 Research
 Open Access
 Published:
Existence and stability of positive steadystate solutions for a LotkaVolterrasystem with intraspecific competition
Advances in Difference Equations volume 2014, Article number: 159 (2014)
Abstract
In this paper, we investigate the existence and stability of the positivesteadystate solutions for a LotkaVolterra system with intraspecific competition byusing the LyapunovSchmidt reduction technique. To do this, we must firstly obtainthe semitrivial steady states as their base, which extend the method used in theprevious studies. Our results show that the two competition species withintraspecific competition can coexist for bigger regions of the diffusion rateμ and also complete the existing works.
MSC: 35K57.
1 Introduction
The maintenance of biodiversity has received increasing attention from ecologists andmathematicians. Moreover, resource competition is thought as an important factor indriving evolutionary diversification, in which intraspecific competition for resourcesplays a major role; see [1–4]. The classical LotkaVolterra competition system [5–8] gives a better description of the population competition, whose dynamicalbehaviors have been studied extensively; see [9–16] and the references therein. In this paper a competitive LotkaVolterradiffusion model of two slightly different species is studied, namely, the two speciesare identical except for their intraspecific competition rates. For this system, oncethe diffusion is involved biologically, it would be very interesting to find out whetherand when the two species competing for the same limited resources to survive cancoexist.
To begin with, we present the classical LotkaVolterra system with spatiallyinhomogeneous terms as follows
where Ω is a bounded domain in ${R}^{N}$ with smooth boundary ∂ Ω,μ is the diffusion rate, $u(x,t)$ and $v(x,t)$ denote the densities of two competing species, and$a(x)$ represents the intrinsic growth rate of species.
The system above shows that the two species have the same diffusion rates and intrinsicgrowth rates. However, the two species with slight difference, such as the originalspecies and the mutant, are common in biology; they correspond to the perturbed systemsof (1.1), see [17–21] for example, and are originally introduced in [17]. The two species with different diffusion rates discussed in [17] correspond to the system
where $\tau >0$. It is shown from $(u,v)\to (\theta ,0)$ (θ given in the next section) as$t\to \mathrm{\infty}$ that the slower diffuser can drive the faster one toextinction, and thus the two species cannot coexist. In [18], it is further stated that the slower diffuser may fail with a time periodicfunction $a(x,t)$ instead of $a(x)$. In [19], the model with small variations of intrinsic growth rates is given by
It is shown that the stability of the two competing species varies in a complicated wayas μ increases, which implies that the two species only coexist for smallregions of μ if they could. For $a(x)+\tau g(x)$ taking the form $a+\tau g(x)$ in [20], the authors further illustrate that the mutant can always survive andinvade, but the original species can only coexist in some cases. Furthermore, in [21] the system with different interspecific competition rates is given as
Then a new structure of coexistence states is obtained and the two species can coexisteven for bigger regions of μ.
Motivated by the studies above, in this paper, we continue the analytic works foranother perturbation system
where τ is a small positive constant (i.e.,$\tau \ll 1$) and $g(x)$, $h(x)$ are smooth functions, which indicates that the two speciesonly have different intraspecific competition rates. Our main purpose is to study theexistence and stability of the positive steadystate solutions (that is, the coexistencestates) of (1.2) by using the LyapunovSchmidt reduction technique, the implicitfunction theorem combined with finite covering theorem and the perturbation theory forcompleting the previous studies. To do this, however, we must firstly get thesemitrivial steadystate solutions of (1.2) according to [22], which is different from the corresponding results already known, and extendthe method used in the existing works. The main results we obtained show that the twocompetition species with different intraspecific competition can also coexist for biggerregions of the diffusion rate μ, but it is also important to caution thatthe conclusions for stability obtained in this paper are contrary to the ones analyzedin [21].
The rest of this paper is organized as follows. In Section 2, from [22] we firstly give the semitrivial steadystate solutions of (1.2), and thenanalyze the stability of the solutions. In Section 3, on the basis of thesemitrivial solutions obtained in Section 2, we investigate the existence andnonexistence of positive steadystate solutions of (1.2) by using the LyapunovSchmidtreduction technique, the implicit function theorem and the finite covering theorem. InSection 4, combining stability theory with perturbation theory, we further discussthe stability of positive steadystate solutions in detail.
2 Existence and stability of semitrivial steadystate solutions
In this section we firstly establish the existence of semitrivial steadystatesolutions of (1.2) by [22], whose proof is omitted. Then we simply analyze the stability of thesemitrivial solutions.
