Here, we reply to the stimulating comments from Sagoff [ 1 ] and Rossberg [ 2 ] on Segar et al. [ 3 ]. Sagoff posits that species assemblages are largely fortuitous and ephemeral, which thwarts opportunities for coevolutionary processes [ 4 ]. Given the dynamic nature of ecological communities, have populations from different interacting species had sufficient time in which to generate selective pressure on each other? As Rossberg points out, in long-lasting and highly intimate bipartite networks, ‘frequent co-occurrence of the two taxa’ is required for evolutionary lockstep between vulnerability (v) and foraging (f) traits. Fitness ‘seascapes’ [ 2 ] stem from constant community turnover: but the adaptive troughs and peaks of the shifting seascape can persist and allow reciprocal evolutionary change if allelic turnover is rapid and selection strong enough. How do we specify ‘frequent co-occurrence’? Since Janzen’s 1985 appraisal of coevolution [ 4 ], Colpoda protozoans have been through over 53 000 generations: resistance to mosquito predators develops in 50 [ 5 ]. We do agree that ecological (nongenetic) fitting is widespread. However, biotic selection within ecological networks does occur, is detectable, and its effects are far from trivial.