Replacers and their evolutionary stability in the Moran process on graphs
Evolutionary dynamics in finite structured populations are commonly modeled by the Moran Birth-death process. A key quantity is the fixation probability of a single invader attempting to take over a population of residents. A recent work introduced a new neighborhood-aware phenotype called a replacer. A replacer never wastes their reproductive turn by always replacing an individual of the other type (if available). In this work, we study the evolutionary stability of resident replacers who are invaded by mutant replacers. We find that residents are strongly protected against such invasions, and we quantify the strength of this effect by showing three types of results. First, we show that on well-mixed populations of size $N$, the invader fixation probability is exponentially small in $N$, even when the invader has a fixed relative reproductive rate $r>1$, and the same holds for all high-degree graphs. Second, we study bounded-degree graphs. We prove that on cycles, the fixation probability of an advantageous invader decreases only by a constant factor. However, we also present graphs with maximum degree 4, where the invader fixation probability is exponentially small in $N$ whenever $r\le1.9$. Thus, high degrees are sufficient for evolutionary stability, whereas with low degrees the evolutionary stability depends on specific features of the underlying spatial structure. Third, we prove general bounds for arbitrary graphs. Namely, we show that for any graph $G$ the invader fixation probability drops below the natural baseline given by the standard Moran process with oblivious individuals on a well-mixed population, both for $r\approx 1$ and for $r\ge 2$. Together, our results establish that the evolutionary dynamics of replacers is better characterized by the phrase ``survival of the first'' rather than the classic ``survival of the fittest''.