Lemma 2.1 (see [22])
Suppose that λ is a positive parameter and thateither$\beta (x)>0$on an open subset of ∂ Ω, orthe boundary condition is a Dirichlet condition on partof ∂ Ω, or${\int}_{\mathrm{\Omega}}m(x)\phantom{\rule{0.2em}{0ex}}dx<0$holds. The principaleigenvalue${\sigma}_{1}$of
is positive if and only if$0<{\lambda}^{+}<\lambda $, where${\lambda}^{+}$is the positive principal eigenvalue of
If$\beta (x)\equiv 0$and${\int}_{\mathrm{\Omega}}m(x)\phantom{\rule{0.2em}{0ex}}dx>0$, then${\sigma}_{1}>0$for all$\lambda >0$.
Consider the following system:
where $d(x)\in {C}^{1+\alpha}(\overline{\mathrm{\Omega}})$, $d(x)\ge {d}_{0}>0$, $\beta (x)\in {C}^{1+\alpha}(\overline{\mathrm{\Omega}})$, $\beta (x)\ge 0$, $f(x,u)$ is Lipschitz in u and is a measurable function inx which is bounded if u is restricted to a bounded set and$f(x,0)=0$.
Lemma 2.2 (see [22])
Suppose that$f(x,u)=q(x,u)u$with$q(x,u)$of class${C}^{2}$in u and${C}^{\alpha}$in x, and thereexists$K>0$such that$q(x,u)<0$for$u>K$. If the principaleigenvalue${\sigma}_{1}$is positive in the problem
then (2.1) has a minimal positive equilibrium${u}^{\ast}$, and all solutions to (2.1) that areinitially positive on an open subset of Ω are eventually bounded byorbits which increase toward${u}^{\ast}$as$t\to \mathrm{\infty}$.
Lemma 2.3 (see [22])
Suppose that the hypotheses of Lemma 2.2 are satisfied andthat$f(x,u)=q(x,u)u$with$q(x,u)$strictly decreasingin u for$u\ge 0$. Then the minimal positiveequilibrium${u}^{\ast}$is the only positive equilibrium for (2.1).
For the further discussions, we make the following assumption on $a(x)$.

(H)
$a(x)$ is Hölder continuous on $\overline{\mathrm{\Omega}}$ and ${\int}_{\mathrm{\Omega}}a(x)\phantom{\rule{0.2em}{0ex}}dx>0$.
Then, from Lemmas 2.12.3, it is well known that
has only a positive solution $\theta (x,\mu )$ denoted by $\theta (\mu )$ when the condition (H) is valid, and it is also easilyfound that
and
respectively have unique positive equilibriums $\tilde{u}(x,\mu ,\tau )$ and $\tilde{v}(x,\mu ,\tau )$, denoted by $\tilde{u}(\mu ,\tau )$ and $\tilde{v}(\mu ,\tau )$, for (H) and $\tau \ll 1$. Hence, we can find the following result.
Theorem 2.1 Assume that (H) holds and$\tau \ll 1$. Then (1.2) has semitrivialsteadystate solutions$(\tilde{u}(\mu ,\tau ),0)$and$(0,\tilde{v}(\mu ,\tau ))$.
To discuss the stability of semitrivial steadystate solutions $(\tilde{u}(\mu ,\tau ),0)$ and $(0,\tilde{v}(\mu ,\tau ))$, we need to consider the eigenvalue problem
corresponding to the solution $(u,v)$. Due to [23, 24], one can show that (2.3) has a principal eigenvalue $\lambda (\mu ,\tau )$, which is real, algebraically simple, and all othereigenvalues have their real parts less than $\lambda (\mu ,\tau )$. Therefore, the stability of $(\tilde{u}(\mu ,\tau ),0)$ is decided by the principal eigenvalue$\stackrel{\u02c6}{\lambda}(\mu ,\tau )$ of the problem
Similarly, the stability of $(0,\tilde{v}(\mu ,\tau ))$ is dependent on the principal eigenvalue$\tilde{\lambda}(\mu ,\tau )$ of the problem
Theorem 2.2 Suppose that the hypotheses of Theorem 2.1 arevalid. If$\stackrel{\u02c6}{\lambda}(\mu ,\tau )<0$ ($\stackrel{\u02c6}{\lambda}(\mu ,\tau )>0$), then$(\tilde{u}(\mu ,\tau ),0)$is asymptotically stable (unstable).If$\tilde{\lambda}(\mu ,\tau )<0$ ($\tilde{\lambda}(\mu ,\tau )>0$), then$(0,\tilde{v}(\mu ,\tau ))$is asymptotically stable (unstable).
Remark 2.1 In this section, we only give a simple conclusion for the stability ofsemitrivial steadystate solutions of (1.2), whose detailed proofs are analogous toSection 3 of [21].
