Stability in an evolutionary game (Q1070188)

Consider a population of plants or animals, ''players'' whose phenotype (or strategy) can be represented by a real number n. If only two strategies \(n_ 0\), \(n_ 1\) are in use, denote for \(i=0,1\) the fraction of players of type \(n_ i\) at time t by \(P_ i(t)\), and let the population evolve according to the law \[ (*)\quad P_ i(t+1)\sim P_ i(t)\cdot H_ i[n_ 0,n_ 1,P_ 0(t),P_ 1(t)], \] where \(H_ i\) is the fitness of a player using \(n_ i.\) n\({}_ 0\) is called a global [resp. local] evolutionarily stable strategy (ESS) if \(P_ 0(t)\to 1\) for all \(n_ 1\neq n_ 0\) and all \(P_ 0(0)>0\) [resp. \(P_ 0(0)>1-\epsilon]\). The game is called balanced if the fitness functions can be written \(H_ i(n_ 0,n_ 1,P_ 0,P_ 1)=G[n_ 0,n_ 1,n_ i,P_ 0,P_ 1]\), and if \(G(s,s,s,P_ 0,P_ 1]\) depends only on s. A necessary condition for \(n_ 0=s\) to be a local balanced ESS is that \[ (\partial /\partial n_ i)G[s,s,s,P_ 0,P_ 1]=0\quad for\quad 1-\epsilon <P_ 0<1. \] These results are then used to determine the optimal flowering time for annual plants under competition. It is shown that the ESS flowering time is later than the time which would maximize seed production.