We start with a definition: relative fitness is any statistic that is sufficient to predict the relative abundance of the mutant. There is other fitness concepts, like absolute fitness and fitness potentials (used for fitness landscapes) - these are distinct, we focus on relative fitness here. 2/n

What's the most basic relative fitness statistic? It's the slope dx/dt where x is the relative abundance trajectory of the mutant of interest. Clearly, it can predict the future abundance (linear extrapolation), but the slope only works over a short time horizon - see this simulated trajector. 3/n