3 Existence and nonexistence of positive steadystate solutions
The research on the steadystate solutions in a competitiondiffusion system is always ahot issue. In this section we establish the existence and nonexistence of positivesteady states of (1.2). For this purpose, we must make use of the semitrivial solutionsobtained in Section 2 for the following results.
To discuss the steady states of (1.2), we deal with the elliptic system corresponding to(1.2), which takes the form as follows
For convenience of the following discussions, we denote
and give the Sobolev spaces
where $p>N$ such that ${W}^{2,p}\hookrightarrow {C}^{1}(\overline{\mathrm{\Omega}})$.
According to (2.2), it is obvious that (3.1) has nontrivial nonnegative solutions${\mathrm{\Upsilon}}_{\mu}=\{(s\theta (x,\mu ),(1s)\theta (x,\mu )):s\in [0,1]\}$ for $\tau =0$. Then for $\tau \ll 1$ we will look for the solutions of (3.1) near${\mathrm{\Upsilon}}_{\mu}$, namely, the steady states of (1.2).
Theorem 3.1 Suppose that$G(\mu )$and$H(\mu )$have no common roots, andset${\mu}_{1}$and${\mu}_{2}$are the consecutive and simple rootsof$G(\mu )H(\mu )$.

(i)
If$G(\mu )H(\mu )<0$in$({\mu}_{1},{\mu}_{2})$, then there exists$\delta >0$such that for$\mu \in ({\mu}_{1}+\delta ,{\mu}_{2}\delta )$such that (1.2) has no positive steadystate solutions other than semitrivial ones for$\tau \ll 1$.

(ii)
If$G(\mu )H(\mu )>0$in$({\mu}_{1},{\mu}_{2})$, then for any sufficiently small$\delta >0$and$\tau \ll 1$, (1.2) has the positive steadystate solutions$(u(\mu ,\tau ),v(\mu ,\tau ))$satisfying$(\mu ,u(\mu ,\tau ),v(\mu ,\tau ))$in the neighborhood of$({\mu}_{1}\delta ,{\mu}_{2}+\delta )\times {\mathrm{\Upsilon}}_{\mu}$besides semitrivial ones. Here
$$\begin{array}{r}u(\mu ,\tau )=s(\mu ,\tau )[\theta (\mu )+{\tilde{y}}_{1}(\mu ,\tau )],\\ v(\mu ,\tau )=[1s(\mu ,\tau )][\theta (\mu )+{\tilde{z}}_{1}(\mu ,\tau )]\end{array}$$(3.2)
with$s(\mu ,0)=s(\mu ):=\frac{H(\mu )}{G(\mu )+H(\mu )}$, ${\tilde{y}}_{1}(\mu ,0)=0$and${\tilde{z}}_{1}(\mu ,0)=0$. Moreover, the positive steadystatesolution branches$(u(\mu ,\tau ),v(\mu ,\tau ))$connect with the semitrivialones$(u(\underline{\mu}(\tau ),\tau ),v(\underline{\mu}(\tau ),\tau ))$and$(u(\overline{\mu}(\tau ),\tau ),v(\overline{\mu}(\tau ),\tau ))$, and the smooth functions$\underline{\mu}(\tau )$and$\overline{\mu}(\tau )$are defined on$\{\tau :\tau \ge 0\}$such that$\underline{\mu}(0)={\mu}_{1}$and$\overline{\mu}(0)={\mu}_{2}$.
Proof For $\tau \ll 1$, our purpose is to find the positive solutions of (3.1)near ${\mathrm{\Upsilon}}_{\mu}$. Following the LyapunovSchmidt reduction technique [25], we know that $X={X}_{1}\oplus {X}_{2}$ on the basis of the spaces defined above. Then we can set
that is to say, the solution form we will look for is given by
where $s\in [0,1]$ and $(y,z)\in {X}_{2}$ near $(0,0)$.
