle\displaystyle y \displaystyle=\frac{x-\alpha f(x)}{\gamma}\geq\frac{x-\frac{\alpha}{4}\left(x-\frac{1}{2}\right)}{\gamma}=\frac{x\left(1-\frac{\alpha}{4}\right)+\frac{\alpha}{8}}{\gamma}\quad\text{ if }x>\frac{1}{2}, \displaystyle\displaystyle x \displaystyle=\gamma\left(y-\beta f(y)\right)\geq\gamma\left(y-\frac{\beta}{4}\left(y-\frac{1}{2}\right)\right)=\gamma\left(y\left(1-\frac{\beta}{4}\right)+\frac{\beta}{8}\right)\quad\text{ if }y>\frac{1}{2}. If \alpha,\beta\in(0,4) we have 1-\alpha/4>0 and 1-\beta/4>0 which implies \displaystyle\displaystyle y \displaystyle>\frac{1}{2\gamma}\quad\text{ if }x>\frac{1}{2}\quad\text{ and}\quad x>\frac{\gamma}{2}\quad\text{ if }y>\frac{1}{2}. The former inequality eliminates \widehat{\Omega}_{3} and reduces \widehat{\Omega}_{2} into {\Omega}_{2} . The latter then reduces \widehat{\Omega}_{1} into {\Omega}_{1} . ∎ We next eliminate the possible location of stationary solution in \Omega_{1} . Lemma 6 . Let \alpha,\beta\in(0,4) , \gamma\in(0,1/2) and assume \displaystyle\alpha>\beta\cdot\max\left\{\frac{2\sqrt{3}}{9\left(1-\gamma\right)\left(2-\gamma\right)},\,\frac{\sqrt{3}}{18\left(1-2\gamma\right)\left(1-\gamma\right)}\right\}, (9) then the system ( 3 ) has no stationary solution located in \Omega_{1} defined by ( 8 ). Proof. The cubic f(x) satisfies f(x)\leq\frac{\sqrt{3}}{36} for x\in(0,1) . Consequently, \mathcal{N}_{y} in ( 5 ) implies x=\gamma(y-\beta f(y))\geq\gamma\left(y-\beta\frac{\sqrt{3}}{36}\right)=:\Psi(y). (10) Since the nullcline \mathcal{N}_{x} in ( 5 ) satisfies \nu_{x}(\gamma/2)=\frac{1}{2}-\frac{1}{2}\alpha\left(1-\frac{\gamma}{2}\right)\left(\frac{\gamma}{2}-\frac{1}{2}\right),\quad\nu_{x}(\gamma)=1-\alpha(1-\gamma)\left(\gamma-\frac{1}{2}\right) and is concave on (\gamma/2,\gamma), we can only compare these