Substituting the expression above into (3.1), we have
where
and the map $F:{X}_{2}\times ({\mu}_{1}{\delta}_{1},{\mu}_{2}+{\delta}_{1})\times ({\delta}_{1},{\delta}_{1})\times ({\delta}_{1},1+{\delta}_{1})\to Y$ defined above is a smooth function. Obviously, on thebasis of the semitrivial steady states obtained in Section 2, we have
Let $L(\mu ,s)={D}_{(y,z)}F(0,0,\mu ,0,s)\in \mathcal{L}(X,Y)$. Then
which is denoted by L, and L is a Fredholm operator of index zerosince X is compactly imbedded in Y. It is easy to check that$(\theta ,\theta )$ is in the kernel of L. Due to$\theta >0$, we know that 0 is a simple eigenvalue of L,which leads to
Define the projection operator $P=P(\mu ,s):Y\to {X}_{1}$ as
It is easily found that
which results in
According to the LyapunovSchmidt reduction technique, the equivalent expression of(3.3) is
We know that L is an isomorphic mapping from ${X}_{2}$ to Y, then, by applying the implicit theorem tothe second equation of (3.7), it can be solved to get a unique solution$(y,z)=(y(\mu ,\tau ,s),z(\mu ,\tau ,s))$ near $(0,0)$. Furthermore, combining the finite covering theorem, thereexists ${\delta}_{2}>0$ such that
and
Thus, the solvability of (3.3) is converted to that $(y(\mu ,\tau ,s),z(\mu ,\tau ,s))$ satisfies the first equation of (3.7), that is,
Combining with (3.4), we have
On the basis of the definition of P, we can get a smooth function$\xi (\mu ,\tau ,s)$ satisfying
Hence we need to solve $\xi (\mu ,\tau ,s)=0$. From (3.8) and (3.9), we can find
and then there exists a smooth function ${\xi}_{1}(\mu ,\tau ,s)$ such that
Moreover, we know that the solutions $s=0$ and $s=1$ correspond to the semitrivial solutions$(0,\tilde{v})$ and $(\tilde{u},0)$ of (3.1), respectively. Consequently, it remains to solve${\xi}_{1}(\mu ,\tau ,s)=0$.
From (3.11), it follows that
Differentiate (3.10) with respect to τ, set $\tau =0$ and combine with (3.6) and (3.8), which leads to
From (3.3), we obtain
and then it follows from (3.5) that
Hence, according to (3.12)(3.14), we have
For $G(\mu )H(\mu )<0$ in $({\mu}_{1},{\mu}_{2})$, for example, we let $G(\tilde{\mu})H(\tilde{\mu})<0,\tilde{\mu}\in ({\mu}_{1},{\mu}_{2})$, and then we obtain ${\xi}_{1}(\tilde{\mu},0,s)\ne 0$ when $\tilde{\mu}\in ({\mu}_{1},{\mu}_{2})$ and $s\in [0,1]$. By using the finite covering theorem, there exists${\delta}_{3}>0$ taken smaller if necessary such that${\xi}_{1}(\mu ,\tau ,s)=0$ has no solution for $(\mu ,\tau ,s)\in ({\mu}_{1}+{\delta}_{3},{\mu}_{2}{\delta}_{3})\times ({\delta}_{3},{\delta}_{3})\times ({\delta}_{3},1+{\delta}_{3})$. This results in statement (i) of Theorem 3.1.
For $G(\mu )H(\mu )>0$ in $({\mu}_{1},{\mu}_{2})$, we can see that ${\xi}_{1}(\tilde{\mu},0,s(\tilde{\mu}))=0$ with $s(\mu )=\frac{H(\mu )}{G(\mu )+H(\mu )}$ and ${\xi}_{1,s}(\tilde{\mu},0,s(\tilde{\mu}))=\frac{G(\tilde{\mu})+H(\tilde{\mu})}{{\int}_{\mathrm{\Omega}}{\theta}^{2}(x,\tilde{\mu})\phantom{\rule{0.2em}{0ex}}dx}\ne 0$ for $\tilde{\mu}\in [{\mu}_{1},{\mu}_{2}]$ and $s\in [0,1]$ since $G(\mu )$ and $H(\mu )$ have no common roots. Combining the implicit functiontheorem with the finite covering theorem, we have that for ${\delta}_{4}>0$ chosen yet smaller if necessary and$(\mu ,\tau ,s)\in ({\mu}_{1}{\delta}_{4},{\mu}_{2}+{\delta}_{4})\times ({\delta}_{4},{\delta}_{4})\times ({\delta}_{4},1+{\delta}_{4})$, the equation ${\xi}_{1}(\mu ,\tau ,s)=0$ has only solution given by the smooth function$s=s(\mu ,\tau )$ with $s(\mu ,0)=s(\mu )$, which shows that (3.3) has solutions$(y,z)=(y(\mu ,\tau ,s),z(\mu ,\tau ,s))$ with $s=s(\mu ,\tau )$. Due to $y(\mu ,\tau ,0)=0$ and $z(\mu ,\tau ,1)=0$ given in (3.9), we see
and then denote ${y}_{1}(\mu ,\tau ,s(\mu ,\tau ))$ and ${z}_{1}(\mu ,\tau ,s(\mu ,\tau ))$ by ${\tilde{y}}_{1}(\mu ,\tau )$ and ${\tilde{z}}_{1}(\mu ,\tau )$, respectively. From (3.8), it follows that${\tilde{y}}_{1}(\mu ,0)=0$ and ${\tilde{z}}_{1}(\mu ,0)=0$. Therefore, for $(\mu ,\tau ,s)\in ({\mu}_{1}{\delta}_{4},{\mu}_{2}+{\delta}_{4})\times ({\delta}_{4},{\delta}_{4})\times ({\delta}_{4},1+{\delta}_{4})$, the equation $\xi (\mu ,\tau ,s)=0$ has solutions given by $s=0$, $s=1$ and $s=s(\mu ,\tau )$, which shows that (1.2) has not only semitrivialsteadystate solution branches, but also positive branches given by (3.2) which meet thesemitrivial ones whose form is discussed below.
Clearly, $s(\mu ,\tau )=0$ and $s(\mu ,\tau )=1$ correspond to the semitrivial stationary solution of(1.2). For $\mu ={\mu}_{1}$, there is either $G({\mu}_{1})=0$ or $H({\mu}_{1})=0$ due to that $G(\mu )$ and $H(\mu )$ have no common roots. Without loss of generality, we take$G({\mu}_{1})=0$, and then $1s({\mu}_{1},0)=1s({\mu}_{1})=0$. Moreover, ${[1s(\mu ,\tau )]}_{\mu}{}_{(\mu ,\tau )=({\mu}_{1},0)}={[1s(\mu ,0)]}_{\mu}{}_{\mu ={\mu}_{1}}=\frac{{G}^{\prime}({\mu}_{1})}{H({\mu}_{1})}\ne 0$ since ${\mu}_{1}$ is the simple root of $G(\mu )H(\mu )$. Thus, by the implicit function theorem, we can obtain aunique solution of $1s(\mu ,\tau )=0$ defined by a smooth function $\mu =\underline{\mu}(\tau )$ for $\tau \in [0,{\delta}_{5})$ with ${\delta}_{5}>0$ small enough and $\underline{\mu}(0)={\mu}_{1}$, which implies that the corresponding semitrivialsteadystate solution of (1.2) can be described by $(u(\underline{\mu}(\tau ),\tau ),v(\underline{\mu}(\tau ),\tau ))=(\tilde{u}(\underline{\mu}(\tau ),\tau ),0)$. Similarly, for the case $H({\mu}_{1})=0$, we can get only the solution of $s(\mu ,\tau )=0$ still given by $\mu =\underline{\mu}(\tau )$, which corresponds to the semitrivial steadystatesolution $(u(\underline{\mu}(\tau ),\tau ),v(\underline{\mu}(\tau ),\tau ))=(0,\tilde{v}(\underline{\mu}(\tau ),\tau ))$ of (1.2). Moreover, for $\mu ={\mu}_{2}$, similar results hold true, and the unique solution of$1s(\mu ,\tau )=0$ or $s(\mu ,\tau )=0$ is denoted by $\mu =\overline{\mu}(\tau )$ with $\overline{\mu}(0)={\mu}_{2}$. Thus the relevant solution of (1.2) is given by$(u(\overline{\mu}(\tau ),\tau ),v(\overline{\mu}(\tau ),\tau ))=(\tilde{u}(\overline{\mu}(\tau ),\tau ),0)$ or $(u(\overline{\mu}(\tau ),\tau ),v(\overline{\mu}(\tau ),\tau ))=(0,\tilde{v}(\overline{\mu}(\tau ),\tau ))$. □
Remark 3.1 The proof of Theorem 3.1 shows the existence of loops or branchesof positive steady states just as [21], whose details are omitted here.
4 Stability of positive steadystate solutions
In this section, we analyze the stability of positive steady states$(u(\mu ,\tau ),v(\mu ,\tau ))$ of (1.2). For $\tau =0$, we know $(u(\mu ,0),v(\mu ,0))=(s\theta (\mu ),(1s)\theta (\mu ))$, and then the principal eigenvalue of the correspondingeigenvalue problem (2.3) is $\lambda =0$ because of $\theta (\mu )>0$, which shows that all other eigenvalues have negative realparts. Furthermore, for $\tau \ll 1$, since all other eigenvalues also have negative real partsby the perturbation theory [26], the stability of $(u(\mu ,\tau ),v(\mu ,\tau ))$ is determined by the principal eigenvalue${\lambda}_{1}(\mu ,\tau )$ near 0 of (2.3), that is,
Firstly, for $\tau \ll 1$, we can set
in (4.1) with smooth functions ${\varphi}_{1}(\mu ,\tau )$ and ${\psi}_{1}(\mu ,\tau )$. Furthermore, the emphasis is on the sign of${\lambda}_{1}(\mu ,\tau )$. Hence, for small τ, we divide thediscussions of the sign of ${\lambda}_{1}(\mu ,\tau )$ into three cases, that is, μ close to${\mu}_{1}$, μ close to ${\mu}_{2}$ and μ bounded away from${\mu}_{1}$ and ${\mu}_{2}$. To do this, we must first prove the following lemmas.
Lemma 4.1 For$\tau \ll 1$, the principal eigenvalue${\lambda}_{1}(\mu ,\tau )$of (4.1) satisfies
Proof Multiply the first equation of (4.1) by v and integrate overΩ to get
In the same way, multiplying the second equation of (4.1) by u and integratingby parts, we can obtain
From (4.4) and (4.5), it suffices to show that (4.3) is valid. □
Furthermore, on the basis of Lemma 4.1, we can discuss the sign of${\lambda}_{1}(\mu ,\tau )$.
Lemma 4.2 For$\mu \in [{\mu}_{1}+\u03f5,{\mu}_{2}\u03f5]$, $\u03f5>0$, we have the following conclusion:
Proof It is obvious that $(u(\mu ,\tau ),v(\mu ,\tau ))\to (s(\mu )\theta (\mu ),[1s(\mu )]\theta (\mu ))$ and $(\varphi (\mu ,\tau ),\psi (\mu ,\tau ))\to (\theta (\mu ),\theta (\mu ))$ as $\tau \to {0}^{+}$ from (3.2) and (4.2). Then, for $\tau \to {0}^{+}$, it follows that
Thus, we can obtain (4.6) from (4.3), (4.7) and (4.8). □
For the discussions below, without loss of generality, we can assume the case$G({\mu}_{1})=0$ (the case $H({\mu}_{1})=0$ is similar), and then $1s(\underline{\mu}(\tau ))=0$ and $(u(\underline{\mu}(\tau ),\tau ),v(\underline{\mu}(\tau ),\tau ))=(\tilde{u}(\underline{\mu}(\tau ),\tau ),0)$.
Lemma 4.3 Assume that$G({\mu}_{1})=0$. Then we have the following conclusion:
Proof According to Theorem 3.1, we know that $(\underline{\mu}(\tau ),u(\underline{\mu}(\tau ),\tau ),v(\underline{\mu}(\tau ),\tau ))=(\underline{\mu}(\tau ),\tilde{u}(\underline{\mu}(\tau ),\tau ),0)$ is the bifurcation point of the positive solutions to(1.2). Then ${\lambda}_{1}(\underline{\mu}(\tau ),\tau )=0$ and the eigenfunction $\psi (\underline{\mu}(\tau ),\tau )$ satisfies
Multiplying the equation above by $\tilde{u}(\underline{\mu}(\tau ),\tau )$ and integrating by parts, we can get
and then combining with $v(\underline{\mu}(\tau ),\tau )=0$ it follows that $\chi (\underline{\mu}(\tau ),\tau )=0$, where
in (4.3). Thus it leads to
by the mean value theorem for $\tilde{\mu}$ between μ and $\underline{\mu}(\tau )$.
For $G({\mu}_{1})=0$, we know that $(u(\mu ,\tau ),v(\mu ,\tau ))\to (\theta ({\mu}_{1}),0)$, $(\varphi (\mu ,\tau ),\psi (\mu ,\tau ))\to (\theta ({\mu}_{1}),\theta ({\mu}_{1}))$, $({u}_{\mu}(\mu ,\tau ),{v}_{\mu}(\mu ,\tau ))\to ({s}_{\mu}({\mu}_{1})\theta ({\mu}_{1})+{\theta}_{\mu}({\mu}_{1}),{s}_{\mu}({\mu}_{1})\theta ({\mu}_{1}))$, $({\varphi}_{\mu}(\mu ,\tau ),{\psi}_{\mu}(\mu ,\tau ))\to ({\theta}_{\mu}({\mu}_{1}),{\theta}_{\mu}({\mu}_{1}))$ as $\tau \to {0}^{+}$ and $\mu \to {\mu}_{1}$. Then, differentiating (4.10) with respect toμ, we obtain
as $\tau \to {0}^{+}$ and $\mu \to {\mu}_{1}$. From (4.3), (4.11) and (4.12), we can get the relation(4.9). □
In the same way, for μ close to ${\mu}_{2}$, the sign of ${\lambda}_{1}(\mu ,\tau )$ is decided by the result below, whose details are omittedhere.
Lemma 4.4 Assume that$G({\mu}_{2})=0$. Then we have the following conclusion:
Finally, based on Lemmas 4.24.4, the main result of this section is presented asfollows.
Theorem 4.1 Suppose that$G(\mu )$and$H(\mu )$have no common roots, ${\mu}_{1}<{\mu}_{2}$are two consecutive and simple rootsof$G(\mu )H(\mu )$and$G(\mu )H(\mu )>0$in$({\mu}_{1},{\mu}_{2})$. Then, for$\tau \ll 1$and$\mu \in (\underline{\mu}(\tau ),\overline{\mu}(\tau ))$, we have:

(i)
If$G(\mu )>0$and$H(\mu )>0$in$({\mu}_{1},{\mu}_{2})$, then the positive steady states$(u(\mu ,\tau ),v(\mu ,\tau ))$of (1.2) are asymptotically stable;

(ii)
If$G(\mu )<0$and$H(\mu )<0$in$({\mu}_{1},{\mu}_{2})$, then$(u(\mu ,\tau ),v(\mu ,\tau ))$is unstable.
Proof Now, we only consider the statement (i) (the statement (ii) can be provedsimilarly). The key point in the proof of (i) is to obtain ${\lambda}_{1}(\mu ,\tau )<0$ by contradiction. Suppose that there exist${\tau}_{i}\to {0}^{+}$ and ${\mu}_{i}\in (\underline{\mu}({\tau}_{i}),\overline{\mu}({\tau}_{i}))$ such that ${\lambda}_{1}({\mu}_{i},{\tau}_{i})\ge 0$ for $i=1,2,\dots $ . By passing to the limit, we have$\underline{\mu}({\tau}_{i})\to {\mu}_{1}$, $\overline{\mu}({\tau}_{i})\to {\mu}_{2}$ and ${\mu}_{i}\to \tilde{\mu}$ as $i\to \mathrm{\infty}$ with $\tilde{\mu}\in [{\mu}_{1},{\mu}_{2}]$.
For $\tilde{\mu}\in ({\mu}_{1},{\mu}_{2})$, we know that
from Lemma 4.2 and $G(\mu )>0$, $H(\mu )>0$ in $({\mu}_{1},{\mu}_{2})$. Thus, for large i, we have${\lambda}_{1}({\mu}_{i},{\tau}_{i})<0$, which contradicts the assumption.
For $\tilde{\mu}={\mu}_{1}$ or $\tilde{\mu}={\mu}_{2}$, we still only consider the case $\tilde{\mu}={\mu}_{1}$, and then the case $\tilde{\mu}={\mu}_{2}$ can be treated similarly. In this case, we may suppose$G({\mu}_{1})=0$ (if $H({\mu}_{1})=0$, then it is analogous). It is clear that${G}^{\prime}({\mu}_{1})>0$ since $G(\mu )>0$ in $({\mu}_{1},{\mu}_{2})$ and ${\mu}_{1}$ is the simple root of $G(\mu )$. Therefore, we see
which also leads to ${\lambda}_{1}({\mu}_{i},{\tau}_{i})<0$ for large i combining with${\tau}_{i}\to {0}^{+}$ and ${\mu}_{i}\in (\underline{\mu}({\tau}_{i}),\overline{\mu}({\tau}_{i}))$. The contradiction completes the proof. □
Remark 4.1 Theorem 4.1 is contrary to the stability results in [21]. But it is shown that the two competition species with differentintraspecific competition rates can also coexist for bigger regions of μ,which is similar to [21].
References
 1.
Bertness MD: Intraspecific competition and facilitation in a northern acorn barnaclepopulation. Ecology 1989, 70: 257–268. 10.2307/1938431
 2.
Bürger R: A multilocus analysis of intraspecific competition and stabilizing selection on aquantitative trait. J. Math. Biol. 2005, 50: 355–396. 10.1007/s0028500402942
 3.
Svanbäck R, Bolnick D: Intraspecific competition drives increased resource use diversity within a naturalpopulation. Proc. R. Soc. B 2007, 274: 839–844. 10.1098/rspb.2006.0198
 4.
Lankau RA: Rapid evolutionary change and the coexistence of species. Annu. Rev. Ecol. Evol. Syst. 2011, 42: 335–354. 10.1146/annurevecolsys102710145100
 5.
Hastings A: Spatial heterogeneity and ecological models. Ecology 1990, 71: 426–428. 10.2307/1940296
 6.
Takeuchi Y: Global Dynamical Properties of LotkaVolterra Systems. World Scientific, Singapore; 1996.
 7.
Du YH: Realization of prescribed patterns in the competition model. J. Differ. Equ. 2003, 193: 147–179. 10.1016/S00220396(03)000561
 8.
Cantrell RS, Cosner C, Lou Y: Multiple reversals of competitive dominance in ecological reserves via externalhabitat degradation. J. Dyn. Differ. Equ. 2004, 16: 973–1010. 10.1007/s108840047831y
 9.
Győri I: Global attractivity in delay LotkaVolterra differential equations. Fourth International Conference on Dynamic Systems and Applications 2003, 21–24. Atlanta, 21–24 May 2003
 10.
Pao CV: Global asymptotic stability of LotkaVolterra competition systems with diffusionand time delays. Nonlinear Anal., Real World Appl. 2004, 5: 91–104. 10.1016/S14681218(03)00018X
 11.
Kuto K, Yamada Y: Positive solutions for LotkaVolterra competition systems with largecrossdiffusion. Appl. Anal. 2010, 89: 1037–1066. 10.1080/00036811003627534
 12.
Hetzer G, Nguyen T, Shen W: Coexistence and extinction in the VolterraLotka competition model with nonlocaldispersal. Commun. Pure Appl. Math. 2012, 11: 1699–1722.
 13.
Wu DY: Global attractivity of a discrete LotkaVolterra competition system with infinitedelays and feedback controls. Adv. Differ. Equ. 2013., 2013: Article ID 14
 14.
He XQ, Ni WM: The effects of diffusion and spatial variation in LotkaVolterracompetitiondiffusion system I: heterogeneity vs. homogeneity. J. Differ. Equ. 2013, 254: 528–546. 10.1016/j.jde.2012.08.032
 15.
He XQ, Ni WM: The effects of diffusion and spatial variation in LotkaVolterracompetitiondiffusion system II: the general case. J. Differ. Equ. 2013, 254: 4088–4108. 10.1016/j.jde.2013.02.009
 16.
Meng XZ, Liu R, Zhang TH: Adaptive dynamics for a nonautonomous LotkaVolterra model with sizeselectivedisturbance. Nonlinear Anal., Real World Appl. 2014, 16: 202–213.
 17.
Dockery J, Hutson V, Mischaikow K, Pernarowski M: The evolution of slow dispersal rates: a reactiondiffusion model. J. Math. Biol. 1998, 37: 61–83. 10.1007/s002850050120
 18.
Hutson V, Mischaikow K, Poláčik P: The evolution of dispersal rates in a heterogeneous timeperiodic environment. J. Math. Biol. 2001, 43: 501–533. 10.1007/s002850100106
 19.
Hutson V, Lou Y, Mischaikow K, Poláčik P: Competing species near the degenerate limit. SIAM J. Math. Anal. 2003, 35: 453–491. 10.1137/S0036141002402189
 20.
Hetzer G, Nguyen T, Shen W: Effects of small spatial variation of the reproduction rate in a two speciescompetition model. Electron. J. Differ. Equ. 2010., 2010: Article ID 160
 21.
Lou Y, Martínez S, Poláčik P: Loops and branches of coexistence states in a LotkaVolterra competition model. J. Differ. Equ. 2006, 230: 720–742. 10.1016/j.jde.2006.04.005
 22.
Cantrell RS, Cosner C: Spatial Ecology via ReactionDiffusion Equations. Wiley, Chichester; 2003.
 23.
Deimling K: Nonlinear Functional Analysis. Springer, Berlin; 1985.
 24.
Hess P: Periodic Parabolic Boundary Value Problems and Positivity. Longman, Harlow; 1985.
 25.
Golubitsky M, Schaeffer D I. In Singularities and Groups in Bifurcation Theory. Springer, New York; 1985.
 26.
Kato T: Perturbation Theory for Linear Operators. Springer, New York; 1985.
Acknowledgements
The authors would like to express sincere thanks to the anonymous referee for his/hercarefully reading the manuscript and valuable comments on this paper. The work issupported by the Education Committee Foundation of Shaanxi Province, and the NationalNatural Science Foundation of China (No. 11271236).
Author information
Affiliations
Corresponding author
Additional information
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MW performed the theory analysis and carried out the computations. JC participated inthe design of the study and helped to draft the manuscript. LQ conceived of the studyand participated in some computations. All authors have read and approved the finalmanuscript.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0), which permits use, duplication, adaptation, distribution, and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Wei, M., Chang, J. & Qi, L. Existence and stability of positive steadystate solutions for a LotkaVolterrasystem with intraspecific competition. Adv Differ Equ 2014, 159 (2014). https://doi.org/10.1186/168718472014159
Received:
Accepted:
Published:
Keywords
 steadystate solutions
 LyapunovSchmidt reduction technique
 existence
 stability